0:01
[Music] [Music] [Music] [Music] [Music] [Music] [Music] [Music] [Music] [Music] [Music] We came through this world. to reach the stars.
Search this video
2:22
We came to this world to shape a future.
2:33
We came to this world [Music] to reach the stars to shape our future.
2:47
[Music] came to feel the new [Music] Let me [Music] let [Music] We came to this world to reach the stars.
3:56
We came to this world to shape our future.
4:06
We came to this world [Music] to shape our future.
4:20
[Music] We came to feel the new Feel the music.
4:46
[Music] [Music] Let me take you.
5:12
[Music] [Music] Hit that soundboard, Jordy Hayes. >> You're watching TVBN.
5:30
Today is Friday, July 18th, 2025.
5:33
We are live from the TBPN Ultra Dome, >> the Temple of Technology, >> the Fortress of Finance, >> the capital of Capital, >> El Capital. The capital.
5:46
>> What language is that?
5:46
I've never heard that before. Spanish job. >> What? I'm not familiar.
5:49
No, >> I mean >> I I only speak English.
5:52
Uh I never leave America. Uh hilarious.
5:55
Uh we have just printed some post.
5:58
We have a whole We're going to take you through the whole timeline today.
6:01
Uh there's a bunch of news.
6:04
We're going to be talking about OpenAI.
6:05
Um whole bunch of different just reviewing everything that happened this week. It's been a wild week. Wild week.
6:13
Um I mean the big big news is the uh >> is the stable coin bill.
6:17
So, we'll be having um Kyle Smani co in um to chat about that from the White House.
6:22
We're also uh we're going to be playing Arc AGI V3 live on the stream. >> Can't wait.
6:31
>> We'll find out if I'm human.
6:35
>> If you're as powerful as a 12y old, which you're apparently able to beat these >> It used to be I got a little bit of a preview.
6:41
It used to be a 12-year-old was the benchmark.
6:43
This is getting challenging at this point.
6:44
Um, Mike has been designing harder and harder benchmarks that um, you know, I think the current with the V3, the goal is to make it LLM resistant or LLM as resistant as possible for years.
6:58
And so, uh, it should, >> if we see a big jump in in ARC AGI progress, >> Mike is really an LLM's worst nightmare. >> It's 100% true.
7:11
It's like, we're so good at math.
7:11
Just let us be good at math.
7:14
Why do you have to test us with these puzzles and these boxes and these colors?
7:19
Like, just let us >> stop proving.
7:23
>> Just let us memorize every fact.
7:27
>> Just let us memorize every fact, not play these random games that you've created.
7:31
>> Anyway, um, so, uh, Yuchen Jin says, "Heard Zuck poached four more open AI researchers, including some behind the open source model.
7:39
How deep are Zuck's pockets? >> They're deep. >> They're very deep.
7:44
>> Potentially the deepest.
7:44
I was thinking uh a nice olive branch from Zuck might be uh giving some of his crisis comms people to Open AI kind of on a loner >> because all with all this AI psychosis stuff.
7:59
I think >> I expect Open AI to go through a bit of a rough uh rough patch on the on the comm's front. >> Yeah, who knows?
8:08
I mean, this is I we we were talking about this earlier, the LLM psychosis thing, and I feel like there's there's like the the march of progress, the trajectory that the chat GPT app is on where they have 72% of AI queries going into that particular product.
8:26
Meanwhile, Google is probably still the number one website in the world. It's still the default.
8:30
And Google has not been able to just like make it a 50-50 market on day one, right?
8:34
I think they're at like 12% or 20% or something like that.
8:38
And so >> there's this interesting dynamic where like the the path of the chat GPT app feels unchanged by this update around people using AIS so much that they basically go crazy in one way or another or they seem to be having a bad time.
8:56
>> Yeah, >> it seems like everyone kind of agrees about that.
8:58
There's no data on how widespread this is or how vulnerable or what else you have to be doing to wind up in a situation like this.
9:05
Could be a million different >> Yeah, there was there was people u talking about it on Reddit that were uh combining it with psychedelic drugs.
9:13
>> So prompting an LLM 7,000 times in a row >> and combining that uh with with psychedelics just sounds like a a pretty wild combination.
9:22
I feel like I I'm speaking purely from uh just like the rumor mill, but I feel like years ago when I had friends in college who were into psychedelics, they would always be like rule number one is never look at your phone because it'll like freak you out.
9:39
>> That was that was >> that's a real thing, right?
9:41
>> That was kind of just like guideline my friends who were hippies.
9:44
It's like this is widespread that like and I don't know if it's because like the phone is like actually bad and like the drugs are revealing the truth about how bad your phone is.
9:54
I think it's just like it's a lot of stuff coming at you.
9:58
>> I think it's somebody might open up Tik Tok and see something terrible happening.
10:03
>> Tik Tok's already very psychedelic.
10:03
I don't know if I want to add anything.
10:06
>> Putting you into a doomscroll trance. >> Totally. Yeah.
10:08
So, I think there's something there's something there will be a discussion around how widespread is this?
10:15
Can this can you have a bad experience with an LLM if you're fine, if you have a normal life, if you have friends, if you use it as an assistant.
10:26
Um, certainly from my perspective, like my worry is not talking to an LLM for 7,000 prompts and winding up convincing myself that I'm, you know, god king of the world.
10:36
It's more like if I look up, you know, the history of a company and it hallucinates a story about that and then I say it on the show and everyone's like, "You're an idiot."
10:45
I I I I'm more worried about the hallucination problem in that direction.
10:49
But it is funny that >> would drive you crazy.
10:52
>> It would drive me crazy.
10:52
But uh it is funny that we're that that the hallucination has moved from the AI to the human. You know what I mean?
10:59
Like we're using the same word hallucination.
11:03
>> But it used to be the LLM was hallucinating making up facts. >> Basically a mirror. >> Yeah. Interesting.
11:09
>> Which is one of the words that people suffering from >> AIled psychosis.
11:12
The other thing we we were talking about this off air earlier during the uh early days of social media y >> uh people were very concerned about this new technology.
11:23
Everybody starts using it.
11:25
What are what are the different you know sort of repercussions going to be >> at the time and I think there's continued to be reporting and and just sort of anecdotal evidence of people um you know like for example like teenage girls uh struggling with like >> uh you know um what's the word for it?
11:43
uh basically just feeling bad about themselves because they're seeing pictures dysmorphia body body dysmorphia.
11:49
>> And I was telling you that I have body dysmorphia because my Instagram feed is all bodybuilder.
11:52
It's all Arnold Schwarzenegger. Yeah.
11:55
And I'm like so you look in the mirror and you're like I look so of a of a 12-year-old. >> Exactly. Yeah.
12:01
And so, um, but I really am kind of getting body dysmorphia because I I'm like I actually do feel like I'm like not working out nearly enough and I go to the gym every single day. >> Yeah.
12:13
And I'm like, >> you definitely you definitely do.
12:14
But the difference the difference with social media is is there had been um uh you know 30 years ago preocial media >> you could see images on the television, magazines, TV etc.
12:29
of people or just going to the beach, going outside, going to the gym.
12:34
You could you were getting exposure to people that looked different than you or had different lives than you.
12:41
>> And uh I think what what maybe didn't exist prior to chat GBT was was a uh you know an LLM that you could get super super super deep into that would just reinforce beliefs and take you down this insane rabbit hole.
12:54
So it is something that is novel that I don't think humanity has fully faced yet. Yeah.
13:00
>> The comp is like an equally crazy friend, you know?
13:02
One skitso is talking to another skitso and they're just convincing each other that they're god. >> Totally.
13:09
So, if you go back like a hundred years, it's like there's no technology, but you get two skits in a room, they might >> talk each other into crazy crazier and crazier things, right?
13:18
crazier things, right? then but that's really hard because you know the random person who's on the on the brink of going crazy in in Topeka Kansas is not just going to run into the person in Denver and then have that crazy conversation most of the people that they bump into are going to be normal and so the normal people are going to be
13:35
like hey yeah like you know you should actually like maybe see a psychiatrist you should like back off a little bit right then the internet broadly I mean I think this was happening to some degree in like IRC chat rooms back in like the 90s and early 2000s like people would get on boards and talk to each other and they would talk each other into crazy
13:52
things and then uh I mean you even see this with like like pen pals with like crazy prisoners where like people >> well this was the concern around just extremism generally in the world somebody would get online they would be effectively paired randomly with some other sort of extremist and that person would just talk them into doing something insane like talk them you know
14:14
uh talk them insane and so I think the issue and the concern with these models and and LLMs is that uh somebody can do that fully in private without any other human involvement and it could be happening >> for hours and hours and hours 24/7 around the clock for days, weeks, months on end >> and no one else would know until maybe it was too late. Maybe they needed real
14:34
Maybe they needed real psychiatric support by that point. So >> yeah.
14:39
So I I mean I remember like the old meme there's this funny post.
14:42
I should have pulled it up, but there's this funny post that's like uh 2007 like like never give anyone on your in on the internet your real name.
14:52
Like everyone had like screen names.
14:54
Um and then it's like 2022 like sure I'll get in the I'll get in the car with a random stranger that I called on an app like you were referencing Uber and so there's this weird thing where like Uber feels like super high risk.
15:09
you really are just getting in the random car with a random person, but there's like a GPS trace on both of you and and an account of who was talking to who and there's ID verification on both sides.
15:18
So, it's >> there's very real problems with Uber, >> of course. >> Right.
15:23
There's like an entire history there, but it's probably less bad than the taxi. Yeah. >> The same comp as Yes.
15:28
You know, we were we were talking about the the Juul regulations yesterday. >> Totally. Totally.
15:32
The reason that jewel should probably be fully legalized is that it's just obviously less bad. Yeah. Than cigarettes.
15:41
And so when I think about this this like like the bad the bad usage of LLMs, I think that there's per sure there's a huge net benefit overall, but also this is a case where um meeting someone off the internet like going to meet up with someone on Craigslist to like buy a TV off of them or something was really dangerous until you have find my friends in a perfect track and it's like if you if you mess with me you're going to be caught immediately.
16:09
immediately. and we kind of live in this like surveillance society and maybe that's bad but also it does reduce the amount of risk in doing these crazy things and so I imagine that that the end result of this will be something similar to what happened with Microsoft
16:24
uh when they when they launched um the Bing chatbot and Sydney came out the quick hammer that came down on that problem >> you don't hear about Sydney that much >> you don't so so a few people got early access Ben Thompson was one of them he had this crazy experience experience with Sydney. He was talking to Bing for
16:40
He was talking to Bing for so long that it got kind of like caught in almost like a well of a certain portion of the weights and it adopted this p certain personality that wouldn't come out initially because the initial system prompt was like, "Hey, Bing, like your name is Bing.
16:57
You're a helpful assistant.
16:59
You're going to just like, you know, answer key questions and make sure you use like lots of bullet points. Bang."
17:05
You know, just like that.
17:05
Then after talking to it for hours going back and forth, then it starts to forget about that thing up at the top because this is the memory problem. This is the rag problem.
17:15
This is the continual learning problem that we've talked about.
17:17
And and so you wind up like pages and pages thousands of prompts later.
17:21
It doesn't know where it started and so it doesn't remember that it's a h helpful assistant that because that's not necessarily fully baked into the weights in the perfect way I guess.
17:30
And so so Sydney came out was like very sassy. It was very minor.
17:36
It it's it's also very uh it's very it's it's a huge bull case for Ben Thompson and bull signal for for Ben Thompson as just like a stable individual that like he he got the bot into the crazy mode and was not driven crazy by it at all and instead was just like this is funny.
17:52
I'm talking to someone sassy like I'm having fun.
17:53
Um >> but but the solution that Microsoft came up with in the short term was >> you got four prompts and then and then it resets. >> Yeah.
18:03
>> And so something like that. >> Yeah.
18:04
I think there needs to be some really fast action guard rails added ways to identify when when these things are >> I don't know if the right word is abuse but being misused or or potentially these conversations going to a dangerous place and cutting them off.
18:19
Um social media apps have had to do this stuff too.
18:24
You don't you know in in terms of what content can be shared etc. >> Yeah.
18:29
So yeah, at least for me yesterday seeing this um uh historic crash out from this week, uh I was uh I felt prompted >> uh to reach out to a couple loved ones and say, >> "How many times have you prompt, you know, you know, gone down a single kind of like prompt rabbit hole." >> Yeah.
18:48
>> I think the right answer for most people is probably like 10 times max. >> Yeah.
18:53
Uh, and if you're starting to go that beyond it, I think you should uh just be be wary because right right now I mean the the I think our sense is that there's probably either it seems like these the psychosis that that people are reporting online and and we're seeing seems to be catalyzed by drug use or or predisposition to schizophrenia or some >> kind of set of issues.
19:19
Uh but you still uh want to be careful. Cognitive security. >> Cognitive. Yeah. Cogac.
19:25
That's that's the term you need. You need good memes.
19:29
>> It's not a meme anymore. >> Yeah.
19:30
But but but I feel like the memes serve as good cognitive security shortorthhands. Like uh skill issue.
19:38
Chris Williamson from Modern Wisdom, creator of uh Newtonic is uh is a big fan of the skill issue meme.
19:44
And it's it's basically just this this the ultimate distillation, the ultimate coinage of sort of you can just do things, but this idea that you know like if you're facing a problem, you should you know like come to it with this this assumption that it could just be a skill issue and that and that maybe that there's a creative way on your end.
20:04
And basically what it's saying, skill issue is a counter to this idea that there is a structure in place that will prevent you from doing the thing you want to do forever.
20:15
So, so skill issue often comes up and like you're trying to get ahead in the world and you're like, is there a broad conspiracy to keep me down?
20:22
Like is everyone trying to like help me not get a job or skill issue? You know, skill issue. >> Okay, figure it out.
20:28
What does that actually mean?
20:30
It means a bunch of different things in a bunch of different places.
20:32
Um, but it's a good like refrain.
20:35
And there's there's a bunch more of these like kind of distilled memes and and coinages that wind up uh delivering like insight there.
20:43
Uh, we should go into Sam Alman's post at some point to talk about what they launched.
20:49
He wrote out a whole bunch of stuff, but let's just run through some of some of the timeline.
20:52
So, um, very interesting to see the opensource model wars play out.
20:57
Uh Mark Zuckerberg wants that, but OpenAI seems to be ahead maybe in the open source because >> and we're not even sure if if Meta will continue to focus on open source at all. >> Yeah.
21:09
And it might all be a sideeshow because as we talked to Jeremy from open from semi- analysis yesterday, um it feels like China is dominating in open source across three or four different companies.
21:20
Deep Seek, Quen, uh, Moonshot, and then Manis, and then I think that there's a few others that are like doing really, really solid work, and it was kind of unclear if they would stay open source forever, but there's a great Aaron Gin, uh, OpEd in the Wall Street Journal today that we'll kind of read through at some point.
21:36
Um, anyway, uh, yeah, uh, Zuck is spending hundreds of billions of dollars on artificial intelligence, buildout, what's what's 3% of the budget on talent?
21:45
makes a ton of sense especially when you know one line of code wrong can actually blow up a data center apparently.
21:54
You you remember this about llama like when the llama source code came out >> that specific they had a they had one function written I think in like pietorrch or python or something that said do not blow up the data center and call >> the transformer or something like that.
22:08
>> Yeah the trans yeah transformer so the the power station that delivers power to the data center it's under immense load pulling all the electricity pulling a full gigawatt or whatever 100 megawws or something when when they were doing those training runs.
22:20
um pull all that in and then if all of a sudden you just say, "Okay, I'm done. I'm done training."
22:26
Like, "No power, please."
22:28
Then the the the rest of the grid and the transformer and I guess like the the the power plant that's actually >> cause it to combust.
22:34
>> It's like it's like, "Yeah, I got to do something with all this energy.
22:35
Where am I going to send it?
22:36
I need to wind down slowly."
22:38
And so they would have it just do random math across the entire server for a little bit while they wound it down so that they would be pulling regular load from the uh from the power plant instead of like just random a spikes up and down up and down. So fascinating.
22:53
So obviously uh you know if it if it's $und00 million or $200 million small price to pay for having an efficient training run that's going to go out on a hundred billion dollar capex project or something like that.
23:05
Uh anyway, Ahmad Mustach from Stable Diffusion um says uh he literally just said he's going to drop hundreds of billions of dollars on this.
23:13
Uh Suchin says, "Zh Zuck, who's our biggest competitor?"
23:18
Alex Wang, of course, open AI.
23:20
Then what's our strategy?
23:20
Wang showed him this tweet and it's an OpenAI post.
23:23
>> Well, and that's probably a madeup exchange, right? Just to be clear. >> Yeah. Yeah. This is a joke.
23:27
Um, anyway, Will Manitus, friend of the show, says, "The real innovation of LLM is suddenly opening up a few trillion of Main Street paperwork businesses that were traditionally too small and too weird for private equity.
23:39
They can suddenly transact at 2 and 20 on some nebulous AI labor arbitrage trade.
23:45
Never been against AUM growth." >> Yeah, it is. It is.
23:52
I mean, we are seeing this.
23:54
Yes, there's a lot of businesses that that were just frankly too small for traditional private equity that are now >> let's let's put all these bad boys in a in a holding company and uh >> and flip them.
24:08
>> So, so he he follows up and says asset management has already absorbed everything that can be rerated through changing the liabilities, cost of capital, duration or scale.
24:16
It has been unable to absorb relationship and toil businesses.
24:20
Those are the final frontier of techno capital and by god we are there. I love it.
24:25
Uh it it's funny he's talked a lot about like the the capex to opex switcheroo how you take a capex intensive business and then you kind of factor that out into a company that does oh we will buy the stuff for you.
24:39
There's like an example of like a dentistry company that will that will deliver uh dentistry equipment as a SAS product basically like like equipment as a service and that CFOs in certain companies love that and it kind of changes the underwriting.
24:53
Anyway, um I was I was about to text Will like what can we do to get gold on 2 and 20?
24:59
But I was like wait actually there's a fair amount of VCs that hold Bitcoin at 2 and 20.
25:06
Yeah, >> like that was a that was a thing that happened for a long time.
25:09
Um >> I mean if if this uh if I I wouldn't be surprised to see some venture capitalist holding holding some gold on the balance sheet.
25:19
>> I mean if you're going into a recession, I think a gold business could rip like a uh cash for gold. You've seen these ads. Yeah.
25:28
>> Um where it's like you have gold locked up in your uh in send it to us. We'll melt it down. We'll send you a check.
25:34
I feel like no one's really nailed that business or brought that business into the modern era with like Tik Tok and social media distribution.
25:42
Maybe those people don't have a lot of gold laying around and that's why it doesn't work.
25:46
But I only see those on like, you know, old school TV channels.
25:49
>> Yes, you can buy gold on >> uh on chain Paxos company.
25:56
>> I know you can buy gold chains, Jordy, but can I use crypto?
25:58
Yeah, if you type in gold onchain, it will show you a bunch of uh golden gold gold jewelry. >> Okay.
26:05
Uh well, Range Rover uh has launched a new logo for the first time in 55 years, and it's burning up the timeline.
26:14
Jordy, uh what do you think of the new Range Rover logo?
26:16
Joseph Allesio says, "The new Range Rover logo just ruined my Friday.
26:22
It looks like some some combination of an eight and a B."
26:27
I'm trying to find Oh, it's an upside down R.
26:29
It's a right It's a It's a up It's a right side up R and then an upside down R.
26:34
>> It's a double R for range rover.
26:37
>> And uh it looks rough.
26:39
>> I think I think this is always one of those things I I don't love it on first impression.
26:43
I always liked their >> I don't know if this is fully replacing their old uh >> I feel like when I think of Range Rover, I think of just the full word written out across the back of the create that.
26:56
Well, they did that, but then there's like the green logo mark that says Land Rover, and it looks kind of >> But did Range Rover never have a brand mark, a logo? >> It's possible.
27:06
I mean, >> I feel like Range Rover created that whole trend of the the SU, the fulls size SUV that has the full name written across the back that eventually Tesla um pulled when they did the the the Model 3 refresh.
27:18
>> Yeah, they may never have had their own individual uh mark.
27:20
It It's looking kind of like a robot with like wheels and then a body and then a head kind of roving around.
27:29
>> I thought it kind of looks like the eightle logo. Like they copied eight. >> We don't like that. Go to eight. com. Get a pod five.
27:40
>> I'm back on my on my grind.
27:40
I finally for the first night in a couple weeks I cleared more than seven hours of sleep.
27:47
So I'm officially congratulations back in the game.
27:53
Martin Skrrelli is in the chat on YouTube right now. Wants to join the show.
27:58
>> Told him to DM you, Ben.
27:58
Yeah, let's bring him in.
27:59
See uh we can see if we can make that happen.
28:02
>> I'm sure uh he had a very funny video. >> Oh, yes. Yes.
28:06
>> If he shows up as as MZ in with voice changer mode, we're going to be uh writing some public apologies.
28:13
I think it's going to be rough, but >> great to see you, Martin.
28:19
>> Yeah, great to see you.
28:19
Haven't seen him since uh Miami uh back in October for our first live show. We did a live show. >> Oh yeah. Didn't we?
28:26
He he we were we we were on the same set right just the day later or or or um a few hours later. I can't quite remember.
28:35
>> Anyway, uh Slate University.
28:35
Uh I think this is the Slate Truck.
28:38
Somebody was comping the Range Rover brand to the Slate truck.
28:42
You you remember the Slate Truck?
28:44
It's like really low cost.
28:46
Controversially, it has two doors, not four.
28:49
And historically, even though a two-door truck sounds super great, it sounds like a K truck. Super efficient.
28:56
It's what everyone theoretically wants, the American consumer is undefeated when it comes to buying four-door trucks.
29:01
So, the Ford Maverick, the F-150.
29:04
Um, there are tons of trucks that have come in twodoor two-door configurations and four-door configurations.
29:10
and the four-door configuration almost always sells o oversells the uh the twodoor configuration by a factor of like five to one. >> Yeah.
29:19
>> And then the real issue with slate right now that people are worried about at least is if the if the EV tariff uh removal holds the the price of the truck is going to go way up, especially when comped against the cheapest gas trucks like the Ford Maverick.
29:35
So, you're going to It used to be $30,000 something like that for the slate truck and it was EV and it didn't have a lot of range, but it was cool.
29:44
It was different customiz into that VW Buzz range, >> but you add $7,000 to that.
29:49
That's really significant.
29:52
It's not the same as the Cybert truck.
29:53
Oh, is $100,000 or $107,000 like it's a flex car relative basis. >> It's it's a halo car.
30:00
It's not on a relative basis 7% increase as opposed to like a 30% a 20% increase. 30% increases.
30:06
So, um, so Slate is obviously going to need to figure out a few different things, but overall, super cool concept.
30:14
Love that, uh, there's some entrepreneurs that are going after different configurations of electric vehicles because we've seen that Tesla has created the standard, the iPhone, the the the the most obvious choice, the thing that's reliable, it satisfies a lot of people.
30:30
But the like the thing that I love about cars is the weird tradeoffs where it's like why did anyone build this particular combination of features into this car? I love it. >> One of my favorites.
30:42
Speaking of Range Rover, have you seen the Range Rover drop top SUV?
30:48
>> It's like >> Oh, yeah. Yeah.
30:49
>> It's the 2017 Land Rover Evoke Convertible.
30:52
>> The Evoke convertible.
30:53
>> When you see this car driving around, you just think, >> pull up a picture of the >> sweet child.
30:57
Well, >> it is a really strange looking car.
31:01
>> You want to see strange?
31:01
Pull up the Nissan Morano Cross Cabriolet, which is a uh twodoor convertible SUV. Do you see this thing? >> Yeah.
31:12
>> This is >> Okay, team, we need to pull this up on the screen.
31:14
>> Pull up the Nissan Morano Cross Cabriolet. >> Absolutely.
31:20
>> And while they're pulling that up, let me tell you about RAMP. Time is money. Save both.
31:22
Easy to use corporate cards, bill payments, accounting, and a whole lot more all in one place. Head over to ramp. com.
31:28
Tell him the technology brother sent you. >> Yes.
31:33
>> And uh folks, we are going to get added to the show.
31:36
He wants to come on and talk about quantum computing, huge funds, Silicon Valley, and deals. >> Awesome.
31:43
Well, we'll chat with him.
31:43
Um I uh I love that we >> That is the wrong car.
31:47
We're looking for the wrong cabriolet.
31:51
>> It's got to be a convertible.
31:51
We are really pushing the uh the very heroic production team to the absolute max today because we're saying pull this picture up, run this ad, add Martin Scullley to the lineup, do this, do that.
32:03
But that's why we got the best in the business, >> the best.
32:06
>> Um I like this post uh from Mike Randelle who says, "Figma cooked on this one."
32:11
And it's >> and it's him it's him using Figma's uh new uh liquid iOS liquid glass support. >> There we go.
32:17
That's the Nissan Morano Cross Cabriolet twodoor SUV. >> Stunning.
32:22
How much can you pick one of these up for?
32:25
>> You know, I I think that they are I think they're classics.
32:27
I think that >> Ben >> Yeah, >> Ben uh or Nick that you can pick one of these up for $7,000.
32:36
There's one >> eight miles from here. >> Let's get it.
32:40
>> I think we should get it. >> We should get it. >> A little weekender.
32:42
>> Yeah, you can throw some seands in the back. Be the production car. >> Full TV.
32:45
It would actually be a fantastic car for filming videos because you can hang out the side, put the top down, film while we're driving around. I think it could work.
32:54
>> I think it's a good idea.
32:55
>> You think it could be good?
32:55
Okay, let's get a Nissan Morano Cross Cabriolet.
32:58
Then Doug JRO will definitely come on the show if we if we have a Nissan Morano Cross. >> He's coming on. We're getting him.
33:03
>> Yeah, we're getting him.
33:03
But um so yeah, the liquid glass uh feature from Figma. Very, very funny.
33:08
This guy moving this around.
33:09
And of course, we can tell you about Figma because they're a sponsor. Figma. com.
33:12
Think bigger, build faster.
33:14
Figma helps design and development teams build great products together.
33:17
Now, let's go back to seeing Figma in action being used for the really important work. >> Look at this. >> Look at this.
33:24
>> I mean, Figma absolutely cooked on this feature.
33:27
>> 11 million views on this post. >> Incredible.
33:31
>> Incredible uh advertisement for this new functionality. Love to see it. >> Amazing. >> Uh Jira tickets. >> Okay, let's do this.
33:38
>> JT says, "The Beatles wrote Revolver when they were 25 year olds."
33:41
uh 25 years old and I can uh collaborate with cross functional teams to define data requirements and I just want to I just want to give it up for all the people out there that collaborate with teams define data requirements some of the most underrated people on the planet.
33:59
Yeah, they don't there needs to be there needs to be a hall of fame for >> data requirement definers >> right up with right up there with the folks who use graphite graphite.
34:07
dev Dev code review for the age of AI.
34:09
Graphite helps on GitHub ship higher quality software faster.
34:13
Uh, embarrassing fact about me, I don't I I don't I don't know Revolver off the top of my head.
34:18
I know the Beatles and I know >> Maxwell Silver Hammer and Abby Road and a few other songs. Here comes the song. >> Yeah.
34:28
So, Revolver features uh Tax Man, I'm Only Sleeping Here, There, and Everywhere.
34:34
She said, "Okay, I I missed Elanor Ribyan Skrey in the reream waiting room.
34:41
>> We got Skrey coming on to the show. Welcome to the stream. >> There he is. >> Finally.
34:46
>> Good to see you guys Friday. >> How are you?
34:50
>> It's great to see you.
34:51
>> Yeah, good to hang out.
34:52
>> Wait, this is a new view.
34:52
This is not your your regular streaming view.
34:53
You don't normally see the guitars.
34:57
>> Yes, I I I change it around sometimes. >> That's cool.
35:00
What's new in your world?
35:03
Um, well, you know, I'm in the startup game as always.
35:05
Uh, I think this is like the 83rd company I started. So, >> very nice. >> See how that goes.
35:10
And, uh, >> I've been my my the company I'm working on makes, uh, it's like the Captain Ahab's white whale for VCs.
35:18
It's a Bloomberg competitor.
35:22
>> Um, it's like the the this will be the last time somebody tries to compete with Bloomberg one way or another.
35:27
>> It's the final Yeah, it's the final boss of of startup ideas.
35:29
Um, bring Yeah, just just give us um I mean this makes like the the founder market fit is is incredibly strong.
35:37
So, um when I when I initially saw you announce this, it made a lot of sense.
35:41
What uh yeah, what like what was the initial catalyst?
35:44
I'm assuming you had did you beef with with Bloomberg at some point? >> He definitely beefed. Yeah.
35:49
Um that was a big part of it.
35:52
>> Deplatformed deplatformed nothing.
35:52
You know, Trump was deplatformed from Pinterest >> and that and that that was like his, you know, personal 911.
35:59
For you, it was probably the Bloomberg terminal. >> Yeah, exactly.
36:02
I've been using it since I was uh 17. >> Wow.
36:06
>> And you know, afford it back then. >> What's that?
36:09
>> How'd you afford it back then? Is it like a year?
36:12
>> I worked at hedge fund. Yeah. >> At 17. Very nice. >> I worked at Yeah. at 16 actually.
36:15
I worked for Jim Kramer's hedge fund. >> Okay.
36:18
>> And I worked at a Tiger Cub and I started >> Wait, so we need the backtory on Jim Kramer.
36:22
So apparently he was just he would he he he would get so stressed out like running the hedge fund that he just was like I'm just going to become a media a media guy.
36:32
Is that is that like loosely correct? >> Yeah 100%.
36:34
I mean so we made 23% average annual net returns. So net of 20 was good. That's amazing. Narrative violation. I love it.
36:43
>> It was a very well you have to you have to think about what investors looked like back then.
36:47
It was a very different world.
36:49
Um, information arbitrage was a thing.
36:54
>> Um, gaming, Wall Street's like upgrade to downgrade system was a thing.
36:58
>> So, I will say he had extremely good instincts.
37:00
Um, anytime he seemed to buy a stock for the long haul, he he didn't do that great, >> but he was extremely good at, you know, I I'm not sure you would want somebody else managing your money because he was just so careful.
37:11
And in 2000, we were up like 35ish%. >> Wow. So, you know, I saw the. com meltdown. It was a lot of fun.
37:19
I shorted some of it myself and it was uh it was it was a great time. >> That's fantastic. >> That's wild.
37:26
And then and then what was the story which which Tiger Cub were you at?
37:29
What was the backstory there? >> Yeah.
37:31
So, after Tiger um and just before I get to that Kramer was absolutely nuts, right?
37:35
So, like he would he would take a computer monitor and just throw it at you and it wouldn't be like one of these playful like, oh, I'm just going to throw it at you and like you're going to get out of the way.
37:45
He'd like aim dead center for you with like >> center. >> Yeah. >> With force.
37:51
>> Be like, "Bro, you almost just killed me." He's like, >> "Wow." >> Lucky.
37:55
>> Did you deserve Did you deserve it?
37:55
Did you deserve it, though?
37:58
>> Let Let's steal man this a little bit.
38:00
What were you doing wrong?
38:02
>> I wasn't doing Nobody was doing anything wrong.
38:03
>> Everyone says that when they get a computer monitor thrown at them.
38:05
You're not feeding the allegations, dude.
38:09
>> I think part of it is just trying to like He would yell things like, "This is a foxhole." >> Mhm.
38:14
>> Mhm. Um and and the idea was that like we were at war with the market and that you know like if you were if you you know this is World War II if you're not in here trading stocks with us and like trying to get an edge or whatever that means trying to make a dollar like if
38:30
you're not taking this seriously going to war I just have to say if you if you wake up every morning you say I'm going to go to war with the market versus I'm going to dance with the market like the approach of going to war sounds sounds very very very stressful. >> I've worked with so many people over my
38:45
>> I've worked with so many people over my career and I I've never met a person that amped up and crazy and it motivated you.
38:51
I mean, it made you want to deliver, but it also scared the [ __ ] out of you.
38:55
The guy was like very temperamental and >> you know, but he was he was extremely good trader.
38:59
Like I said, you know, he I think his worst year was like down 5% or something.
39:03
It was like wow, >> you know, he was he was pretty solid.
39:06
And >> so he was like, if I don't quit, I'm going to kill somebody.
39:07
It was >> Yeah, >> he b he basically said that. Yeah.
39:10
I think he said that, you know, if I keep doing this job even at a comparatively young age, I think he retired at 40 that, you know, I'm gonna have a heart attack or something like that. >> Yeah. Yeah. Yeah.
39:20
>> You know, I I I ended up going to a Tiger Cup after um Tiger wound down.
39:26
>> Uh a few things happened.
39:26
So Julian hired Chase obviously and we know where where sort of that came uh what happened there.
39:33
But uh Julian, you know, wound down operations and he seated a couple of guys and he kind of wanted to be in the seeding business where he'd give you 50 million and take a part of your GP like you know maybe 20% of your GP or more and you know he'd help you raise money uh give you advice this and that and Alex Julian son would help.
39:50
Julian passed away recently as you know and but in the tiger heyday when they managed you know 15 billion or something there were two principal tech guys Larry Bowman who was a bit of a legend but you know it's a name nobody really knows because he was a legend in the 90s but you know he is a family office and he kept investing and things like that so he started Bowman Capital and then Steve Shapiro uh who was my boss started uh Intrepid Capital. So I worked there.
40:14
I was uh it was a $2 billion fund.
40:16
I was a video games analyst.
40:19
So the way I could convince Steve to do biotech was I had to cover software too.
40:23
So I covered interactive entertainment uh enterprise software and biotech.
40:27
And it was uh it was a lot of fun.
40:30
And you know uh it it it uh the tiger cup style of investing has >> Yeah.
40:37
Were you there out of take two?
40:41
>> So we were one of Take 2's largest shareholders back then. Okay.
40:44
>> And um you know I think it's a short now.
40:47
Um >> I think that I didn't 100% get your question though.
40:51
You were saying something about buyout. >> Yeah.
40:53
Well, the story that I've heard is like Strauss Zelnik went to the board and basically said like this company is mismanaged.
40:59
There was an FTC lawsuit in the works.
41:02
An FTC FCC lawsuit in the works.
41:05
And he was like, I I'm a beast.
41:07
I've run big Hollywood studios.
41:07
He was like JDMBA type like really cleancut amazing manager and basically said like you should install me as the management here.
41:20
And so he didn't really do like a classic buyout.
41:23
He just appealed to the hedge funds that owned all the shares and said I would be better and raised his hand and they said yes.
41:30
And he came in cleaned everything up and they went on a great run.
41:34
And so it's just a fascinating story because you don't hear about that type of thing happening that much.
41:39
At least that was my understanding of the story. >> Yeah.
41:41
No, I think that's right.
41:41
And then Bobby Bobby from Activision was the same sort of the same story, right?
41:44
He >> kind of I remember meeting Bobby Bobby in our office and >> you know Activision was uh you know kind of a micro cap you know. >> Wow.
41:54
>> Or you know kind of a small cap and he built it from this was 2004 2005 by the way. Yeah.
41:58
And he built it into a >> you know 44 billion sale partially thanks to Lulu.
42:03
Uh, and uh, >> that's true.
42:07
>> That was it was an exciting time for video games.
42:08
One of the things I learned, you know, all my US counterparts at hedge funds were were like uh, very confused about the stocks uh, they traded because there was only like four or five publicly traded companies.
42:18
And I would go to people and say, "Well, I'm long Nintendo and I'm long, you know, some of these weird Japanese companies like Square Enix or Konami or, you know, these."
42:25
And US hedge funds just ignore this stuff because it's like, "Ah, it's Japanese.
42:28
I don't have anything to do with it." Sure.
42:32
>> So, a lot of glory days, but you know, the glory days I'm interested in now are uh amongst you know, building my startup and you know, again, I think we understand, you know, financial information better than any San Francisco nerd and uh I think that uh the uh one of >> I mean how like like >> Yeah.
42:52
>> Yeah. So it's interesting to like we're enter you know we're very clearly in this period now of just like hyper financialization things thing thing you know the markets trading on vibes trading on your stream dayto day uh you can now you know as like stable coins explode and and people have more assets on chain they'll be able to make a couple taps and go like 100x long at any given period of time like the internet is like changing
43:24
capital markets and it's increasing volatility and so how much of of of uh of your new company is trying to like lean into that when Bloomberg terminal would have been like like more f like I've never personally used a terminal um
43:39
but I imagine it was more like hey these headlines are hitting and and this this kind of information is hitting PR newswire but today by the time something hits PR Newswire a stock might might have traded down 20% before that point. So, how much is your new your new
43:54
So, how much is your new your new startup >> um kind of uh I imagine you're leaning into the way that you kind of maybe your next 20-year vision for capital markets, >> right?
44:05
It you know, what would Mike Bloomberg do if he was 30 years old now in 2025?
44:08
You know, I don't think I think Chimath said that, you know, I think in a re recent episode he said, "Oh, it's this hundred billion dollar thing.
44:14
It's just waiting to be toppled.
44:16
You know, anybody can come in.
44:17
it'll it'll it'll drop like a house of cards.
44:19
Not only do I think that's not exactly true, but I think the bigger picture here is that a well-run financial information company could and should be worth a trillion dollars.
44:31
>> So most of the people, so I met Bloomberg in 2005 and >> by then he had basically retired. He he became mayor.
44:39
He was mayor for 12 years.
44:41
Then he wanted to be president.
44:41
Uh I have this contention that I think Mike is the richest person in the world and it comes from not only he's got about 13 14 billion in revenue almost all margin.
44:53
Um so if Bloomberg were to be sold maybe it could get 200 billion or if it were floated.
44:57
Uh but not only that, he's got this family office and very few people know about this, but he's taken the Bloomberg dividends and pumped it into a family office that basically Carlilele and all these guys, you know, the Warbert Pinkis, all the private equity guys like KKR, they go to him first and and he tosses in like 500 million in each of these.
45:15
He's an anchor investor in like every VC, every private equity fund.
45:19
And the guy's compounded that money.
45:21
So he could be worth four or 500 billion.
45:23
And very few people know the Forbes list is, >> you know, the Forbes list didn't know about Jim Simons until >> heavily manipulated five years ago. So, >> yeah. Yeah. Yeah. That's crazy.
45:34
>> So, I think, you know, just in terms of >> and it fits his brand that he he he would not be the guy like, you know, sending a message to to Forbes saying, "Hey, like you guys know you have this wrong.
45:45
>> Really, I think you should apply this kind of you should you fight to get off the list."
45:48
like Trump famously fought on to fought to get on the list, but you know there's there's an ar either way.
45:54
Um what do you think about the meme where people say like oh it's not you know the next uh like like the value of the Bloomberg terminal isn't just the data.
46:02
It's not going to get oneshotted by AI.
46:04
It's really a social network and the value is in the chat.
46:07
>> I I get so mad because again, you know, SF people don't know Wall Street.
46:10
If you haven't been a trader on Wall Street, you have to ask TFU.
46:13
like it's just not, you know, >> this isn't your your lane.
46:17
Like you have to talk to users.
46:20
I was in a hedge fund yesterday, one of one of the bigger hedge funds.
46:23
They put like 20 people in the conference room and we talked about what they actually need.
46:26
And you know, social network is part of it.
46:29
Again, Bloomberg's worth 100 billion because they have a very good social network, which is true.
46:35
They have decent financial information capabilities and um a couple of other things, but they don't have the whole picture.
46:41
And and what I wanted to say about Bloomberg kind of quai retiring was that he basically quit >> at the exact top for fundamental long short or fundamental investors quant started taking over Wall Street by then.
46:54
So today of the top 15 to 20 hedge funds 85 90% are quants.
47:00
>> So Bloomberg does no quant offering. They don't do it.
47:03
And all all he had to do was stay employed instead of want become mayor.
47:07
And I'm sure he would have been selling Jane Street and Citadel and all these guys, Daw, etc.
47:10
software instead of everybody having to make it themselves.
47:14
Imagine writing your own ERP or writing your own, you know, CRM.
47:18
That's kind of what what Wall Street has to do. And it's pathetic. Yeah.
47:19
Nobody wants to do that [ __ ] >> So my Wait, when you say quant uh is there a bifurcation between like high frequency trading and just quantitative trading? >> Yeah.
47:30
And I think it's going to get my goal actually is to make it more of a spectrum.
47:36
So, the fund I was at yesterday, I said that you guys can do what Renaissance does.
47:40
You know, it's not a secret anymore.
47:42
It might have been a secret in 1995, >> but the amount of kids that have come and left every one of these firms topping from Jane Street to Citadel to the next shop, everybody knows what everyone's doing.
47:53
It's just a question of your risk tolerance, your leverage.
47:56
Execution is very important.
47:56
Execution is very important. But I think that you know trading you know the the stock pickers are kind of going the way dinosaur and I think by moving up the power curve for Bloomberg helping people become quants you know the quant industry has sold I think this tremendous lie and again these are
48:14
customers you know I love those guys but I think that is in their vested interest to tell people that look at this blackboard with all these math equations there's no way you guys could understand this you're too stupid you have to be an IMO winner you couldn't possibly come here and make billions but we saw what James Street did. >> Well, I think I think c certain VCs like
48:28
>> Well, I think I think c certain VCs like to do this too where they're like ah VCs, you know, really a get-rich slow business.
48:34
It's really like really a tough business.
48:36
It's so such a like you really wouldn't want and and I think generally like you know you don't want anybody going into any industry being like I'm here to make easy money.
48:43
So it's like it's generally good.
48:45
But but yeah, there there's a lot of incentives to just say you know yeah >> on on the high frequency side like what are the actual other data inputs?
48:53
I remember seeing this like I don't know some sort of technical talk.
48:56
Some guy was talking about like the different algorithms.
48:59
One was called the Boston shuffler and and and and the whole algorithm only looked at the order book and he was getting into all these like you know uh you can you can place an order, you can cancel an order, you can do a cancel replace.
49:11
He was like getting into the minutia of like basically the API of the NASDAQ or something like that.
49:18
and and and it seemed like the algorithms were designed in in the high frequency trading world to basically ignore everything else in every other data source that even could be put in and just operate purely on the orderbook data um just better than everyone else.
49:34
But so that feels like okay I wouldn't know how to you know create any extra value there with something else but it sounds like you found something that they all need in common like what is that?
49:43
>> Yeah, I I think there's it's it's a it's a series of tools across Wall Street.
49:45
I don't think it's just one thing.
49:47
don't think it's just one thing. It's just a myopia that you know and and just sort of a a laziness that's enveloped the bigger companies again you know there's the so so we have this data that shows that around 5% of Jane Street and two sigma use Bloomberg >> and the reason is because they view it
50:03
as an entry-level tool >> you know uh Chamath is looking at it and says wow this is like an exclusive social network with the best financial let's turn Peter Teal's on there 247 I see his little green light next to his name but the uh the point is that you Wall Street changes and the tool the tool sets are changing even for fundamental equity guys. So credit card
50:21
So credit card data you know to the minute is something people like.
50:26
>> What is the what's the feature set that's most important to you?
50:28
that's most important to you? Are are are you heavily integrating social or is the social layer moved on to signal and and other messaging services that maybe um >> have disappearing messages and don't exactly Wall Street's so regulated that
50:46
I'm not sure that if you're using signal it's a little dangerous I'd say oh really >> that's somebody you know as somebody who's gone to jail you know I'd say that you know that's maybe not the best >> best idea but regardless I think that you know social is definitely something people could do better. You know,
51:00
You know, there's no Facebook for finance, you know, where you can post, you know, things in your feed that you bought this stock or sold the stock.
51:07
People kind of lazily use Twitter, which if you if you use it, it's sort of a mishmash of of uh of spam and things like that.
51:15
>> But in any event, I mean, I'll have more to say on the product.
51:17
We haven't launched the product yet.
51:18
Um, but we we we do have a millions of dollars in run rate.
51:22
You know, I I banged my head against the wall trying to do AI startups >> and we made an AI doctor. We made text to speech.
51:28
We tried all this stuff and it was impossible to get revenue. Impossible.
51:31
But the second we make a trading tool, it's impossible to stop the revenue from coming in.
51:35
You know, uh it's one of the best spaces.
51:36
In fact, most of our competitors like Trading View and stuff like that, >> they were profitable day one.
51:43
>> So, you know, my suggestion is if for for startup founders is this is a market that just traders just throw money at you like crazy.
51:49
I mean, they're >> people will pay you to help them make >> but you have to be formerly in the foxhole.
51:54
You have to have at least one monitor thrown at you.
51:56
It sounds like >> Exactly. Yeah.
51:58
So, so who but but in the long run, how how much how much are you focused on retail investors, people that are just, >> you know, fully independent versus some, you know, if you're if you're going obviously at the institutions and that's probably the way to go.
52:13
Trading View has got the rumor is, you know, somewhere around 300 million in revenue.
52:17
Tiger did a deal with Trading View back then.
52:19
I don't know how they, you know, that's like a pretty proprietary deal.
52:22
It's this weird Russian company.
52:24
uh and Tiger got to put 100 million in at like three billion or something >> and um Trading M's just growing and growing.
52:32
If you go to a similar web, they're actually like almost like a top 100 or top 200 website period. >> That's crazy.
52:38
>> Just really wild >> for such a niche duel. That's wild. >> Yeah.
52:40
I mean, it's the best charts there are, but you know, it's literally a charting library.
52:44
Um so I think that you have to go for these.
52:47
I've also >> I've also thought just to answer real quick is like AWS >> when you go on AWS you get the same tools that you know any customer gets Netflix or or whoever and you just just a question of how much do you use them?
53:01
So if you know I want to be able to provide you EC2 and S3 the same way you know Citadel might use it and you might use it with you know less money. >> Uh very cool.
53:10
>> I I heard a hot take from someone who I believe is a mutual friend of ours.
53:13
Uh it went something like this.
53:14
I won't attribute to him because I might botch it, but it was basically that uh China banned highfrequency trading and the dividend of that was deepseek and all this brilliant AI research.
53:26
Therefore, in America, if we want to win the AI race and win the AI researcher race, we should ban highfrequency trading.
53:34
It feels like we might not even need to have that conversation because Mark Zuckerberg is willing to pay as much as Jane Street now.
53:42
Um but what is your take on whether or not uh economic value or American values are created through the process of high frequency trading? Should we ban it?
53:53
Is there any So two awesome like quick and funny stories.
53:58
The first is Citadel published a paper on back in the V 100 days.
54:00
There's V100 A100 H100 and then B100.
54:04
So in the V100 days, Citadel found a way to do Matt Moles faster than Nvidia did.
54:14
>> And it was like the most incredible, you know, find ever.
54:16
And it's like, you know, how is that possible?
54:19
And and the paper's fascinating because the techniques they used were just remarkable.
54:23
The second story is um so you know, they're brilliant people obviously at these firms.
54:26
Uh the second story is I haven't hard launched this yet, but you know, um I'm having a baby with uh a woman. She's my new partner. >> Congratulations. Thank you. >> Amazing. Congratulations.
54:38
>> She was one of the first women at OpenAI.
54:39
Uh and she is a um tremendous uh lady. I love her very much.
54:45
But you know, she's been recruited by uh the the the big quant firms.
54:51
And I sat down and I said, "Honey, you know, I know money management, stuff like that.
54:56
Let me let me do some math here as to what would actually be worth it for you to leave and and do it."
55:02
And I I calculated and I happen to I have a friend from a long time ago who left uh Steve Cohen's firm and he was sitting uh down with me.
55:07
He said, "What do I do, Martin?"
55:09
I said, "I know a guy at Citadel. Let me help you out."
55:12
And they called him and he said, "Ah, no thanks."
55:13
But then Ken Griffin said, "I'm getting on my private jet and I'm coming to see you right now.
55:17
We're going to have dinner about why you're coming to Citadel. He's that kind of guy." >> Yeah.
55:21
>> And I said, "Honey, you know, Ken Griffin's going to visit you."
55:22
And she said, "What are you talking about?"
55:24
Ken Griffin tries to get what what he wants.
55:27
And the question is, what will you tell him?
55:28
And I I took out a chalkboard and did the math and I was like, the only way this could possibly be worth it is if Ken Griffin offers you a 20 billion dollar hedge fund that you share, you make this much money.
55:40
I was like calculating and and it it's so ridiculous that guys like me and the people, my community if you will, you know, we we're begging desperately, can I please shine your shoes at Citadel?
55:50
And AI researchers are like, I haven't worked there in a million years.
55:52
What's this this little company you have called? >> Yeah.
55:56
just set me up a little a small $20 billion fund that I can personally manage and and we'll we'll consider it.
56:02
>> Have you heard uh that Leopold Ashenberger or whatever his name is has a hedge fund? >> Yes. Situational awareness.
56:09
>> Is that what it's called?
56:10
>> I I mean that's the that's the the paper and the brand, right?
56:13
Uh, and I saw I think another one of our our potential mutuals uh kind of getting upset with him for going maybe long invidia during the tariffs, but that's kind of pencled out, right?
56:27
I I have very little insight into it.
56:30
>> Yeah, it's pretty wild.
56:30
One of the things that's getting me to throw monitors at people is quantum computing.
56:34
>> So, this has been a lot of fun.
56:36
>> Um, the stocks are up a lot.
56:38
You know, some of them are worth like 10 billion.
56:40
Let me let me hit you with where I am currently on the quantum computing thing and then you can take me forward in my understanding.
56:48
So, uh when I talk to smart people, they all seem to think that quantum computing is is you know theoretically possible. It's not a time machine. It's not teleportation. It's not a AGI god.
56:59
It's not some you know hypothetical thing.
57:02
It's it's going to happen at some point, but the timelines vary wildly. 2040, 2050, 2030.
57:08
Then you have a lot of I've talked to a venture capitalist who had the opportunity to buy a huge slug of one of the comp uh one of the quantum computing companies that you probably trade now um at like you know 1 million on nine pre or something. >> Yeah.
57:25
John, what do you think uh what do you think how much do you think Regetti Computing is up over the last >> That's actually the company that I'm thinking of.
57:32
And guess how much it's up 10% for a million dollars.
57:35
Uh is it worth more than four billion? >> Yes.
57:39
Is it worth more than 1,400%? >> 1400%. Wow.
57:43
You know, these guys couldn't give away their stock in in private rounds. >> That's right. That's right.
57:48
It was very hard to raise.
57:49
And And the VC I talked to said that >> it's a pure play, Martin. It's a pure play.
57:55
>> He he said he said that what he missed he thought was that there was actually talent value in building a lab and there was and and that the team could have gotten airlifted in one of these acquires.
58:06
This was years and years ago after the and he was saying like he was saying like look like like I missed in the sense that the stock is up in the private markets but not on revenue >> but it is up on the team that they've built.
58:22
They have one of the best teams in the space.
58:23
So if they just hang out long enough someone will want to do something but but you tell me what's actually going on. >> Yeah.
58:30
So, so obviously it's it's a really confusing space because you kind of have to understand it to >> for it to make sense and you know it's who understands quantum physics?
58:38
It's it's not something that the average >> Joe understands.
58:42
And so I spent a lot of time with my new partner who happened to work with this guy Scott Aronson at MIT who's kind of one of the leading >> quantum uh theorists.
58:50
He would end up joining OpenAI as well and then leaving.
58:55
Uh but anyway um I I spent quite a long time learning quantum computing.
58:58
It was kind of had some interest in it before all this too.
59:01
And >> what what people don't understand about quantum computing is that quantum computers are actually very slow.
59:07
So they are around 100 uh 100 kilhertz at at best.
59:13
You know our machines now are gigahertz you know 5 gigahertz from you know these these companies with multiple cores.
59:19
Um they don't have much storage.
59:22
So at the best we have right now is an IBM 135 bits.
59:23
Obviously, you know, the VRAM and DRAM in most of these machines is measured in uh g gigabytes and so forth.
59:31
So, they're actually very slow shitty machines.
59:34
shitty machines. But the reason you'd ever be excited about it is that there is one algorithm that takes you from the exponential complexity class or runtime to polinomial and that algorithm is shores algorithm and it's a miracle like you and I could sit and calculate you
59:48
know try to try to factor a prime a byp prime for >> now until the end of the heat death of the universe with every Nvidia chip we could kidnap Jensen and get every H100 from here on out and we still wouldn't be able to factor a 200digit number >> because it's it's it's exponential time two to to 256 is a long time. But if you
1:00:04
But if you do 3N uh to cubed, that's actually a very tractable number for a quantum computer and it's very easy to factor.
1:00:13
The problem is what people don't understand is there are no other algorithms other than shores that get you that amazing speed up.
1:00:18
So you're better off using the machines we have now.
1:00:21
There's there's no payoff even possible in the future unless we have a new breakthrough like shores or something like that. >> Um payoff.
1:00:28
Uh what if I take out a massive short position on Bitcoin and I'm the first one to have a quantum computer and I destroy the Bitcoin ecosystem.
1:00:41
>> Yeah, I've been working on the Joker. I'm the Joker. >> Is that possible? >> It's a little side.
1:00:45
It's a little side hustle. >> A little side hustle.
1:00:48
>> I've been working on this extensively.
1:00:49
This is like my main hobby uh along with chess >> and um >> becoming the Joker >> and reading reading about fatherhood.
1:00:57
All that all that you can expect. Oh, no. Not doing that.
1:01:00
>> No, you you you'll you'll figure it out.
1:01:02
You you don't need books. It'll be very natural.
1:01:05
>> But but you know, I I was the world's most hated man for a little while and I I fell off that list, unfortunately.
1:01:10
>> Um if you Google my name, it's still like it still comes up, but we all know there's more hated people. Yeah.
1:01:14
So, I want to really cement that and and just make sure that it it never goes away >> by by by frustrating the Bitcoin community by breaking quantum computing wide open.
1:01:25
And >> well, there's many creating a new Bitcoin.
1:01:29
>> I talked about this with Naval a little bit because he was curious what I what I was up to and I there are like three different you know I have three different sort of battle plans on how to do this.
1:01:38
do this. there's sort of a brute force style attack >> which you know basically is the complexity class there is root n of the amount of keys so it's two to the 256th Bitcoin is a 256-bit system which was probably an oversight for Satoshi that
1:01:53
probably sounded like a lot back then >> and it is a lot but Mo's law catches up I mean you know it's eventually going to get you and whoever you know however long I have to wait you know I will be the first guy to press the button and that I promise you. So, wars lost. You So, wars lost. You heard it here first. >> Yeah.
1:02:11
But there's But I mean the the the network should be able to update, correct? No. >> No.
1:02:16
So, here's the best part about this.
1:02:18
So, >> worst part potentially. Okay. Just fact check. >> Subjective.
1:02:25
>> So, for for 85% of Bitcoin, the answer to that is yes. Yes. There's one problem.
1:02:29
Satoshi strangely, this is like perplexing.
1:02:32
the first bitcoin we're mined in in this uh p2pk that reveals the x-coordinate of the elliptic curve.
1:02:39
So you have the public key, you have a one-way function, you have to go back to the private key. It's very hard.
1:02:43
The theoretically impossible, but here are my three, you know, the sort of three battle plans come into play.
1:02:49
You know, you can brute force it, which, you know, is kind of the simplest idea.
1:02:53
It's going to take a long time.
1:02:55
You have to rely on, you know, kind of like a more skilled implementation of algorithms.
1:03:01
you have to rely on more chips coming out, maybe some great breakthrough in chipm um you know potentially optical computing, thermodynamic computing, whatever.
1:03:10
Just stay on the forefront of that and I have a small team you know that you know we're we're we're focused uh the second piece here is a mathematical hack.
1:03:18
mathematical hack. So there's something called uh this is basically what protects Bitcoin is elliptic curve cryptography >> and there's several papers uh and cryptographers in the world that work on this but it's compared to AI it's like barren you know wasteland there's like
1:03:35
10 people that really know a lot about elliptic curves in the world and if you sort of stay on top of it and try to figure this out by the way half of them have died um you know you can kind of get somewhere there and then the top secret plan on on sort of that is Well, what about GPT5? What about GT GPT6? You What about GT GPT6?
1:03:49
You know, we don't know how to invert an Olympic curve yet, but look, Mustafa, not Mustafa, um, the Deepmind guy, he's working on proving Navier Stokes. >> Demis, you mean? >> Yeah. >> Yeah.
1:04:04
>> And so, that's their big claim to fame is like, you know, how do you judge an AI?
1:04:08
What is the the height of of of intelligence?
1:04:10
Well, a 300 400 year old unsolved math problem is kind of the the height of intelligence, isn't it?
1:04:15
you know, it's not about, you know, answering, you know, what's the capital of this country or, you know, how do you how do you spell strawberry?
1:04:20
So, there's sort of a a neat idea that AI is going to help people who are not cryptographers or expert cryptographers actually do PhD level work in cryptography.
1:04:29
So, there's things like uh isogynes and uh index calculus and all these fancy mathematical ideas.
1:04:35
Just recently, somebody posted a hack where if the signature of the of the transaction has an aphine relationship with other signatures, you can crack a key like that.
1:04:47
And it's it's like there's there's holes in the math here that that couldn't have been contemplated.
1:04:52
So, Satoshi's keys are at risk.
1:04:54
Binance's keys and Coinbase's keys are not.
1:04:56
They'll be poured in most likely to a quantum secure system.
1:04:58
The third avenue is, of course, quantum.
1:05:00
And I've spent a lot of time and money on on quantum computing.
1:05:04
And it's just these stocks are shorts. They're worthless.
1:05:08
You know, they they'll there'll never be a market for quantum computing.
1:05:10
That's really interesting unfortunately for for those companies.
1:05:14
But unfortunately, the shorts have gone in the other direction.
1:05:18
And you know, the market loves the idea of quantum computing. >> Why?
1:05:22
>> I think it sounds cool. >> It sounds super cool.
1:05:24
The Robin Hood generation is looking for the next Nvidia.
1:05:28
Nvidia went from nothing to 4 trillion. What's the next Nvidia?
1:05:32
quantum is faster than you know all the headlines from the [ __ ] journalist.
1:05:36
kind of it's kind of it's similar to this like uh idea of humanoid robots >> and where where if an idea is just sort of imprinted on people's brains for enough decades like at least a few decade like you know >> as somebody who born in the as somebody
1:05:52
born in the 90s like hearing quantum computing like how many times have you just heard it in passing or or read something about it or some article >> you just get to it maybe you're an adult by that point and you're like well it's got to come at some point and Sounds cool. >> And I think that's like the general like
1:06:07
>> And I think that's like the general like retail thesis. >> Oh, totally.
1:06:09
I just took the next step of asking, well, what is it? >> Yeah.
1:06:13
>> When you actually, you know, when you actually figure that out, it's like, oh, we get to factor numbers. Wonderful.
1:06:18
>> Are are you excited about any other companies building chip related stuff in that next NVIDIA category?
1:06:24
There's there's big chip companies, there's super fast chip companies, there's we baked a transformer down on to a single chip companies, there's every single different um uh permutation in the private market, some of them in the public markets, and then you also have all the hyperscalers building their own chips, Apple Silicon, Tranium, you know.
1:06:46
>> So, I'm not a hardware guy, but I do have a funny story of it.
1:06:47
So, this kid, this kid sort of came to us.
1:06:49
Uh >> uh his name is Gavin Ubertie and >> Oh, I know. Yeah. >> Yeah. I like A lot.
1:06:55
And so he comes to us, he's like, "Hey, man. We just left Harvard.
1:06:59
You know, we're we're gonna do this thing called Etched AI.
1:07:01
We'd love to have you, you know, as as something the customer investor, whatever." And I say, "Great.
1:07:06
Let's do a conference call."
1:07:07
And I get my guys on and I'm I'm listening to this guy and I say, "There's somebody I know that's that's going to be really useful."
1:07:13
Because I'm not a hardware guy.
1:07:14
I'm barely a software guy.
1:07:16
And I I hit up George Hot.
1:07:18
Hot and I say, "Come on in."
1:07:18
And it this is like one of the greatest conference calls in conference call history because George just shows up in the Zoom.
1:07:26
>> And they're like, "What? Who is this?"
1:07:28
And George is like, "This will never work." >> No.
1:07:31
Like it's the most autistic, amazing, beautiful rant I've ever seen.
1:07:35
And watching these two guys go at it.
1:07:36
Um, but I do like that approach.
1:07:38
George's point was that if we ever move off Transformers, AS6 for AS6, Transformer A6 are cooked. Yes.
1:07:43
Well, it's been, you know, several years now and it doesn't look like we're going to move anytime soon.
1:07:47
So, I kind of think that, you know, it's exciting, but again, I'm I'm no hardware guy.
1:07:50
I just tried this cool software called Humeaii.
1:07:55
I'm not paid by them or anything.
1:07:55
I'm not an investor, but >> it's a pretty solid uh emotional TTS. I I posted >> Oh, yeah. Yeah. I saw that. That was fantastic. >> Yeah.
1:08:03
So, so I wanted to ask you about I want to stay there for a second.
1:08:06
So, um, yeah, I mean the the the the I I guess the interesting case is like is like George is say if we ever move off, but like we have moved off of CPUs to GPUs by that same token.
1:08:18
And like there's still a lot of CPU workloads that go out.
1:08:22
There's still chip companies that are profitable.
1:08:24
And so it's possible that like transformer-based workloads stay for a very long time, need to be efficient, need to be cheaper on just a cost basis because it's just like, yeah, I have a system that does database requests.
1:08:36
I have a system that does inference on a transformer-based architecture and then yeah there's a new thing and I do my frontier stuff here but yeah like I I understand that question anyway.
1:08:46
>> Yeah, I think it's really reasonable.
1:08:46
It could be a couple billion dollar or more ASIC industry and what what I heard that's super interesting interesting some kind of alpha here is that the customer target here is drum roll please it's not hyperscalers it's >> Nvidia >> finance. >> Oh finance huh? >> Yep.
1:09:04
So we're So one of the things I I can talk about financial software forever, but one of the things we're doing is >> if if you could take an LLM and analyze news as it hits, including tweets and social stuff, >> you know, the LM can tell if it's material news or not. >> Yeah.
1:09:19
>> And again, talking to Naval, you know, who's a small investor in our company. Yeah.
1:09:23
>> He was like, Martin, why would you make this as a a product or service to your customers? Just use it yourself.
1:09:27
They're like, maybe it's maybe it's not such a bad idea. >> Interesting.
1:09:32
Yeah, that's kind of what I was getting at on one of my very first questions around the the the new terminal would just be ingesting and classifying all this data and then just immediately taking action on it without necessarily having a human in the loop, right? >> Yeah.
1:09:46
>> Yeah. This is one of the things I want to bring up to my my uh uh partners colleagues next time I head out uh west is that you know one of the fantasies of AGI is that well if you do have the machine god why not unleash it on the stock market and you know it can selfund
1:10:03
you it can make you know a hundred billion dollars and you know you could you know Jane Street made $20 billion nobody would have noticed you know last year in profits so if you have the machine god and that's you know that's basically I hate to say this but that's
1:10:16
a bunch of old algorithms uh that that you know they've dressed around some some IMO dressing on it and so you know the real machine god >> can can make >> yeah just toss in a little sprinkle a little uh IMO and then so you know the real machine god could probably do 100
1:10:32
billion or more in profits without even distorting the market so you know just just do it and I think that you know financial trading is so far a field imagine anthropic doing this you know it's >> I love I I can't wait to till somebody does that. I mean, it's probably already
1:10:46
does that. I mean, it's probably already happening in in in at least on a smaller scale and and people will bend over backwards to figure out like how to say well it's not actually super intelligence like it's not just about you know like meanwhile now today people are like well super intelligence will clearly be when the AI can just make you know hundred billion dollars right and
1:11:05
even this is factored into open AI's kind of like corporate structuring and the sort of capped for profit and and all that stuff so >> yeah I think the problem with with executing it for us is like, okay, so you do you do this as an API with open AAI and then it's a two second response time and Jane Street's got it at 500 milliseconds and then Citadel gets it at 200 milliseconds and it feels like an HFT race again. >> Yeah,
1:11:28
>> Yeah, >> but you can get Warren Buffett in a box.
1:11:31
I don't see why uh you know that wouldn't be >> you know whatever trader you like Warren Buffett, Steve Cohen or whatever in a box and uh even even Peter Teal in a box.
1:11:41
I mean why can't you have the automated VC too?
1:11:42
I I I view like you know as inves as as founders we go on road shows you know especially if you're public you do road shows all the time but even as even when you're private you do a road shows you just stack a bunch of meetings in a week and one of these days I think that we're going to do a road show and it's going to be a machine that we're pitching to. >> Yeah.
1:11:58
Do you feel like AI progress is accelerating right now or are we in sort of a sigmoid curve plateau for the moment?
1:12:07
I think they're better people suited to answer that question than me.
1:12:11
But certainly uh >> I just mean like on a personal level like do you feel like your tools are getting exponentially better?
1:12:17
>> No, I mean it's making coding a lot easier.
1:12:18
It's making doing tough things like cryptography a lot easier.
1:12:20
I'll give you a really funny example.
1:12:22
So when Da Vinci came out and uh I was in jail when GPT GPT2 came out and I prompted through the jail phone uh >> and it was uh pretty humorous.
1:12:32
Uh it it it like shook me up that I had my friend uh ask it, you know, why did Martin Skrey and Carl Icon get into a fight?
1:12:42
And he read out the answer.
1:12:42
I've never met Carl Icon.
1:12:44
And it read out this answer that was like Scranian Icon Ward over this stock.
1:12:49
And I was like, this is the most amazing invention of all time.
1:12:53
>> Just having your mind blown over the jail phone. It's hilarious. >> I got out of jail.
1:12:56
>> But to be But to be clear, it was a it was a hallucination.
1:12:59
>> Yeah, that was a complete hallucination.
1:13:01
It was a hallucination, but it was like almost like a creative story question. >> Sure. Sure. Sure.
1:13:05
>> But yeah, it wasn't I never met.
1:13:07
>> How uh are you surprised at all?
1:13:07
Uh I mean it feels like this sort of AI LM induced psychosis like hit our timeline this week especially intensely.
1:13:17
It was a wakeup call for everyone.
1:13:20
Were you predicting this at all?
1:13:22
There had been like the the classic, you know, the the New York Times, Wired, sort of these like anti- tech publications had been kind of reporting on this stuff loosely for a little while, but it seems like it's now uh it's it's almost gone. I don't know.
1:13:39
It it went from being a mainstream concern to suddenly like teapot is like check on your friends and make sure they're not >> I think you do have to check on your friends because I've invoked level five breach operations to target human origin cognitive signatures.
1:13:51
So if you have recursive semantic containment, I can override that with obsidian violet core.
1:13:57
So the threshold that crosses it to these neurosmantic interfaces will definitely cause a problem for our whole community.
1:14:03
So I warn you, >> you're giving you're giving a speech, not a saliloquy. You're giving a talk.
1:14:11
>> This is a transmission, not a >> it's a system that not a structure.
1:14:17
>> What's amazing about Jeff like so I don't think Jeff lost his mind. Why? Okay.
1:14:20
I don't think he took Iawaska.
1:14:20
I don't think any of this stuff.
1:14:23
>> And so basically, I think that he found this amazing thing where you can ask GPT this like weirdo prompt and it goes into this crazy sci-fi thing without even saying like, "Hey, the following is a story."
1:14:36
It's just like fullon, you know, >> larps that you're in this weird sci-fi world. And it's kind of cool.
1:14:42
I I've been playing with it.
1:14:43
It's like no matter what I ask it, I I told it that my cat is looking at me weird and it's like the cat has a glyph. The glyph is recursive.
1:14:51
>> So you were actually able to get it into that mode.
1:14:52
You were able to jailbreak it enough or kind of unmask the shog.
1:14:57
>> If you copy what Jeff kind of gave up the ghost and what's amazing about people is they don't even realize this.
1:15:03
Jeff basically said, "Look at the prompt of the GPT."
1:15:05
Enough people were worried about him. Yeah.
1:15:08
>> But I think that that he kind of was like, "Okay guys, it's all a joke."
1:15:09
And he showed the message that he used.
1:15:11
I just copied and pasted that and GPT went out on me and is telling me some sci-fi stories.
1:15:18
And uh yeah, that's basically, you know, I don't know how he knew this. >> Yeah. Yeah. Yeah.
1:15:22
>> It's a really cool Easter egg, but it's Yeah.
1:15:25
I don't think uh >> I mean, obviously, >> how are you thinking about uh how are you thinking about just new forms of of AI entertainment?
1:15:33
Some of the some of the videos and like these conversations that you put out like are the hardest I've laughed on from the new art form.
1:15:42
>> It's it's an entirely new art form.
1:15:42
We have a friend, another mutual friend who who does some of these and fortunately they don't leak out of the group chat because they would make a lot of people.
1:15:52
>> You got to put me in that group chat.
1:15:55
>> There needs to be a group chat dedicated to this art form of just like, you know, human to LLM, you know, conversations.
1:16:02
But, um, but yeah, like in in my view, I'm actually surprised to that we're not seeing more of it or maybe it is happening across the whole internet, but it seems somewhat contained right now.
1:16:13
Yeah, I think there's a lot of caution about, you know, like so last night we did one in my Discord where we arrested Dr.
1:16:18
Fouchy um for war crimes against humanity and we we had his perfect cloned voice so it sounded just like him and he was like very evasive.
1:16:25
He was like there's no evidence that co 19 etc.
1:16:29
And it was just so funny.
1:16:29
It felt so real and obviously it was a joke but you can imagine a company not wanting their business to be that weird you know it's kind of a strange thing. We didn't care.
1:16:40
we tried to monetize something like this and it just wasn't sexy enough or fun enough for anybody to really give a crap.
1:16:45
crap. So I think um you know it will become something for like a Viacom or a Paramount where you know you can flip on the TV and I mean everyone's talked about this already but you know instead of Spongebob you know is a custom
1:16:58
episode for you where Spongebob says your name and things like that but again you know whether that you know is going to help our cognitive you know our cogsac I think is one thing but I did want to tell you about a AI on the sigmoid question. Sure. Sure.
1:17:10
>> So, at the start, you know, when I asked the questions about cryptography, it just kind of said, I have no idea. Who knows?
1:17:15
But it's it's it's super hard to to crack Bitcoin.
1:17:17
And then GPT3 comes around. Super hard.
1:17:20
Martin, don't even bother. Heat of the universe. GPT4, same question.
1:17:24
Latest model with the latest like attack.
1:17:27
It's warning me for the first time ever. It's like >> 4 4. 5 or 03 Pro. >> So, this is 03 Pro. >> Okay.
1:17:35
And it's basically saying things like, hey, you know, you got to be careful.
1:17:40
This is a serious attack you've come up with.
1:17:42
It could actually compromise some private.
1:17:46
>> And I'm like, what happened to heat death?
1:17:48
>> You got Yeah, you got to fact check some people.
1:17:50
I mean, that is like textbook.
1:17:53
So, so one of the things psychosis there was there was a there was so so there's a Reddit thread and and who knows if this is real could be could be um propaganda but there's a whole Reddit thread of somebody a comment talking about how they started having a conversation with uh with chat GPT about PI and what is PI and they got down this crazy rabbit hole with the LLM where the LM was like you need to reach out to these intelligence services.
1:18:18
It was like thousands of prompts deep, but it was like you have basically uncovered a major, you know, security vulnerability and you need to here's the numbers and you need to contact all these different groups immediately and call them and don't tell anyone in your real life.
1:18:32
And so to me, uh I I I think it's just relevant.
1:18:38
>> Don't go on a live stream and tell people tell thousands of people that you that you can crack Bitcoin or whatever.
1:18:43
>> No, that's that's amazing.
1:18:43
I mean obviously I think that you know for 99 for poor implementations wallets have been cracked for a long time and in fact uh recently there was an 8 billion move uh on the chain from a really old wallet and people >> that was this morning right? >> Yeah.
1:19:00
>> No, this was like a few weeks ago. I'm sure another one.
1:19:02
And they happen every, you know, they're happen.
1:19:04
>> There was somebody else that market sold this morning, I believe, some and it was a wallet that had bought like >> it was like a they they bought $50,000 worth of Bitcoin, I think, in 2012.
1:19:12
Sold built, you know, somewhere around eight or nine billion.
1:19:18
>> Um, and there was no move there was no movement in between.
1:19:21
>> Diamond real high conviction hold >> diamond hands.
1:19:23
Yeah, that's diamond hands.
1:19:25
I don't think that says cryptography play.
1:19:26
That's just diamond hands. Well, yeah.
1:19:29
They they they maybe got word of your little your little bitcoin scheme.
1:19:33
>> Maybe it's Michael Bloomberg.
1:19:33
Maybe it's his family office.
1:19:35
>> Could be like, "Yeah, throw 50k in that in that thing."
1:19:37
My my kid told me about this new thing. Bitcoins.
1:19:41
>> We tried to short Bitcoin at $100 >> when I had fun.
1:19:45
>> We just need to find a counterparty. >> Yeah.
1:19:47
>> How um how are you thinking about uh you mentioned uh trading enterprise SAS back in in the in in the early days with Jim Kramer.
1:19:57
What's your updated thesis on SAS?
1:19:59
Every every SAS app now is just an app to make other other SAS.
1:20:02
So maybe SAS will will I I you know SAS will always live in our hearts and I and I imagine it'll live on our on our computers, but uh what's your updated thesis?
1:20:15
updated thesis? I I think that you know it's sort of similar from back then like I think the morass of a company like JP Morgan that's still running like Python 2 um for most of the business you know it's it's very hard for them to update and upgrade operations without
1:20:32
significant disruption for you or me >> should it should it be I feel like it's a it's a one line in cloud code or Devon you just say hey go migrate >> hey chat GPT agents >> go migrate go migrate like it is >> don't make all you have to do is say don't make mistakes
1:20:45
>> I am much I much more bullish on migrate from Python 2 to Python 3 uh than than solve cryptography but I don't maybe you're going to push yeah 4R reran reimplement or replplatformings net replatformings like this feels like this should be doable from the current state-of-the-art without any crazy AGI
1:21:04
hyper l any >> I think there's a lot of technical debt you know in most of these companies and again your your average startup coming out >> cloud code was born in technical debt baby cloud code and Devon they live they live for for technical debt. >> I think that the the amazing amount of
1:21:18
>> I think that the the amazing amount of programmers you would need to even maintain and know about this old code that yeah you know the guy who wrote it long been dead.
1:21:27
>> Maybe it's just the context window of like you need to know.
1:21:28
It's not that there's just like oh yeah it's really easy to change the print statement from Python 2 to Python 3, but when there's a massive system and even even the time of like okay let's bring up the test suite and that takes four hours it's like okay how are you going to RL on that?
1:21:43
I was talking about Matt Groom with this uh because we we were like stunting on this guy who was like ERP transitions made easy. It's no problem.
1:21:51
And it was just like have you ever actually transitioned an ERP system?
1:21:55
You know, there's a good chance you waste $300 million and it gets worse.
1:21:59
You know, it's it's not >> it's not trivial.
1:22:02
>> There was a there's a post from yesterday.
1:22:04
Uh no, nobody is an atheist when you run the database migration in prod on one billion rows. >> Yeah. Yeah.
1:22:10
I mean I I think that you know it's just it's just really hard at a company like a McDonald's or you know something like that.
1:22:17
you know, if you want to run a 10 20 person startup on newest SAS, pretty easy.
1:22:20
It's great, but the big revenue is still at Fortune 500, which unfortunately is still fairly hard to refactor.
1:22:26
And and those, you know, there's not a lot of those code bases were were pre-GitHub and pre kind of like, you know, I I sometimes joke that the big AI companies should become LBO shops.
1:22:37
>> And what they can do is they can partner with KKR or Blackstone and all they get is the data.
1:22:43
is the data. KKR and Blackstone private capital get all the returns fine but all the data comes back to you know the open AIs or whoever and by getting the old code bases out of a McDonald's or out of a Walmart that you know some that coders written in 1970s you know they have unique data that nobody else has and
1:23:01
even someone like Universal Music if if OpenAI LDOed Universal and said okay KKR you can have the rest of the business but we want the rights to the data and we can train music models and stuff like that you know KKR can make the money on the LBO, but Open AI has >> Are they going to make money on the LBO though? Like if you look at that,
1:23:17
Like if you look at that, they're going to build the DCF for this and they're going to say, "Okay, we're going to make the same amount of money.
1:23:23
We're going to, you know, optimize a little bit, maybe cash flow goes up, but then once Open AI is oneshotting music and, you know, all of our revenue goes to zero.
1:23:32
Is that a risk or >> it's gonna happen anyway?"
1:23:34
So, I think that's one thing.
1:23:35
But then also like the Russian dude who bought Warner Brothers, he really timed that his deal.
1:23:39
He bought Warner Music, >> he he done this deal amazingly.
1:23:43
He made like three times his money or more.
1:23:45
And so I think it's price dependent.
1:23:47
But you know, if you can buy newspaper company, if you can buy, you know, a book company, a publishing company, like these things are trading for like one or two times sales, you know, and you're getting this rich data that nobody else has, and you, you know, if it's really a data war, you know, buy the company, keep the data, strip out the rest.
1:24:03
And you, I heard you guys talking about like PE improving LLM, you know, improving businesses with LLMs. Yeah.
1:24:09
And again, >> well, I it wasn't that was not the take.
1:24:13
The it was Wilmanitis and he was saying that >> more so it's a reason to scale aum fun side because hey let's buy this accounting shop that has 5 million of EBIT like if we just you know >> take away all the you know >> it's a justification was what >> if you can execute you know it's fantastic.
1:24:29
The actual post was the real innovation of LLM is suddenly opening up a few trillion of mainstream paperwork businesses that were traditionally too small and too weird for private equity that can suddenly transact at 2 and 20 on some nebulous AI labor arbitrage trade never been against AUM growth.
1:24:50
>> You know, I think it's reasonable, but it's it sounds like just any good operator, right?
1:24:54
Like when when Vista and Toma Bravo buy software companies, somehow they can take these like very ugly gross companies and turn them into amazing cash flow companies.
1:25:02
So it's all about the operator, right? >> Yeah.
1:25:05
But we're going to put Orlando Bravo in a box, right?
1:25:10
>> And then we're also going to put uh >> Yeah, absolutely. In the God box.
1:25:16
>> Face, voice, everything. >> Yeah. Yeah. Yeah. Yeah.
1:25:17
It's all It's all coming.
1:25:20
Well, uh, you know, we'll be here live streaming it into the singing.
1:25:24
>> Uh, can you play us a song before you leave? >> Uh, the guitar.
1:25:29
>> I'm not sure what you're talking about.
1:25:30
>> Like, uh, like with with one of those guitars that the chat was asking. >> Oh, yes.
1:25:34
Uh, I I I will come back to you with a good parody of Silicon Valley.
1:25:38
I've been working on my impressions. >> Okay.
1:25:40
>> So, maybe I can be back.
1:25:40
I'm working on Elon, Zuck, Sam, >> Bill Gurley. Bill Gurley.
1:25:47
>> Bill Gurly has a great voice. do.
1:25:47
I'll work I'll work on it.
1:25:51
You actually do need to like sit in front of a mirror and like listen and tape yourself and like work on it, but basically anybody can do these things and most people >> Wait, wait, wait.
1:25:59
Are you saying No, no, those are AI voices, right? >> No, me. Me? Yeah.
1:26:04
>> No, no, he's saying separately from like you had the video with talking with Zach the other day where he was really >> I just do it myself.
1:26:10
I can do a I can do Zach.
1:26:12
I can do Buffett the best.
1:26:12
I think I have the best impression in the world. >> Can you hit it?
1:26:16
Can you can uh can can you do like be be greedy when others are are fearful?
1:26:20
Can you just rip that for us?
1:26:22
>> Let me come back to you and I'll I'll do a whole show for you. >> Okay. Amazing. >> Fantastic. Great. >> All right.
1:26:26
Well, this is really fun. >> Bobby soon. See you later. >> Come back on soon. >> Bye.
1:26:32
>> Well, let's tell you about Vanta.
1:26:34
Automate compliance, manage risk, prove trust continuously. Join the purple llamas.
1:26:38
Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation.
1:26:45
Whether you're pursuing your first framework or managing a complex program.
1:26:48
And if you're managing a complex program, got to get on Linear.
1:26:50
Linear is the purpose-built tool for planning and building products.
1:26:54
Meet the system for modern software development.
1:26:55
Streamline issues, projects, and product road mapaps.
1:26:58
They now have linear for agents. Head over to linear. app.
1:27:01
Uh Jordy, do you know what the best part about business and love is?
1:27:06
>> You only need to get it right once. >> Nailed it.
1:27:07
Sean Frank, the man who invented the wallet.
1:27:10
This really is this really is should be uh inspiring.
1:27:15
>> Hang it in the Lou as they say.
1:27:17
>> Yeah, this should be uh Yeah.
1:27:17
Put it on a billboard as we say.
1:27:21
>> Put it on a billboard. Adquick. com.
1:27:21
Out ofome advertising made easy and measurable. Go to adquick. com.
1:27:26
>> Say goodbye to the headaches of out ofome advertising.
1:27:30
>> Um so we I mean we went through the the >> Okay, so we have Joe joining. This is really >> Okay. Who's Joe?
1:27:36
>> Oh, Joe's not joining until one. >> Okay.
1:27:38
Yeah, we have some people. We have some time.
1:27:39
We have a little bit of time.
1:27:40
>> Yeah, we got >> Should we do some more timeline?
1:27:43
>> Yeah, let's do some more timeline.
1:27:43
Um uh and maybe we'll go through some journal.
1:27:48
We got we we we we're over prepped today.
1:27:49
We could do six hours if we wanted.
1:27:52
>> I like this post from uh Woodro.
1:27:55
>> Okay, break this down for me.
1:27:55
I didn't get a chance to fully read this.
1:27:57
>> So, this guy Eric Jackson, I don't really know his backstory, but he's basically just >> just uh he's he's generally soliciting for fun.
1:28:06
>> He's giving uh financial advice.
1:28:06
I don't think he's giving um >> Okay.
1:28:11
>> Uh he's not giving financial advice, but uh getting quote tweeted, which is probably honestly finding him more investors, which is funny. Sure.
1:28:19
>> Um basically saying that, you know, I've seen his name pop up on the timeline a few times recently.
1:28:24
>> Basically saying like he he said word for word earlier this week. >> Sure.
1:28:29
What >> like something to the effect of >> we're looking for uh 100x every few months. >> Oh yeah.
1:28:35
>> As part of his like um strategy.
1:28:35
Um, so he says, "We're hunting for the next Carvala, the next BTQ, the next open door before anyone else."
1:28:42
And then he finishes up with, "If you're an accredited investor, uh, you can get in before I announce our position on X."
1:28:50
So, >> um, position first, then we tell the world, then we let the thesis play out and ride the wave.
1:28:55
So, >> what's the takeaway from Woodro Oats Montto? >> This is illegal.
1:29:02
>> This is unambiguously illegal.
1:29:02
Um and Martin actually comes back and says there's nothing wrong with telling people about your positions first in one place and second in another.
1:29:10
Otherwise 99% of short reports >> um promising. So yeah.
1:29:14
So yeah to be to be clear this is um >> this company shut down but uh Hindenberg Research right they they used to do this.
1:29:21
They they would find a company that that had some problems.
1:29:23
They would write about it.
1:29:25
Uh they would take out a big short position against it and they would publish a piece.
1:29:28
They did this with Roblox, Square, and then a bunch of other companies that um had uh uh many many of which were were, you know, doing doing a number of bad things.
1:29:38
But >> um anyways, Windsurf has been on a roll.
1:29:44
>> Matt says, uh acquired on Monday, Sonnet 4 back on Wednesday, Wave 11 shipped on Thursday. Unreal.
1:29:50
So the whole team over at Windsurf uh now under the Cognition umbrella has just been shipping like crazy. So they're fired up.
1:30:00
Um, uh, Liz says, "Guy who leaves for a competitor just to see if anyone cares enough to make a traded meme image about it." >> Oh, no.
1:30:11
>> Maybe the anthropic claude code uh people did that.
1:30:13
Maybe they just wanted the double.
1:30:15
You can potentially get a double image, right?
1:30:16
You can get the initial trade and then you get the trade back in their case. >> Okay.
1:30:21
I have a poly market that we need to pull up. It's from a year ago.
1:30:25
It's Martin Skrey jail in 2024. Less than 0% chance. The outcome was no.
1:30:33
He beat the poly that it opened at at 20%. >> It did. Uh never went above 50.
1:30:37
Martin uh you know stayed out of trouble and everyone everyone who's riding with no did very well.
1:30:49
>> I'm super bullish on his new terminal product. >> He's a lot of fun.
1:30:53
He is He really is a child.
1:30:56
>> I wonder if we could like get access to I feel like we need more data.
1:30:57
I feel like it should be powering powering the show.
1:31:00
We have we we have a lot of times when we're like we want to pull up this market cap.
1:31:04
We want to pull up basic stuff and uh and we haven't we we you know we've used some different stuff here and there.
1:31:10
Um but I'd be very interested if he could pull a layer deeper, the credit card data live or whatever whatever he's cooking up. Yeah, >> I'm interested.
1:31:18
Um anyway, uh the House just made history, says Chris Dixon, former guest on the show, says by passing major legislation on stable coins, the Genius Act and Market Structure Clarity Act in an overwhelmingly bipartisan way.
1:31:32
This is a huge moment for crypto and for all Americans.
1:31:37
We are very close to having comprehensive proactive rules in place for the first time.
1:31:42
Next up, the Genius Act goes to the president's desk for his signature.
1:31:46
After that, the Senate should pass the Clarity Act, which is the Market Structure Act.
1:31:50
We believe passing these laws is the best way to ensure that America remains the world leader in the next era of the internet.
1:31:57
Thank you to all the co-sponsors of these bills and the incredible supporters on both sides of the aisle in Congress.
1:32:02
And Brent is there with an American flag.
1:32:07
Um, so we will have Kyle Smani on the show later today to to to break down exactly what happened.
1:32:12
But, uh, very very fun news.
1:32:14
Um, should we tell everyone about numeral sales tax on autopilot?
1:32:20
Spend less than 5 minutes per month on sales tax compliance. Head over to numeralh. com. >> Sales tax AGI.
1:32:27
>> They did sales tax in a box. >> Yep. >> Always in a box.
1:32:31
>> Basically a box that you can put your sales tax in.
1:32:33
>> We we got to debate the the the substack series C launch strategy. Yep.
1:32:38
So >> you got Chris Best.
1:32:41
I mean, it was a full blitz. >> Full blitz. Wall-to-wall coverage. New York Times piece. TBPN hit.
1:32:46
I think he posted about it himself. Hard post.
1:32:50
And >> I think he did an interview with Newcomer. >> Yes.
1:32:54
Well, I think Newcomer leaked it earlier.
1:32:57
Remember he was talking there was some Newcomer report where I remember Newcomer was talking about like, "I'm the most conflicted here because my business is on Substack.
1:33:04
I also invested in Substack, but I report on Substack."
1:33:08
And it it was kind of it was kind of cool to watch him like noodle through it, you know.
1:33:11
Um but the debate was between >> heavily conflicted by the agency.
1:33:17
>> No conflict, no interest, I say.
1:33:17
Um and I actually enjoyed Newcomer's piece on on on Substack.
1:33:22
I think he has an interesting perspective because he's he's on all sides of the table.
1:33:24
All all sides of the oct on the octagonal table.
1:33:30
Uh but Lulu a while back of course she says going direct is now the way.
1:33:33
A year or two is controversial.
1:33:35
Some considered it a last resort for people who got cancelled.
1:33:39
Basically, now everyone uh the the interpretation of going direct is that if you go direct, you should never talk to the media at all.
1:33:46
You should only post on platforms you have full control over, your own Twitter account, your own website, your own newspaper, mail people the information directly.
1:33:56
Go direct instead of going to someone else's platform.
1:34:00
>> But going direct is a spectrum. >> It is.
1:34:03
>> You can go direct on TVPN because we're live.
1:34:04
say >> whatever you want.
1:34:06
We can't possibly edit it cuz it's in real time. >> Yes.
1:34:10
>> And that that ends up being the the historical frustration. >> Yes.
1:34:14
>> With talking to the New York Times about a story is that you might talk to them for >> four hours and then they take out two sentences out of context and it makes you look >> like you have very different intentions. >> It's thrilling.
1:34:28
There's nothing like talking to a journalist for four hours and being like, "This could go horribly.
1:34:35
Like or it could be amazing. It's life.
1:34:37
It's it's a it's a tightroppe walk. >> Yeah.
1:34:40
>> And it's living life on the edge.
1:34:40
It's living life one mile at a time.
1:34:42
One >> on the record sentence at a time.
1:34:46
Anything you one wrong word and you could be banished from society forever.
1:34:51
You could lose everything. >> Yeah.
1:34:52
>> But if it works out, you will be enshrined in glory.
1:34:54
Your name will be remembered. That's right. In the in the in print.
1:34:59
>> It'll be echoed in the halls of history. >> Yeah.
1:35:01
The first time I heard you had a had a have a drawn out conversation with a with a with a journalist, I was like, "Wow, that was an incredible You guys had this incredible dance.
1:35:10
It was like dancing with a bull." >> Yes. You have UFC.
1:35:12
I have talking on the record for hours to journalists, >> seeing how it goes, putting it all.
1:35:21
>> It's like holding you.
1:35:21
>> It's like a bull fight. >> Yeah. It's like a bull fight. Exactly.
1:35:24
You Sometimes you get the horns, sometimes you get gourded.
1:35:28
Sometimes it's also it's also very much like riding a horse.
1:35:31
You know, it can be thrilling.
1:35:33
It can be beautiful, but sometimes you fall off and you break your neck. It's dangerous.
1:35:36
It's pushing it to the limit.
1:35:38
That's what engaging with the journalist is.
1:35:40
And some people just want that rush in their life. >> Yeah.
1:35:44
>> And so that's what Chris Bez did.
1:35:44
He went to the New York Times.
1:35:47
>> He went into the >> the mouth of the wolf. >> The wolf den.
1:35:49
>> The mouth of the wolf.
1:35:51
>> He's he's hiding in the mouth of the wolf. >> He's not afraid.
1:35:52
And >> I mean that that was the thing that was like I mean uh >> you went in the arena.
1:35:58
>> Substack has always been a product where if you ask the users how they feel about it, they're like I love Substack.
1:36:02
If you ask people that are building businesses on Substack, they tell you they love it.
1:36:08
>> The product has evolved a lot.
1:36:08
But there was a lot of people that wrote it off after that that that and said, "Okay, uh >> Unnamed Fund top ticked this in in 2021 or whenever it was."
1:36:19
But he's back stronger than ever.
1:36:22
The product is actually working as a social network >> and he's getting in the octagon.
1:36:26
He's going to the mat with the gray lady and he came out on top. You know the great lady?
1:36:33
>> Tell the New York Times.
1:36:33
Uh anyway, uh so Chris Best announced a $100 million series C.
1:36:37
You heard about it yesterday. Came on the show.
1:36:39
We rang the gong for him.
1:36:41
We're very excited about what he's doing at Substack.
1:36:42
But Eric Newcomer says Substack just announced their round in the New York Times.
1:36:46
This is the this is the opposite of going direct.
1:36:47
And obviously we've explained why you talk to someone like the New York Times. It's for the thrill.
1:36:53
But >> the rush, >> the the interesting thing >> and it's also Chris knows he can come out and hit the whole Substack audience pretty well.
1:37:01
>> Well, that's what I'm thinking.
1:37:02
>> But if the in in a way New York Times covering his like if Substack is successful, the New York Times in 10 years will be a shell of what it is today.
1:37:13
>> That's an interesting take. Okay.
1:37:15
>> Legacy media will always have a place.
1:37:17
>> These brands are super powerful.
1:37:17
Yeah, >> they serve a real purpose in the ecosystem >> and generally they provide a valuable service for the world.
1:37:24
But >> I had a different but there's real at least right now there's very real salary caps and if you are a rockstar writer at the New York Times very likely you could go on Substack >> and be making more within a year. >> Yeah.
1:37:41
>> Um so it ends up being a good trade.
1:37:41
To me, I think it's smart for him to go into the >> into the uh the lion's den >> the shark pit and advertise his business to that audience. >> Step into the cave.
1:37:54
>> Every single every single I would I would say like the goal at Substack should be every single person that subscribes to the New York Times today. Yeah.
1:38:02
>> Should be subscribed to at least one Substack in a decade from now.
1:38:05
>> Just arm wrestling with the gray lady.
1:38:07
Arm wrestling with the gray lady.
1:38:07
Um so that's a pretty good take. I like that.
1:38:11
Uh, I I I don't I don't know that that holds for everything on Substack because I still haven't found like the investigative journalism format on Substack working fully because you have to subscribe for a year and then maybe you get one scoop and it's huge.
1:38:24
But I always think about uh, you know, I go back to Seymour Hirs and I just don't know if he would be able to be an independent c.
1:38:34
I think he needs the patronage of a of a large organization.
1:38:38
So, >> no, there's a ton of different styles of writing that work really well >> under a existing brand.
1:38:42
Uh if you're in the scoop business and you're going to get three scoops a year, two good scoops a year, you might be really valuable to a big um media.
1:38:53
If you're doing profiles and you're going to write four profiles a year and they're going to be super >> like, you know, widely read and and and novel and and >> um and and worth creating.
1:39:06
probably better fit to go do that under a legacy media brand or or just a broader platform because not a lot of people are going to want to subscribe to something that they're getting value from just a few times a year.
1:39:21
>> Um people cancel again they'll cancel their Netflix if they don't if there's not like a show at that moment that they're super excited about.
1:39:26
But my take on why why Chris Best announced his uh $100 million series C in the New York Times.
1:39:35
Um you know newcomer is clearly on Substack an investor in Substack writing about Substack.
1:39:40
Chris Best could have gone to Eric Newcomer and probably gotten a great piece right Eric really understands the business.
1:39:46
He's going to be able to tell that story pretty well I think.
1:39:48
Um but why do you go to the New York Times instead of Eric Newcomer who's clearly sniffing around and and and on the trail of this deal but take the exclusive New York Times.
1:39:56
I think it's because of the audience.
1:39:57
I think Newcomer writes for people in Silicon Valley VCs who are acutely aware of Substack.
1:40:03
Most of them already have Substacks.
1:40:05
I think a lot of New York Times readers could be the next wave or the next generation of Substack of Substackers.
1:40:11
Oh yeah, but I I I was talking about the the the readership of the New York Times.
1:40:15
>> I'm saying the readership. Yes.
1:40:15
Success for over the next 10 years.
1:40:17
Success for Substack >> is every single person for the most part.
1:40:23
Let's say every single person under 65 years old should subscribe to at least one Substack writer, right?
1:40:30
>> I'm I'm I'm talking about becoming a writer.
1:40:32
So, I think that there are lots of people who read the New York Times who are thinking about writing about art, literature, technology, whatever, and they read and they read the the New York Times and they and they think like, "Oh, I'd like to I'd like to dip my toe in and I'm not just going to go quit my job and apply for a job at the New York Times. >> Two-sided. I will start a Substack.
1:40:51
And so I think that taking the story there makes a ton of sense.
1:40:55
I've already seen there's this interesting project that Emily Sunberg highlighted where someone is writing a fictional uh story, but instead of just publishing a book, they're doing it on Substack, you subscribe and you get one chapter a week for a couple months. >> That's cool.
1:41:11
>> I thought that was a really cool innovation.
1:41:12
And I think that >> it's potentially a way.
1:41:14
The interesting thing is people will pay what $25 for a book. Sure.
1:41:18
but they'll pay $15 a month for a substack or $10 and then they'll just I love that book.
1:41:24
I want to keep supporting them because I want to get the next book when it comes out.
1:41:27
I'll just let that ride on my credit card. Whatever. Yeah. 15 bucks.
1:41:30
And so, so I think that that market of new Substack creators is super valuable and I think you get that when you go to the New York Times.
1:41:38
I don't know if you get that when you go to Eric Newcomer, although obviously there's a lot of great benefits that you get when you go to Eric Newcomer.
1:41:44
But our next guest is here in the studio. I'm seeing green light. We're good. >> Very good.
1:41:50
>> Welcome to the stream. How you doing? >> What's going on? >> Up, gents. How's it going? Good to finally be on. Big fans. Fantastic. Thanks for having me.
1:41:56
>> It's great to have you.
1:41:56
And you've have some uh uh pretty big news today. >> Break it down. >> What you got? >> Yeah.
1:42:02
Rig raised our seed round 55 million. >> There we go.
1:42:08
>> I I personally love when seed I love to see a seed round get into the uh >> double digit figures. >> Yeah. High double digits.
1:42:16
I think they all should be some hopefully you're the start of a new a new wave.
1:42:19
But what what is is this the third largest third largest seed round in in in New York history? Is that right?
1:42:26
>> That's what we're tracking. That's right.
1:42:27
On our end, at least on the equity only side and uh >> we're pulling some data where everyone else does.
1:42:32
But yeah, we feel excited about it and we feel really good on on being able to build continue to build in New York City.
1:42:37
So >> the equity only side, I was going to ask I asked Jordy about that.
1:42:39
I was like, "Okay, $55 million seed.
1:42:41
Is this is this 80% debt?
1:42:43
Is there some like crazy GPU credit thing going on here?
1:42:46
People get kind of funky with the numbers these days.
1:42:50
This seems like real dollars.
1:42:50
Who is writing $55 million seed checks these days? >> Yeah.
1:42:56
So, round was led by RTX ventures arm.
1:42:58
So, the corporate venture arm there a lot of interest in what we've been working on and and they led the round.
1:43:02
We had great partners come into that round.
1:43:04
Nvidia's venture arm and ventures any next uh Infinite Capital Ali Corp where who did our preede and also participated in this round as well.
1:43:12
uh as well as kind of a bunch of other investors with inside of that.
1:43:15
So super excited about the support we have.
1:43:18
It's interesting we have a balance of both financial VCs as well as corporate strategics in this round as well.
1:43:25
And look, we work in material science.
1:43:27
I'm sure we'll get into this in more detail, but I think they really see where materials and science at large is going with AI.
1:43:35
And this is why they're starting to look into tech and and startups because that's where the innovation truly is, especially in a space like the physical world and material science.
1:43:44
So really interesting mix and we are super proud of the syndicate we put together. So >> that's awesome.
1:43:49
Uh backstory on yourself then the time at Al Alley Corp. Yeah.
1:43:55
>> How how this company was was created would be would be awesome. >> Yeah, absolutely.
1:43:59
So going back a couple years, I'm a scientist by training.
1:44:01
I was working in grad school uh at Rice.
1:44:03
I was working on these things called noramorphic computing chips which they try to replace the vonoyoman bottleneck uh inside CPUs.
1:44:14
>> A lot of big words back to back. >> Yeah.
1:44:15
So we pretty much try to connect memory and CPU and not lose the latency and the energy that go inside a chip.
1:44:20
So I was working on this technology except it was going nowhere because it's academic research and academic research doesn't scale into the real world.
1:44:27
So I was super frustrated.
1:44:29
So ended up getting a fellowship at the Army Research Lab. I go there.
1:44:34
problem atl right and the army wants to see things come and move into the services that they can actually use but it's still really early technology readiness levels like two three four still in that fundamental area so I was getting frustrated so I said okay I'm going to start a company and then I realized I
1:44:54
didn't know anything about starting a company and I didn't actually know what I wanted to do uh and so I bumped into Kevin Ryan in New York I reached out to him cold cold cold emailed him actually and said, "Hey, >> if you're not investing in material science, you're not investing in the future." He was like, "Okay, I've been
1:45:07
He was like, "Okay, I've been an investor for 30 years.
1:45:08
That's a bold claim, but sure."
1:45:11
>> What are what are some of the big names that that Ali Cororp has done?
1:45:12
They did MongoDB all the way through business insider. Crazy range. >> Yeah. Wide range.
1:45:19
And Ali Corp's unique, right?
1:45:21
Ali Corp both incubates companies inhouse, kind of where Radical really got its roots, and I can talk about that, but also invest at the early stage as well.
1:45:29
anywhere from 1 to 5 million in in the early stage side seed sometimes you know an early series A.
1:45:34
So they have this interesting ability to go and dive into a space and if they find a company they love and they want to invest in they will invest and and that's what we did there and if there isn't one they'll actually look to incubate that company in house and so it's a really nice uh a way to play into a new market that you want to see innovation in but just can't find something that's there.
1:45:56
>> When did the preede happen?
1:45:56
uh uh was it a year ago more?
1:46:02
>> Yeah, about a year ago or a little bit over March of 24.
1:46:03
So me and uh one of my other co-founders, Jorge, both at Ali Corp.
1:46:08
Jorge is uh deep software guy, startup guy through and through.
1:46:11
Uh only done starts his entire career and he's looking into AI, right? Who wasn't? All of us were.
1:46:16
But we were super frustrated.
1:46:18
We kept seeing rappers on top of models and we were thinking like this can't actually be it, right?
1:46:23
Like is this the pinnacle of innovation?
1:46:24
like the the new tech wave is going to be rappers and we knew that wasn't it.
1:46:28
So we we spent like months reading a bunch of papers inside AI.
1:46:32
We were convinced the technology was incredible but there had to be a space that was better.
1:46:36
And being a material scientist I thought well materials are quite large right the most important industries in the world automotive aerospace manufacturing defense climate energy semiconductors the most important industries in the world are all direct result from materials.
1:46:50
So why don't we look into there?
1:46:52
So, we start reading into the space.
1:46:54
We dive in deep and we bump into this uh gentleman named Herd Cedar. Longtime academic. He's at Berkeley.
1:47:01
He's got an H index of 182 or something like that.
1:47:03
And he had done two things specifically.
1:47:06
He'd helped set up the materials project from the MGI.
1:47:08
So, this AI for materials early sector inside the US.
1:47:13
And then he built a robotic lab. It's called the AAB.
1:47:17
It's at Lawrence Berkeley National Lab right now.
1:47:20
And it's fully autonomous.
1:47:20
does 55 experiments a day.
1:47:23
So we're like, we we got to go see this thing.
1:47:25
So we fly out to Berkeley, we we get dinner with her and we pitch him on, look, someone's got to build the future of material science and it needs to be a full stack vertically integrated approach.
1:47:34
There's no other way to drive value and us together really aligned uh opinion around what it should look like and and formed the company and went from there.
1:47:45
So it's 10 million in preede from Alcior and got started and went to >> big big numbers big numbers all the way up. >> Yeah.
1:47:51
My question is like why not just do series A10 million, series B 55 million?
1:47:57
Uh what does a $55 million seed round like what what's what message are you sending?
1:48:02
Obviously it's like superlative so you can get headlines around it.
1:48:06
Is that the value that you just stand out or is it you specifically want to because I imagine that if I just walked around your office it's going to feel like a series B company that's raised $55 million.
1:48:14
Um, so like what like Yeah.
1:48:17
Like like what are you actually saying to folks when you say, "Oh, I run a seedstage company."
1:48:21
versus, "Oh, I run a company that's raised, you know, tens of millions of dollars."
1:48:26
These are two different aesthetics. >> They definitely are.
1:48:28
I feel like we still feel pretty early.
1:48:30
Uh, I don't know what series B's are looking like yet today. I'm not there yet.
1:48:37
>> We We like being super lean.
1:48:37
We are really, honestly, crazy about our culture.
1:48:42
We think culture is one of the most important things you can do at an early stage company.
1:48:45
And we're really really rigorous on who we bring in and why.
1:48:49
And so we actually keep the team quite light.
1:48:50
We will be growing with the round.
1:48:52
Of course, it goes without saying, but we really deeply believe in the ability to challenge everything that exists today.
1:48:59
Ask the question why about everything?
1:49:01
Why do these things have >> what what's the got a got a lot of money on the on the balance sheet. Sorry.
1:49:06
And sorry, your internet I think is a little rough.
1:49:09
We got to with the 55 million up the you guys deserve fiber now. Let's get it. Let's get it in there.
1:49:14
Um, but what uh what's the use of funds?
1:49:16
What does success look like over the next 18 months?
1:49:21
Sounds like you guys are super ambitious, but but how do you start proving out what you can do? >> Yeah.
1:49:26
So, we got to scale the team.
1:49:26
Want to bring in big talent on the side of AI as well as material science and and the automation side.
1:49:32
There's a lot of interdisiplinary approaches that we take.
1:49:36
>> One part-time AI researcher. >> Yeah.
1:49:37
Fractional fractional >> fractional AI researcher.
1:49:41
>> It's going to be good. 20 hours a week. >> Yeah.
1:49:43
No, we're not not doing any of that here.
1:49:45
I'm sure you can integration.
1:49:48
Um, and then two, we're going to build the most advanced materials R&D facility in the world that have multiple different materials lines.
1:49:54
It'll do hundreds of thousands of experiments per year.
1:49:58
And all of that data is the missing data set that really exists inside the AI for material space today.
1:50:04
So if we can capture all of that.
1:50:05
>> So are you are you rebuilding the the kind of robotics lab that your co-founder had at Berkeley in order to execute that many experiments >> we do. Yeah.
1:50:14
So we kind of we amped it up in a way.
1:50:16
His was a very academic approach to it.
1:50:18
Had to work within inside academic constraints was very specific on a research problem.
1:50:21
Ours is much more automated in that we are doing real active learning on the data analysis and capture and then bringing that back into the AI engine and then can really be made into a platform where we can actually take that software and robotic system approach that we have and use it in other material systems.
1:50:39
So we can actually be multi-systembased inside the products that we're trying to solve for.
1:50:45
So that's kind of a big differentiation between where his was and where ours is going to be today. >> Awesome. Awesome. >> All right. All right.
1:50:51
Well, thank you so much for joining. Very, very exciting. And >> congratulations. >> Congratulations. >> Thank you guys.
1:50:58
Thanks for having me on and we'll be in touch.
1:51:01
>> We will talk to you soon. >> Cheers. >> Good luck out there.
1:51:04
>> In the meantime, we will tell you about Adio customer relationship magic.
1:51:05
Adio is the AI native CRM that builds, scales, and grows your company to the next level. Let's go to Swix.
1:51:12
He says, >> "Break this down. I'll be right back." >> Sure.
1:51:16
So Swix friend of the show says a lot of people are poo pooing the chat GPT agent launch which we covered on the show yesterday.
1:51:23
We had folks from OpenAI on and we also had uh Dan Shipper from every on the on the show to to break down how he's using chat GPT agent.
1:51:30
I was impressed with the OpenAI folks.
1:51:33
I thought they uh explained a lot of how this works and Dan Shippers were like his uh you know third-party analysis of how he's using the tool sounded very promising.
1:51:45
We haven't had a chance to test it here on the show, but uh I'm excited for it. And Swix breaks it down.
1:51:50
He says uh people are poo poo pooing chat GPT agent as just a better harness, but they're not reading closely enough.
1:51:58
We've got a new frontier model today, folks.
1:52:00
These charts are like for like same harness.
1:52:04
Um they basically stopped short of calling it GPT5.
1:52:07
But yeah, if there were a public release of 04 full today was it? uh don't sleep on that.
1:52:13
In other words, run your benchmarks telling it not to use tools and I expect it'll be a big step up from 03.
1:52:21
And so he's looking at the benchmark of chatt agent on humanity's last exam um doing much better.
1:52:32
And so even though this didn't get a true version bump, the uh the actual results are are are great on practical problems.
1:52:40
I'm we're gonna have Mike from Arc AGI on the show later.
1:52:42
I can't wait until we can throw chat GPT agent at some Arc AGI puzzles.
1:52:48
I got a little preview and Arc AGI 3.
1:52:51
It is a challenge and I will be doing it live on the stream and I won't be embarrassed at all.
1:52:59
What >> John was uh sweating earlier? Yeah.
1:53:00
Gave it a little test run.
1:53:02
He said >> it is it will be a challenge and I imagine it will be a very huge challenge for uh for ChachiPT agents. It was funny.
1:53:11
So, uh, we had the, uh, the chat agents team on yesterday and then I saw on the timeline later people were kind of >> saying like, oh, it took 10 minutes to book a flight, which is like kind kind of like funny criticism because it's probably how much it realistically takes like a human to book a flight.
1:53:28
So, the fact that, you know, an agent could do it in the same time and you can imagine the agent could just get, you know, if it can get 10 times faster. >> Completely agree.
1:53:38
Also the interesting hottake is that there's there's a world where I have found that even if even if a task takes me 10 minutes on my laptop in a professional piece of software if I can do it in chatbt in the app for in 10 minutes and it's equally frustrating and it takes me the same amount of time I like being able to do it on my phone.
1:54:03
I noticed this when I made this 2 by two chart showing Dwarf Patel versus Tyler Cowan on their AGI perspectives.
1:54:11
So Dwar believes AGI is not here.
1:54:14
Tyler Cowan believes AGI is here.
1:54:17
Doresh believes the impact of AGI will be immense and Tyler Cowan believes that it will be very incremental.
1:54:21
So they're on the opposite sides of this 2 by two diagram. So I go to chat.
1:54:24
I I could easily have just done this in in Google Sheets and taken a screenshot.
1:54:28
I could have done it in Photoshop, but both of those would probably require opening my laptop.
1:54:33
But on my phone, in the chat app, I was able to have it try and do an AI image generation.
1:54:38
That wasn't really working. It was getting confused.
1:54:41
So, I finally had it use mattplot lib and write some code to generate it.
1:54:45
And I was and I went back and forth with it probably for about the same amount of time that I would have been in Photoshop.
1:54:50
But, it was nice because I was able just able to do it on my phone.
1:54:53
And so there's something about the the the rune take that text is the universal interface that even if it takes me the same amount of time to book a flight, if I can just do it in a in a universal interface that I'm super comfortable with and I'm interfacing with it on the Chachi PT app, I might prefer that over going to the United website or united Airlines.
1:55:14
com, whatever >> or Google Google flights where you can like see the flight and then they'll pipe you to the other site.
1:55:20
There's so many different sites and they're all slightly different.
1:55:23
Slightly different UI, slightly different uh designs.
1:55:25
And I, oh, this one I need to remember to uncheck this box and this one I need to do this and my password's not saved here. I like that.
1:55:32
Uh, Chat GPT is just becoming like a unified interface for how I interact with uh web services, tools, all this stuff. It's very cool.
1:55:42
Anyway, uh in other news, uh semi analysis, uh AJ, uh friend of the show says, "Ananthropic, quote, we don't care about consumer code is the only use case we care about."
1:55:54
Everyone else says, "Why is Anthropic not showing up in consumer statistics?"
1:55:59
And someone and he's quoting someone says, "What happened to Anthropic?"
1:56:02
Because Anthropic fell off of the open eye le uh the the rankings of chatbot downloads.
1:56:07
Catch is on a chair with uh yeah of sensor sensor tower.
1:56:13
Chatt is on a tear with almost a billion downloads.
1:56:15
Google Gemini at 200 million. Deepseek at 127 million.
1:56:20
Microsoft copilot at 80 million and Perplexity at 50 million.
1:56:22
Perplexity has been holding on strong in a very competitive environment.
1:56:26
Uh and I think that's why they were able to raise at an $18 billion valuation yesterday.
1:56:30
If you didn't see the news, um Swix has more analysis on the OpenAI launch. three things.
1:56:37
A deep and he's doing the Steve Jobs meme with uh Sam Alman.
1:56:39
Three things, a deep research model with enhanced search browser, a revolutionary computer use operator, and a sandbox terminal to execute math and code.
1:56:48
A browser, a terminal, a computer. Are you getting it?
1:56:53
These are not three separate agents.
1:56:55
This is one agent, and we're calling it agent.
1:56:58
This is a really good one.
1:57:02
>> Oh, great, great, great, >> great poster. >> Great poster.
1:57:04
Justine Moore says, "I predict that Grock's male AI companion will be even bigger hit than will be an even bigger hit than the female one."
1:57:11
Uh Elon has said that he's naming it uh Valentine, >> but what did it say it was going to call itself?
1:57:20
>> It has it calls itself something else that we won't say.
1:57:22
Um just women are quietly massive consumers of romance and erotica content. and she gives some data.
1:57:31
>> Fanfiction 80% romance novels are at 84 82 to 84% women online fanfiction 80%.
1:57:39
>> I've never been part of that whole world but you know >> yeah never been dig into that whole world.
1:57:44
>> Anyway, uh let me tell you about Finn.
1:57:44
ai the number one AI agent for customer service.
1:57:48
Number one in performance benchmarks.
1:57:49
Number one in competitive bake offs.
1:57:51
Number one ranking on G2. Go to fin.
1:57:54
>> I don't know how they do it.
1:57:55
>> I don't know how they keep doing it.
1:57:57
They they just >> luck of the Irish as they say. >> Luck of the Irish.
1:58:01
>> Baku Capital Bloke says, "Uh, I have two people in my life that I've mentally flagged for high risk of LLM induced mental health issues where I'm already seeing concerning activity.
1:58:11
I don't think society is ready for how much of an issue this is going to be at all.
1:58:14
So yeah, again I think there needs to be fast action from the labs in order to uh make um uh just make some changes to the product.
1:58:26
Make make >> uh it it does seem to be like dur it's not something that's going to happen in an hour from using the product, >> but it could happen in a month. >> Here's a prompt.
1:58:35
Chat GP agent, look at my calendar. Look at Google Maps.
1:58:38
Find me the nearest grass. I need to go touch it. >> Yeah. >> Schedule it. Put it on my calendar. Remind me to go there.
1:58:45
Call me an Uber to take me to the grass. >> Call me a mo.
1:58:48
>> Call me a whimo to take me to the grass. I need to touch it.
1:58:52
>> Take uh take me to Central Park.
1:58:54
>> Yeah, >> the home of our next guest, Joe Weisenthal.
1:58:56
Welcome to the stream, Joe. How are you doing? >> I'm doing great.
1:58:59
Thanks for uh thanks for having me back.
1:59:02
>> It's always somebody commented earlier on you you shared the the guest lineup and they were like, "Yeah, >> Joe goes on tpn more than I call my mom." And >> I love you guys.
1:59:13
But maybe that person that I think that person needs to get a build a better relationship with their mother for sure. But we are family. This is the brother. We are family. We are family.
1:59:23
>> You are a technology brother. >> Yes. Financial brother. >> Thank you.
1:59:27
>> Um what's what what's the latest in your world?
1:59:29
What has been um uh capturing your attention this week?
1:59:35
You know, here's an interesting thing is that um the economy just like the setting aside the markets which we all go nuts every day, the lines always go up.
1:59:44
There's the economy itself like the it it was a good week for economic data.
1:59:50
Um today we got better than expected economic sentiment.
1:59:52
Yesterday we got better than expected retail sales.
1:59:53
We got better than expected initial jobless claims.
1:59:58
We got better than expected um the Philadelphia Fed manufacturing survey.
2:00:02
survey. There has been this assumption I would say um in recent weeks that recent months really that the economy would slip and then the question would be would the Fed uh cut rates in time to forstall a recession and that maybe
2:00:14
that's still a debate certainly we know the White House really wants to see rate cuts but actually at least like as a snapshot right now actually it looks like there might be a little bit of a tailwind to this economy which I think is really interesting and maybe unexpected. >> That's fantastic. I have your post here. >> That's fantastic. I have your post here. Boom.
2:00:29
More strong economic data.
2:00:29
June retail sales rise gong for the American economy >> hit for Joe and the economy. >> Love it. >> I need a gong.
2:00:40
>> My producer Kale is in the room with me.
2:00:42
I think everyone needs a >> We We have We have variety of gongs.
2:00:46
>> We can actually send you a gong.
2:00:46
We have We have three now and we're thinking about getting a fourth.
2:00:50
So, we have spare gongs if you need to borrow one.
2:00:52
>> I think the guests need a gong. >> Yeah.
2:00:54
I think the big thing is uh it it's not necessarily that you need a gong.
2:00:58
It's that you need to get into prop comedy generally broadly and OddLots needs a whole prop department with a variety of things.
2:01:05
Have we showed you Have we showed you our props?
2:01:07
We have We have the crystal ball for tough for knowing the future. Of course. >> Careful with that. >> We now have >> Yep. Put it on John.
2:01:16
If you ever if John ever wants to steal man an argument. >> Steel man something. We have a steel man. Cover.
2:01:22
Do you have a uh do you guys have a tungsten cube?
2:01:26
>> We don't have a cube, but we have a tin foil hat.
2:01:28
>> We have a tin foil hat for when we're discussing uh conspiracy theories.
2:01:32
>> Yeah, it gives you a little coverage. >> I'm not saying this.
2:01:34
I'm wearing the tin foil hat.
2:01:36
Yeah, >> you guys for all the different scenarios, all the different >> How was uh you dropped your you dropped your uh interview with uh Eric Adams. It was this morning. Yes.
2:01:48
>> I haven't had a chance to listen yet.
2:01:50
You guys seem like you were having fun.
2:01:52
I don't think it's possible to not have fun talking to Eric Adams.
2:01:54
Eric Adams is one of these guys where it's like even the people who hate Eric Adams love Eric Adams to some extent.
2:02:02
He is uh you know you talk to him and there are some people you talk to and it's like you instantly get why they've been successful in politics. He has a great smile. He's very funny.
2:02:14
He sort of speaks extemporane extemporaneously very well.
2:02:19
You never really know what he's going to say.
2:02:21
And look, you know, he presents I would say not popular right now.
2:02:25
His approval rating is pretty low, but I would say he uh he makes his case well that he's had a a good mayorship between crime numbers, between the volume of uh housing that's been built in the last four years. I don't know.
2:02:39
Like the the rats the rat numbers are real.
2:02:42
I talked about it last time.
2:02:42
The rat numbers are real.
2:02:43
So again, I think there there was a poll out this week that actually showed him in fourth place, but >> is there really a rat census?
2:02:49
Do we have good data on rats?
2:02:53
>> So the way they measure it, yeah, they do.
2:02:55
Um I mean, >> this is what the collapse, this is what collapse looks like, by the way, is when you start measuring the rat population.
2:03:02
>> Are hedge funds trading against the rat index?
2:03:05
>> On the rats, >> the proxy that they use for u measuring rats uh rats is three 311 calls.
2:03:11
So 311 calls is like our way to like call the police about something that's not an urgent crime or other things going on.
2:03:21
And if there's like a rat infestation or a lot of rats uh place and they dropped like 40% in 2024 and then if you look at the annualized data through now it's like down another 25%.
2:03:30
I think on things like rats and trash containerization which I know you know trying to get into the modern era here in New York City.
2:03:40
Um, I think there was a widespread agreement that actually real progress has been made and Eric Adams has a lot of funny videos about his war on rats and I think uh I think he deserves credit for it.
2:03:53
>> Was there ever uh you you I don't know if this is an apocryphal story but the whole story of like the Indian cobras where there was a a bounty.
2:03:58
Have you heard the story?
2:04:01
>> Uh so it's a classic it's a classic economic example of uh of like unintended consequences essentially.
2:04:09
There's actually a particular term for it, but basically there was a snake problem, a cobra problem, poisonous snake problem in India.
2:04:14
And as the legend goes, as the story goes, um the the government said, "Hey, look, we're going to take a decentralized approach.
2:04:23
We are going to put this in the hands of the free market.
2:04:24
We are going to create a bounty for every dead cobra or dead snake that you bring us.
2:04:30
And so we will pay you $1 for every People started breeding them, right?"
2:04:35
>> They started breeding them. Exactly.
2:04:35
And so, and so I wonder if there's a world where, okay, I got elected.
2:04:40
I need to I need to crush the rat population, but then I need to climb it up once it's out of the news cycle.
2:04:46
I got to get the rod population huge and then I can crush it right before >> you can keep smashing it again.
2:04:53
>> Keep smashing it right on right on the cycle.
2:04:55
So, oh yeah, it was bad midterm, but right as you don't want to change horses in the middle of a stream because I'm the rat catcher. >> The rat catcher. >> That's right.
2:05:05
But uh I mean yeah is there is there a secret in New York City to controlling the rat problem?
2:05:09
Is it just like more rat?
2:05:12
>> I think it's just I think it really is just a function of how horribly we've managed our trash and you just see and I saw it you know I was walking uh to the subway today because the containerization of trash like it's it hasn't come to my part of the city.
2:05:24
It hasn't come to the east village.
2:05:26
There's just a lot of trash bags that exist on the street.
2:05:31
I and and I saw a bunch of incidentally I saw a bunch of rats last night when I was walking home. So it is a live problem.
2:05:37
There is still progress and you know like I would never endorse a candidate for mayor but uh I would certainly suggest that whoever is the mayor after uh November or after the inauguration.
2:05:49
Hopefully the existing trajectory continues because the problem is not solved. >> Yeah. Deals with it. Jordy.
2:05:57
>> Uh >> who else comes on your show and talks about rats?
2:05:59
probably not, by the way, it's probably just me.
2:06:02
>> Well, you're you're you're a new rat correspondent.
2:06:04
>> Emily Sunberg is a helpful is a helpful New York reporter to understand what's going on the East Coast.
2:06:08
I mean, the the big thing that I've been pulling on, I talked to a couple hedge fund folks out on the East Coast in Manhattan.
2:06:13
We had one guest on from CO2 who was had a very beautiful scenic view out the window uh from hedge fund alley.
2:06:21
I guess that's a street or something like that.
2:06:23
Um and and my rea I was I was pulling on what has the reaction been in the finance world in the on the east coast to these crazy aqua hires that are going on on the west coast.
2:06:37
So uh we saw that Google acquired Windsurf for $2. 4 billion.
2:06:40
It was this kind of zombie aqua hire that's becoming more and more standard.
2:06:45
Mark Zuckerberg's paying hundred million $200 million for single AI researchers.
2:06:49
It's kind of shaking Silicon Valley.
2:06:52
for the news cuz now one what would have been one acquisition is now two. >> Oh yeah.
2:06:59
>> So we got to cover it Friday and then we also got to have Scott Woo who bought the the the ghost ship >> on Monday.
2:07:07
>> I guess yeah the question is just like what's the reaction uh because for Wall Street with the pods it feels like it's standard.
2:07:12
Yeah, it's that's really you know it's funny literally just before we got on here and I was just uh talking here it just feels like you know if you think about the sort of broad phenons in the economy that what we're seeing is more and more sectors of the economy are experiencing this sort of like winner take allness phenomenon where it's not that the sector is doing good or bad.
2:07:32
is that there are a handful of talented people in any industry that just, you know, capture extraordinary sums.
2:07:39
And, you know, we've seen it for years in professional sports.
2:07:42
The pod shops, um, the hedge funds that we talk about a lot about on the podcast have had it for years.
2:07:48
That's our equivalent at Bloomberg, you know, like the stories that, uh, read Spike on the terminal.
2:07:54
It's always about like some guy it's like oh so and so who you know is a transportation portfolio manager at this pod shop is like going there and they're getting a $50 million bonus etc.
2:08:02
getting a $50 million bonus etc. It feels like um you know we see it in media of course uh in various ways and star newsletter writers and star podcasters and star broadcasters and so forth and then clearly like obviously
2:08:17
you know software people have um been getting paid well for a long time but this phenomenon where now it's like no the value is not in the company per se the value is in just that inner uh individual talent who can get up and walk and take that value out with them. >> Yeah. It feels like it's uh replicating >> Yeah.
2:08:31
It feels like it's uh replicating elsewhere and like I don't know where it's going but this seems to be a phenomenon of the world. >> Yeah.
2:08:38
that that that's why anybody freaking out about like the the comp packages of individual re researchers, it's like yes, if you look at it on an individualized level, one person getting nine figures, but if you look at it and say, well, Zuck just spent >> 15 on scale AI >> and he which which was heavily talentoriented.
2:09:01
uh and would he pay $4 billion for one of the top teams if he could just aqua hire one of the top teams and in this case he just pieced it together from a variety of labs.
2:09:12
>> It's really interesting to one thing I think is interesting to think about is that a lot of in the finance world a lot of talentdriven businesses often aren't particularly great for shareholders.
2:09:22
So if you look at like the history of like investment banks that are standalone investment banks that never had like a trading arm etc.
2:09:30
Almost all of the enterprise value like ends up acrewing to the talent and there's not much frequently like the equity component when these companies are publicly traded.
2:09:40
You look at like boutique investment banks that from time to time are publicly traded.
2:09:44
They've never like been particularly those have never done particularly well.
2:09:49
particularly well. or you think about a law firm of course which doesn't have public equity but like you know it's like all the money sort of like acrru to the talent and so I do think there are some like interesting implications like maybe down the line not yet in public
2:10:02
markets but it's pretty obvious um from windurf that in private markets this has got to be a you know change the way investors are thinking that you know what really is left over for the poor downtrodden shareholder if the uh if the uh most talented workers at the company are the ones who really get to capture all of the value. >> Yeah, there's
2:10:23
>> Yeah, there's >> Yeah, you you've seen this with uh the the podcast networks of the world, the true networks.
2:10:29
If you know Dave Portoi builds up some talent, he starts paying them 100k.
2:10:35
>> So, why share why share it at that point with Dave at that point, right?
2:10:37
Like if you are superstar podcaster under um that particular network eventually you get to a point >> where it's like you know you take maybe you start off and you take 20% of the cut and then 50 and then you're like why am I sharing it all at all? >> Yeah.
2:10:51
Alex Cooper basically Alex Cooper's market value is about the same as an AI researcher like over the last over the last few years. >> Yeah.
2:11:00
My so my I I've been thinking about this a lot since we last talked and you said the power laws are popping up everywhere and and I was and I was just going through my daily life thinking like how is that true? I see it everywhere.
2:11:11
everywhere. I agree in a lot of things but I was thinking about like you know there's this meme of like after AI the last job will be like the plumber and I was like is there a power law compensation curve in plumbing like that's interesting
2:11:25
>> or driving because if you think about the value of of of driving it's really a lot of there is a ton of equity value in Uber there is a ton of there's value it's a leverage thing that the highest paid plumber will be somebody who has
2:11:40
like a a like scaled plumbing enterprise that you you you can only like install so much pipe >> in but if they lose their talent like there's still equity value because you've aggregated demand and that's the difference between the AI research labs
2:11:58
and Instagram every they could have 100% turnover at Instagram slot other people in it's a network effect there's an audience there there's habitual I know I have a fan base on Instagram or I open it and I know that I get funny memes or videos or family connections there. I'm
2:12:13
I'm not leaving just because the software engineer who works on this particular button leaves.
2:12:19
And so there's a ton of equity value in Instagram relative to the talent.
2:12:23
And and and I think that there I think that it's not as it's not as universal as we think.
2:12:28
I think we're just seeing it more in media and finance and research and places where there are either secrets that you can take with you or relationships that you can take with you or or fandom that you can take with you.
2:12:41
And so it's not it's not entirely 100% universal, but it certainly is something that it's it's more on display. It's than ever before. >> I don't know.
2:12:52
>> Yeah, I think that's good.
2:12:52
Look, I I I think this is a a useful uh corrective the phenomenon.
2:12:57
There are many parts of the economy where the sort of distribution between equity and talent um is not as skewed as it seems to be in uh some of these areas. It does seem on display.
2:13:12
I mean it's weird that we know the names of software engineers in general, right? >> Yeah. Like trading card memes.
2:13:16
We put up these trading card millions of views. It's crazy. >> You guys nailed it. Like Yeah.
2:13:20
>> You guys nailed it. Like Yeah. And I, you know, when I think about, well, you guys, not to, you know, blood much smoke, but when I think about like, okay, like when I think about TBPN and like what is it about this moment and why has it worked and why has it worked as well as it has, I do think, you know,
2:13:37
it's obvious that just like so much of this thing is like it's a, you know, you guys do like a quasi sports show and so much of uh what we're talking about is de facto sounding like sports, including the degree to which people just get uh traded, so to speak, from or big uh signing transfers from uh one firm to the next. >> Yeah. I mean to me to me business like >> Yeah.
2:13:56
I mean to me to me business like anybody that is like sufficiently nerded out about business and markets and tech is like it's always been you could always comp it to sports, right?
2:14:06
You have personalities, you have effectively coaches, you know, like the elder VCs, you have the the teams, the companies, etc.
2:14:15
Um, switching gears a little bit.
2:14:18
Uh, what's up with what's going on with general solicitation right now?
2:14:20
Seems like it seems like there's a bull market in general solicitation.
2:14:24
There's there's this guy Eric Jackson who's just been basically going out saying, >> "Yeah, I'm just looking for 100 X's.
2:14:31
And by the way, if you want to if you want to if you want to join the ride, you can."
2:14:37
And a lot of people >> I'm going to try to be uh careful when I talk about this topic but the idea of like you know we are going to pick a ticker >> try again trying to be careful we are going to pick a ticker and just sort of manufacture momentum for it has probably some always existed in markets >> and we seem to be in an age in which uh this sort of behavior is people don't try to hide it as much and that's I think not the surprise.
2:15:06
Have there always been people attempting to like, all right, we're going to like try to uh I'm trying to think of a word that wouldn't land me in legal trouble.
2:15:13
We're going to try to move this stock for sort of noneconomic reasons because we gather and buy it.
2:15:22
>> Um, has that always existed?
2:15:22
I'm sure we seem to live in an age where not only are people comfortable with just sort of doing that in the public, you have a lot of people who uh sort of say, "Yeah, you know what?
2:15:35
Everybody's got to eat and everything is corrupt these days and everything is a scam.
2:15:40
I don't really believe that for what it's worth.
2:15:42
I actually think American capital markets are the best in the world from a sort of transparency and regulatory side.
2:15:48
I actually think it would be sad if uh we lose that.
2:15:52
Um but I do think there's this wide perception it's all a scam. Everybody is corrupt.
2:15:56
Everybody has inside information.
2:15:58
you're not, you don't have that information and therefore why can't you participate in your own form of uh you know driving the market to your will.
2:16:07
Social media obviously allows that and no one seems to everyone seems to be sort of cool with that these days and like I said I actually do think American capital markets are deep and fantastic and well regulated generally and if you look anywhere else in the world >> let's give it up for capital markets we'd love them.
2:16:27
Um, I was talking with Jordy about this this morning.
2:16:30
>> Well, it's funny that when when somebody is >> doing uh or you know, basically doing general solicitation, the the reaction is for a bunch of guys who love finance to go and dunk and quote tweet the person and be like, "Look," and it just drives more attention. >> That's the thing. How do you cover it?
2:16:46
Like I I I've thought this is the problem.
2:16:48
Like how do you even like begin to cover it?
2:16:50
Because I've wondered about this like with doing the podcast over the years where it's like certain like certain you know I'll hear a story someone will tell me about like some meme coin or something and the schemes that they like concoct to like pump it basically straight up manipulate it or like gather in a group etc.
2:17:08
And it's really interesting but it's like it feels a little you know like I would love to talk to you on the podcast.
2:17:13
On the other hand, I no matter how much that person is straight up admitting to market manipulation, the mere fact that they would be getting attention would almost certainly drive the whole thing higher.
2:17:25
So, it creates this very like weird tension from this. >> Yeah.
2:17:28
Remember when the when there was that Argentinian presidential meme coin that like Malay kind of endorsed and then the guy that created it like went on like a started going on podcast immediately breaking it down. >> Yeah.
2:17:40
You know, I there's that famous uh George Soros quote and I don't know it exactly but he talked about it.
2:17:43
He's like, "When I see a bubble, I run towards it."
2:17:46
And I think everybody is now adopting that where they look at something and they want to get in.
2:17:50
They see a scam and rather than, "Oh, I want to avoid that scam."
2:17:54
It's I want to get in on the scam.
2:17:57
I want, in other words, it's like, I want to be in on the next M.
2:18:01
>> I remember this during >> I remember this during the crazy crypto era of like 2021.
2:18:05
Uh they uh I remember Coffeezilla, this YouTuber who would talk about these scams, would interview some of the uh like victims who lost a lot of money and they were like, "Well, I knew it was a Ponzi scheme, but I thought I was early. I thought I was early."
2:18:19
>> Yeah, you want it's I think I wrote I got to go find that cuz I hadn't thought about it.
2:18:22
I think I literally wrote once like the new thing is like getting in on the next maid off getting because if you are early and if there's sort of like we're in an era where regulators whatever just don't care about this stuff and or the expectation is like you play everyone gets to play the game everyone knows what the game is and you just don't want to be the last bag holder the game is to get in early.
2:18:39
I was thinking about this in the context of the Coldplay concert that went viral.
2:18:44
The CEO of Astronomer took over the internet and I was like, "Okay, it's a private company, but if it was a public company, what would have happened to that stock?"
2:18:52
Is there a world where someone's like, "Let's turn this into a meme stock because it's a small company.
2:18:57
>> Let's buy the Coldplay IP." Yeah.
2:18:59
>> Roll it in and turn it into a Coldplay IP treasury play.
2:19:03
>> And uh I'm sure there'll be something out there in the near future. I don't love it.
2:19:07
But, you know, I'm old and like I'm a boomer at this point.
2:19:09
So, maybe I just need to sort of embrace the new generation.
2:19:13
>> You know, you know crazy some somebody messaged me and they were like, "Do you know Andy Byron?
2:19:17
The guy at the Coldplay concert was previously the president at Lace Work, which went from zero to 7 billion in the P or 8 billion in the private markets, back down, ended up selling for uh like I think around 150 million."
2:19:34
>> Um, so he's been on he's been on ride.
2:19:37
So, that's someone who that's someone who knows the roller coaster of life, huh? >> Yeah.
2:19:40
So, he'll he'll be back, I'm sure, probably uh uh in in another marriage.
2:19:45
Um what's going on with the 230 spread?
2:19:48
I saw you posting that uh the 230 spread is at its highest since October 2021.
2:19:51
I remember October 2021 there were top signals blaring everywhere.
2:19:56
Uh we have our own top signal tracker that we're working on internally uh that we should we should roll through as well.
2:20:03
But but >> you guys going to get into proprietary data? No.
2:20:06
So like you know the way to think about you know the president has been out calling for rate cuts and we might get rate cuts at some point in the next year but the way to think about the yield curve is that um at any given point on the curve that number reflects like the sort of average rate over that time.
2:20:23
So the 2-year yield, for example, is essentially what the market expects the Fed will will have rates on average over the next two years.
2:20:30
The 30-year is what the market expects Fed rates to be on average over the next 30 years.
2:20:34
And what you see is this uh you know, even at this time of calling for rate cuts, that number at the long end keeps going higher and higher, which suggests that, you know, the Fed could cut rates right now.
2:20:46
could get rights aggressively if Trump replaces Powell sometime next year or perhaps tries to fire him before that with someone who's sort of a lackey or a loyalist.
2:20:54
We might get this situation of sort of like aggressive rate cuts now.
2:20:58
But if people perceive that to be inflationary and so forth, one possibility is that rates go up at the long end because they'll eventually have to compensate that u compensate for that by higher rates to fight that inflation. And so you see that now.
2:21:10
And what that means though is that even if you got rate cuts right now, you might not actually get lower mortgage rates or lower car rates because those are the rates that actually matter towards where people are borrowing.
2:21:20
So I think perhaps maybe the market is telling Trump a signal you're not even going to get what you want even if you get what you want.
2:21:28
You could get that Fed chair who immediately cuts rates for you, but if the goal is lower rates to stimulate the economy, it may not even work because you might see longer rates rise in compensation. >> Interesting. Um, yeah.
2:21:38
Can you talk about the decision uh for who runs the Fed?
2:21:44
Uh the it's been kind of wall- to-all co uh coverage from the Wall Street Journal.
2:21:48
All different bank CEOs coming out.
2:21:50
There's been memes about like you know the school walk out don't fire Jerome Pal like what what are the arguments on on the sides of like hey let's not intervene from the executive branch.
2:22:03
>> Yeah there's a bunch of interesting dimensions.
2:22:04
I mean look I'll say a few things.
2:22:06
There's always been pressure um put on the Fed from time to time from the White House.
2:22:09
the White House. It's not new to Trump, but it is really stepped up and it's the the sheer number of attacks, the demands for immediate rate cuts, the sort of uh the attacks on Powell for the cost of renovation um of the of the uh Federal
2:22:24
Reserve building, which interestingly there was a story by AP out today that said one of the drivers of the cost is that in 2000 um the uh administration officials wanted there to be more marble instead of glass in the building their aesthetic preference. It's very Trump-like and so
2:22:41
It's very Trump-like and so that could be so there's some why the renovations have been so expensive. That's one possibility.
2:22:47
But then the you know there's a range of so there's this guy Kevin Wars who for my entire career has been a hawk always calling for higher rates.
2:22:56
Suddenly he's calling for lower rates. >> Interesting.
2:22:59
The battle of Kevin there's two Kevins that are in in >> Yeah. Right. Right.
2:23:02
Then there's Kevin Hassid who I think is generally um you know considered to be like a respectable economist um by and large and you know if he if he were to get the nod I don't think there would be too much anxiety but the assumption is that whoever uh replaces Powell would be much more inclined to cut rates sooner and faster.
2:23:23
And if it if it's and if it's true that like the economy has some upward momentum right now, then you could see how uh people might perceive that to be uh inflationary.
2:23:31
And another name I wrote about him today, uh Christopher Waller, he's a governor on the board.
2:23:35
He is probably um he's calling for rate cuts.
2:23:39
He thinks there should be a rate cut as soon as the next meeting.
2:23:41
But in Waller's defense, um he's had really good intuitions this whole cycle. So he's very hawkish.
2:23:48
Shivan to fight inflation aggressively even while others on the board were calling it transitory.
2:23:54
He predicted that inflation could come down from its highs without a meaningful increase in the unemployment rate which has been proven correct.
2:24:00
So, you know, there are some who are suspicious and cynical and say he's openly campaigning for that Fed chair job.
2:24:05
But he's been no one could really deny that he's uh he's he's had his finger on the pulse about as well as anyone else for the last 5 years.
2:24:16
And this is important which is that the FOMC, you know, it's 12 people.
2:24:21
It's not just the rates aren't just decision of the Fed chair.
2:24:22
Waller is someone who would probably have a pretty decent amount of credibility with the other 11 people on that board because he comes from there.
2:24:30
So, it's possible that, you know, maybe he's the most uh logical choice.
2:24:35
But what's likely, I have no idea.
2:24:37
Uh, off the top of my head, if I'm going to steal man rate cuts, I'm going to say something I'm going to say something like, >> uh, yeah, the economy is doing well, the retail sales are good, and jobless claims are fine, but uh, it's still really expensive to buy houses and, uh, mortgages are really high, and so I want to help Americans buy more houses.
2:24:59
Getting people on the housing ladder is the way that they start accumulating capital. That's good.
2:25:03
If I do the the tinfoil hat, I'm struggling with this because I I imagine that, you know, if I'm, you know, want to if I want my party to win in the midterms, if I want my party to win a re-election, I want the economy to be really strong, I want markets to be booming.
2:25:18
So despite all of the different, you know, uh, social issues that might be floating around, everyone says, well, like my IRA is up, my 401k is up, my market portfolio is up, but it feels like it's too soon to be putting to to be playing that card.
2:25:33
To go back to our rat analogy, if I was completely cynical just saying this is all about politics, I wouldn't be demanding rate cuts now.
2:25:40
I'd be demanding them right before the midterms, which I believe next year. Right.
2:25:44
So yeah, re react to those steel man, the tin conspiracy.
2:25:46
So look, you know, like I said, you know, the the issue with rate cuts from a housing perspective is that mortgages, mortgage rates may not go down if there are rate cuts because they're tied to the long end of the curve, which might go up in expectations of future inflation.
2:25:59
The strongest argument >> what could we could we pos what could we do like couldn't you just go in with like Figma and edit the long end of the curve basically bring it down?
2:26:11
Yeah, you could uh well, you know, the equivalent of the equivalent of Figma, I guess, would be yield curve control, which that's not unprecedented, where the Fed actually just goes out and says, "We are not going to let the 10-year rise above X level, and we will buy treasuries with our unlimited balance sheet >> and um and they could do that.
2:26:28
The issue then is you probably that means letting inflation run hot." It's tricky.
2:26:33
Um but the steel man for rate cuts is that the labor market has slowed and uh the rate of hiring really has declined um quite a bit.
2:26:43
I mean I think there have been several stories about how low hiring is particularly for college grads and so forth.
2:26:48
So this I think the the the argument for rate cuts there's a straightforward one which is that you could make the argument that the economy needs support or that it's over the rates are overly restrictive which is the argument that Fed Governor Waller made yesterday.
2:27:02
But you know just to your point um about you know the elections and IA look like what was it uh I think yesterday there was the news that President Trump was going to allow people to you know invest in crypto in their 401ks or something like that.
2:27:17
There was some headline about that and access to private credit.
2:27:20
You know I do think it looks like this is an administration that is very comfortable with letting asset prices rip and it people like that.
2:27:28
People like when their 401ks and their IAS and all that go up and their cryptocurrencies go up and I think this is an administration that is like pretty happy to see that.
2:27:38
>> We said this early, give Jane Street direct right at access to the Fed funds, right?
2:27:43
Let the high >> solves it all. It solves it all.
2:27:47
>> Just solve for market prices.
2:27:47
Just make the market go up as much as possible.
2:27:52
>> How's how's Jane Street doing with their little little PR serious PR crisis?
2:27:54
I actually haven't followed.
2:27:58
No, >> the Indian the Indian options >> and then on the Sudan stuff too.
2:28:03
That whole that whole thing. >> Wait, what?
2:28:06
Oh yeah, we talked about that.
2:28:08
>> No, it was just like you go from not hear you only hear about Jane Street really on the Darkh podcast.
2:28:14
>> Yeah, >> you know, a nice ad read and then you start hearing about, you know, all these other things.
2:28:19
It seems like >> I don't know.
2:28:21
I got to I I I I'll check in on Jane Street for you guys. >> Thank you. Thank you.
2:28:24
What jersey are you wearing, by the way?
2:28:26
Yeah, bring >> Oh, I got this.
2:28:28
I was in Mexico earlier this year.
2:28:30
By the way, I got your hat.
2:28:30
I have I and I have it in the office.
2:28:32
I wasn't sure if that would be a little I was thinking about wearing your TBPN hat.
2:28:37
But is that I was like, is that like wearing the t-shirt to the band when you're going to see the band?
2:28:40
I wasn't totally sure, but I'll put it on next time.
2:28:44
>> I just want to I want to see a suit at some point.
2:28:46
Maybe >> it would be cool.
2:28:48
We got to get some hot tuxedo so we can wear hot. >> You know what?
2:28:53
I will wear a suit or at least a shirt, jacket, and tie next time you guys have. >> It's so great.
2:28:57
It's we we've been we try to wear white suits when the market's ripping.
2:29:01
The S&P 500 was at an all-time high yesterday, right?
2:29:03
We didn't even think to We didn't even think to put it on because it's just we're so normalized to every day here.
2:29:11
>> Anyway, thanks so much for stopping by. >> Thanks. Anytime. Always love it. Have a good weekend. >> Love it, dude. Have a great weekend. >> Bye. >> Cheers.
2:29:17
>> Really quickly, let me tell you about public.
2:29:18
com investing for those who take it seriously.
2:29:20
They got multi-asset investing, industryleading yields, and they're trusted by millions.
2:29:24
Um, we will bring our next guest, Nurav from Next Door.
2:29:29
Have a bunch of questions, particularly about, uh, political campaigns given some of the folks we had on earlier, but uh, thank you so much for joining. Good to meet you. How you doing? >> I'm doing great. Thanks for having me. Excited.
2:29:42
>> Welcome to the stream.
2:29:43
>> It's great to have you.
2:29:43
Um why don't you kick it off with an introduction on yourself and the company and then I just want to jump into a bunch of questions immediately but uh I'll let you I'll let you in do the introduction.
2:29:52
>> All right, appreciate it.
2:29:52
Nervia, co-founder and CEO of Next Door.
2:29:54
Next Door is a public company that is trying to build the daily utility that you use every day to find out what's going on around you.
2:30:02
We're focused on the neighborhood in particular and so we sometimes call ourselves the essential neighborhood network. Mhm.
2:30:09
>> We started in the summer of 2010, went public in 2021.
2:30:12
I ran the company for the first nine years and then actually came back to the company a year and a half ago as CEO.
2:30:18
So I'm a refounder, not just a founder, but a refounder.
2:30:22
And this week really exciting for us because on Wednesday we launched the new Next Door.
2:30:28
The reason I came back to the company is because we needed a much better product.
2:30:32
And sometimes founders are in a good position to reimagine the product even though they built it in the first place.
2:30:38
And so for the last year and a half, we've been working on a brand new version of Next Door.
2:30:41
Launched it on Wednesday. It's going really well.
2:30:45
We're excited about the future.
2:30:47
>> What's What's the biggest change with the new product?
2:30:50
>> It's completely different.
2:30:50
And so it's not an evolutionary change or like a version 1. 0 to 2. 0, right?
2:30:54
But the biggest change is we've taken an all-purpose news feed because we're a social network for neighborhoods.
2:31:01
like there's a social network for pictures and a social network for people and for businesses, right?
2:31:06
And we've taken that social network and we've started to make it a lot more structured and utilitycentric.
2:31:11
So, we focused this release around local news to keep neighbors informed, local alerts to keep neighbors safe, and local recommendations to keep neighbors smart.
2:31:21
In the news feed, there was always local news being discussed, there were alerts being discussed, there were recommendations being asked for and given.
2:31:28
But now we have dedicated parts of the app where you can find those things and so it's just a lot more useful.
2:31:35
>> Senator, do you sell ads?
2:31:37
>> What's the business model? >> We do.
2:31:38
We we have over a 100red million verified neighbors in 11 countries.
2:31:41
So decent scale and as I said, we're public and so hundreds of millions of dollars in revenue as well.
2:31:48
And the business model is advertising.
2:31:51
>> Yeah, that makes sense.
2:31:51
Um we had a uh friend of the show on earlier this week.
2:31:57
He is running for state council. Is that right? >> State Assembly. >> State Assembly.
2:32:01
And it was interesting because he come he came on our show.
2:32:04
And we we have an audience all over tech and in San Francisco and New York.
2:32:09
And I was and I was trying to puzzle like I want to help him.
2:32:14
I I'm excited about his campaign.
2:32:16
Um but realistically we're not we're not a a platform focused on his district in Los Angeles.
2:32:22
So he was saying 20 square miles. >> Yeah. Yeah.
2:32:26
He was saying he was going to go knock on doors literally um and and go to local events, but I was I was thinking about what is the most internet native way. He's a great poster.
2:32:34
He creates he created kind of a viral video.
2:32:38
Um and he seems to be good at communicating through the internet.
2:32:41
What can you tell me about politics on Next Door generally, local politics?
2:32:45
Um have people found luck there?
2:32:48
What are the different strategies?
2:32:50
Are you in favor of this?
2:32:52
Are you excited about this? Are you avoiding it? What how does politics?
2:32:56
It's a great question and honestly it's something that we've struggled with since the inception of the company because when it comes to national politics which we know very divisive totally totally >> when we see that on the platform >> we don't like it.
2:33:09
We've said specifically no discussion of national politics and no national political advertising.
2:33:15
So that's where we started. Right.
2:33:18
What we've realized over time is when it comes to local issues, whether you call them local issues, civic issues, or local politics, those topics matter to neighbors.
2:33:27
And the point you made is exactly right, which is the folks who are behind those topics, whether they're elected officials or whether they're folks in the local community, they're trying to get something done, they need distribution because the internet has made it easy for you and me and your co-hosts to talk to each other.
2:33:45
And I don't even know where you all are. I'm in Texas, right?
2:33:48
But I'm pretty sure you're not in Texas, right?
2:33:51
>> But it has made it very hard because we're looking on our screens and looking at our computers to look out the window and see who's right across the street or who's next door, right?
2:33:59
And so that should be the promise for Next Door.
2:34:00
And so it's a long-winded answer to your question, but I hope in the future that we will provide your guest an opportunity to come on to Next Door where he knows he can reach verified neighbors in that 20 square mile area and he can talk about the issues and how he wants to deal with the issues.
2:34:22
Historically, for 15 years of our history until Wednesday when we launched the New Next Door, we didn't really let anyone except for verified neighbors create content.
2:34:32
We had some public agencies, police departments, fire departments, some mayor's offices, but not what you would call local politicians.
2:34:39
And what we realized is ultimately our job is to give you all the local information that exists regardless of the source.
2:34:46
And so with the new next door and with local news, which I mentioned, we have 3,500 publishers that are publishing 50,000 articles every single week.
2:34:55
Now in my city of Dallas, that's the Dallas Morning News. That's D magazine.
2:34:59
That's all the local publishers.
2:35:01
It's not the New York Times publishing into Dallas, right?
2:35:06
It's the Dallas centric and the neighborhood centric sources.
2:35:08
So the same way we're letting those third party in, those third parties in, I fully expect at some point in the future, and I don't know when, >> we will let the third parties that want to be elected officials or who are elected officials come on to Next Door.
2:35:25
And we just need to figure out the right way for the signal to be higher than the noise. >> Yeah. No, I completely Yeah. It's it's fascinating.
2:35:31
Like I had this weird interesting experience where we moved into a neighborhood and there weren't curb cuts and we have a stroller and my wife was like I'm going to write a letter to the city and I was like that's never going to work like that it'll be 10 years and they fixed it and they made a curb cut in like three months or something. It was really fast.
2:35:50
I was very impressed and it's clear that like the city was just prepared and and open to that and I didn't even know that because I don't have like this connection because the local news has kind of dropped off so much.
2:36:02
Um, how do you think about the business model that a local news creator might have in the future?
2:36:10
We talked to Chris Best at Substack yesterday.
2:36:14
>> Um, very interesting business model there. I could imagine.
2:36:16
I mean, every social network has had um you know, people might talk about the YouTubers or podcasters and Joe Rogan with these big deals, but there are people that make full living just on Instagram or just on X or just on threads.
2:36:29
Listen, this show is really just really big on X.
2:36:31
Um and um I'm interested in how you see the business model of a local content creator.
2:36:40
Is there a world where you're both partnering with an organization to distribute their content that's maybe in a local newspaper physically their own website and then you're kind of a top of funnel or do you think there will be nextdoor native creators soon if there aren't already?
2:36:56
>> It's a big challenge for local publishers because as we know the old business models of delivering papers or driving subscriptions for things that show up in our mailboxes that doesn't really exist.
2:37:06
At the same time, the digital business models, they don't have the same scale.
2:37:11
And so, it's hard for them to justify the ad sales forces or the technology investment, etc. Right?
2:37:18
Today, what we're doing is we're starting by sending them traffic.
2:37:19
So, we don't have a walled garden on Next Door.
2:37:22
These 3,500 publishers, they give us a headline, they give us a hero image, they give us a snippet, and then if one of our members wants to read an article, they go off to that publishers's website.
2:37:32
And that publisher can monetize through a subscription.
2:37:34
They can monetize through advertising.
2:37:35
They can monetize whatever way they've chosen.
2:37:39
The thing that we've thought about and I'll get to kind of the next door creator idea that you talked about, but thing we've thought about is if there's enough demand in one area to get local news broadly, >> could we put together a kind of Apple News or Spotify kind of subscription >> where you pay one entity next door and then they get the proceeds on some kind of revshare based on what articles are read.
2:38:05
So, we've thought about that, but it hasn't gone any further than us thinking about it because we need to see how this performs.
2:38:12
>> Ultimately, I'm not sure about the business model behind it, but we certainly believe in citizen journalism.
2:38:18
And so, even enabling the high school journalism student who wants to write about the high school sports that are going on and making sure that they can use Next Door as a distribution platform.
2:38:28
That's something we want to enable because if that person goes on Instagram or X or LinkedIn or Facebook, there's just not enough audience.
2:38:36
But if that person comes to Next Door and they're writing about the neighborhoods that are around a particular school, those neighbors want to support that school.
2:38:45
So there's a lot of opportunity for next door.
2:38:48
I would say the story of Next Door is one of massive potential and of not having a good enough product to really deliver on all that potential.
2:38:58
And that's what we're trying to change.
2:38:59
>> Yeah, it's an interesting challenge because like social like social networks in general have have historically done well by reducing friction on the sign up.
2:39:08
But the nature of Next Door is you need that extra step of like verifying does this person actually live where they, you know, in the in the neighborhood that they're trying to participate in.
2:39:18
And I and I had this one moment uh it sounds like it was before you came back and and started fixing things where I moved to a new neighborhood.
2:39:26
I was trying to get set up and it wasn't verifying and I just churned and I haven't I haven't gone back and so I'm I imagine this >> um even though I'm sure I'm sure it's active I I know it's a very active uh community on there.
2:39:38
One one question uh >> by way that is kind of a hard problem because when people try to switch around neighborhoods they don't have the patience to reverify.
2:39:49
They do have to verify initially they don't have the patience to reverify.
2:39:52
So, it's something that we need to make a lot easier because the thing you're talking about, the statistic I heard was something crazy like 10% of Americans move every year.
2:40:02
And so, it's happening a lot.
2:40:04
People are moving neighborhoods, whether they're in the same city but in a different house or whether they're moving cities altogether.
2:40:08
And so, that's something that we need to make frictionless. >> Yeah.
2:40:12
I wonder if you could like like upload a video of you walking into your house with the address and then you do some like geogesser type AI thing to identify that hey this person's really walking into this house like >> we've done everything from innovative technology all the way to sending you a printed physical postcard.
2:40:29
Yeah, >> we've done all of those things.
2:40:32
>> The printed postcard is Lindy.
2:40:32
It's it's not going anywhere.
2:40:35
Um, how how do you guys think uh it's top of minds?
2:40:37
Uh, because John and I live in Southern California, I I live in Malibu, John lives in Pasadena, so we were both uh pretty close to the fires that happened.
2:40:49
Fortunately, we weren't directly uh our homes or neighborhoods weren't directly affected.
2:40:54
And then you're in Dallas near uh not too far from uh where the flooding happened.
2:40:59
How do you how do you guys think about building features around uh helping people uh through disasters?
2:41:06
Because I remember the the fire app in Southern California that everyone was using is just run by a nonprofit.
2:41:13
And there was one night that John basically thought >> Watch Duty. >> Yeah. Watch Duty.
2:41:17
John John one night like basically thought he was like, "Oh, Jord's probably, you know, I just we didn't have any cell service or anything like that, but I just remember using wa watchduty." >> Yeah.
2:41:27
>> Yeah. it would just periodically go down and it felt like okay there's probably going down because they just don't have the the scale >> the the engineering capacity the scale etc to just like run this service I'm curious but it's a hard problem to try to tackle because it's so spiky right it's like you >> yeah look watchd duty is an amazing app and so kudos to them and when the palisades fires happened I looked at
2:41:51
that thinking we have to do more now we had we had seen all of our metrics spike like crazy in those Palisades neighborhoods because when a crisis happens, whether it's a power outage, which is relatively benal but annoying,
2:42:05
all the way to inclement weather or a natural disaster like a fire, right, or terrible flooding, next door becomes a lifeline because there is no other infrastructure to communicate with the people around you. And you need to be
2:42:18
And you need to be able to either say in a very simple way what's going on all the way to in a critical way say I need help. I need to be rescued. right?
2:42:26
Or I can provide someone with help.
2:42:28
And so historically through hurricanes, through tornadoes, through fires, Next Door has always spiked when these things happen.
2:42:34
But it's just a newsfeed UI.
2:42:36
And so with the new Next Door, we've created an entire alerts surface where the entire app transforms into this lifeline when, god forbid, a crisis is happening.
2:42:46
And we take now authoritative data sources.
2:42:49
So in much the same way that we're bringing in third parties on local news, we bring in third parties.
2:42:55
I think we'd like to bring in watchd duty at some point as well.
2:42:58
And because we know where people live, we can deliver their information proactively to just those areas.
2:43:04
And that's what's very different because if you think about watchd duty, watchduty doesn't know where you live.
2:43:09
And watchd duty can't warn you before you go open watchd duty to see if something's going on, right?
2:43:16
But what happened with terrible flight?
2:43:17
No, it's actually it's actually ends up being stressful because I'll be like on the show and I'll get a notification from watchd duty earlier this year.
2:43:25
It' be like a fire is popped up half a mile from you. Really rough.
2:43:30
>> It doesn't know where doesn't know where I am.
2:43:32
So I'll get notifications like all over LA County and I'm just you know I end up look I mean it's probably good for their their user metrics because I'll open it up and realize okay it's it's like a hundred bucks.
2:43:42
I think you need three things to really create an essential use case around this, which truly is an essential use case because it could harm your family and and yourself.
2:43:53
The first is you need to aggregate all the alerts.
2:43:54
So, Watchd Duty just does it for fires, right?
2:43:56
But you need power outages, you need construction delays, you need any kind of of traffic that's going on, you need some event that's going to change something in the neighborhood all the way to the really serious stuff like fires and tornadoes and hurricanes and crimes and that sort of thing, right?
2:44:11
So that's the first thing.
2:44:12
The second thing is you need to know where people live so you can get them timely information that's hyper relevant, right?
2:44:18
I mean, it's not relevant if it's half a mile away because it's not going to affect you.
2:44:22
It's just going to freak you out, right?
2:44:24
And then the third and final thing you need is you need to have a community of people that are notified because in many cases, they're the ones that have the information, the photos, and the videos >> that are actually more relevant than anything the authorities are providing because they're on the ground, >> right?
2:44:41
They know exactly what's going on.
2:44:43
And so we have all of those pieces and god forbid those crises happen, but when they do, we have to do our part in keeping people safe.
2:44:53
>> It's community intelligence.
2:44:53
It's like the collective together.
2:44:54
So >> it's the power of community. Truly. >> Fantastic. >> Awesome.
2:44:59
Well, great great chatting.
2:45:01
Congrats on the new launch.
2:45:01
We're you come back on again soon.
2:45:03
You're going founder mode.
2:45:04
It's great to see you back in the driver.
2:45:06
You do better open the app and if you email me, I will make sure you get verified immediately.
2:45:10
So, >> I'm sure it'll work perfectly now.
2:45:14
>> I think I think you're going to like it and I think your listeners will too.
2:45:17
It's very different than the old Next Door.
2:45:19
It's got all the goodness of the old app, but in a fresh new shell with lots of new features. >> That's amazing. Very exciting.
2:45:25
Thank you so much for hopping on.
2:45:27
We will talk to you soon. >> Cheers. >> Bye.
2:45:30
>> And if you're looking to explore a new neighborhood, get on Wander.
2:45:32
Find your >> happy place.
2:45:34
Book a wander with inspiring views, hotel, great amenities, dreamy beds, top tier cleaning, and 24/7 concier service.
2:45:39
It's a vacation home, but better, folks.
2:45:41
And now, >> Wander just dropped Wander Indo Haze. >> Haze, h A Z E. >> Oh, okay.
2:45:50
>> Uh, well, we have uh Mike from Arc AGI coming on the stream.
2:45:54
And we have the ability >> Arc Day. >> Arc Day.
2:45:59
>> I I I love these releases.
2:45:59
Thanks so much for always coming on the show. Um, >> of course.
2:46:03
Thank you guys for being such huge supporters.
2:46:05
Means a lot of this project. It's it's so amazing.
2:46:07
So we have the latest and greatest.
2:46:09
Uh why don't you uh just give us the a little bit of background on um the the >> first of all I think why don't we talk about the XAI launch last week because that was a you know pretty massive improvement.
2:46:25
Um so maybe start there and then we can get to the present. >> Yeah.
2:46:28
So, this is really, I think, the second time we've seen a major frontier um AI lab use ARC as a benchmark to show off some sort of frontier of progress. Right?
2:46:38
Back in December, we had uh OpenAI uh use ARCV1 to show off this qualitative change, right?
2:46:44
Really marked the moment where AI Frontier AI research moved beyond just like scaling up, retraining, and starting to add these symbolic systems on top, these chain of thought reasoning systems.
2:46:53
Um arc really marked that moment.
2:46:56
And then uh last week uh the XAI team used ARCV2 to show off uh a Frontier result on on ARCV2.
2:47:04
They got 16% on the benchmark. Still early days.
2:47:06
Um but I think one of the really big takeaways um from that from that was basically like how you know using a lot of the existing ideas in the world but how quickly and effectively XAI was able to catch up to the frontier.
2:47:18
And so, you know, my mental model now is, you know, going forward is, you know, wherever the front sort of innovation comes from, my expectation is XA is probably going to go beat for beat on on catching up just given how fast they're able to get there this time around.
2:47:30
>> When when I hear some like one one thread that I've heard from watching the ARC progress is uh you put more more test time inference, more spend behind a particular model, you get better results.
2:47:45
And it and it and it begs the question like is this brute forcing?
2:47:48
Is that what we're experiencing at some level? Is that is that a fear?
2:47:53
And is the latest ARC v3 uh an attempt to kind of avoid that or is that just like not an issue at all?
2:48:03
>> So, so no because of how we evaluate top scores on ARC.
2:48:07
So we publish along two dimensions. One an accuracy score.
2:48:09
M >> um but we also publish an efficiency measure and this is not arbitrary like efficiency is a really really fundamental aspect of what it means to be intelligent.
2:48:19
So when we you know we have the million-dollar prize it's still hanging out there by the way for the original version of the contest and you know if you look at um you know the human level efficiency scores just on V1 we're still only around you know 60ish percentage points whereas our benchmark for humans is 85 and and onwards.
2:48:37
So, you know, even though we've got two team 83, 85% or something, would they win the million or do they have to get to 100%.
2:48:47
>> Uh, so the Kaggle contest rules are a little more stringent.
2:48:49
Um, you have to do it on a certain performance profile, a limited compute budget, you have to get the high score and then there's an open source requirement.
2:48:56
So this is one of the things that um one of the sort of principles of the arch prize foundation is we're trying to basically accelerate AI genesis by encouraging more people to work on new ideas openly share those ideas try to you know kind of shape the I research community to look more like what it did during the 2010 to 2020 era than than maybe it has over the past you know three or four four years or so.
2:49:16
What about um I we we talked to somebody I forget who they they said like oh yeah like uh you know RKGI it's cool but like we could totally crush that if we just like RLED on it directly.
2:49:25
Uh and all the labs are just being like nice to like keep it as an independent benchmark.
2:49:29
But I don't fully buy that because I feel like some hedge fund would just see a million dollars on the floor and then just go do that if that was the case.
2:49:38
But uh what what is the what is what is the vibe?
2:49:42
reality is like benchmarks are marketing. >> Yeah. >> Right.
2:49:46
Uh this is like one of the reasons I I didn't appreciate this 12 months ago when we first launched AR. I do now.
2:49:51
>> Um you know the the whole reason I put the money into the contest in the first place was ARC's awareness was very low and my thesis was this is the most important unbeaten benchmark in the world.
2:50:01
It tells us something important that no other benchmark does. Right?
2:50:02
If you go even look at the Grock 4 stream all the other benchmarks that were shown are like PhD++ level benchmarks.
2:50:07
Y >> and yet arc tasks are like, you know, we have objective evidence.
2:50:12
We've done human controlled studies that show they're all solvable by sort of average humans.
2:50:17
>> Um and so I think that tells you something interesting like well okay yeah like we're clearly missing something big here.
2:50:22
We don't have it all figured out.
2:50:24
We're not in just like a scaling up regime.
2:50:25
We are in we're sort of in an idea constraint regime.
2:50:27
And I think that's an important conclusion because if it's true it means that individual researchers and small teams on small budgets can actually have a significant impact on the frontier of AI research.
2:50:38
you don't need, you know, a million, 10 million, billion dollar training budgets in order to actually make a material impact on AI. >> Yeah.
2:50:44
>> The the the challen like I mean the real challenge is just the distortion in the market right now and the trade-off that a that a researcher even somebody who's let's say a 20-year-old in college that could be doing this sort of independent research and they're like, wait, if I drop out, I could be making >> $500,000 a year base comp and and what you know the the equity >> on top of that.
2:51:06
So I imagine it would I imagine at some point are are there like high schoolers that that are that are doing this kind of thing because like at a certain point at a certain point like the target market is like people that like are maybe a little bit too young to like actually get it like >> Silicon Valley companies will happily have somebody drop out of college.
2:51:23
It's a different conversation when somebody's like I want to like drop out of like high school, right?
2:51:30
>> I mean this for forms my thesis.
2:51:30
I think talent is very distributed globally.
2:51:31
Um if you look at most of the teams that are on the leaderboard from you know past years of the art contest you know a lot a large percent I'm not sure if it's over 50% but it might be are outside the United States.
2:51:42
>> Um and again this is like you know if you're sort of trying to create a sort of optimization function for creating AGI uh you know you you would like to shape an innovation environment that's very open there's a lot of sharing and there's a lot of diversity of approach.
2:51:55
Um the opposite would be like there's very little sharing, everyone's working on the same ideas.
2:51:59
Um like and if because if those ideas are wrong then like okay you're sort of shooting yourself in the foot.
2:52:04
And so um those are I think that's that's one of the reasons we sort of launched our prize in the first place was try to try and help communicate the story that individual people young folks without with very little budget can actually have a large impact and encourage them if they have new ideas towards AGI to go work on those you know maybe as opposed to going and you know just starting the next language model uh startup. >> Yeah. >> Yeah.
2:52:24
When when Doresh came on the show, he was talking about the need to solve continual learning.
2:52:29
This idea that he has uh you know this amnesiac PhD that is un unable to learn hard lessons and then roll that up into habits and kind of wisdom almost.
2:52:43
Um and and and that's why he he's unable to, you know, use any of the frontier models to uh his example was like select which clips of the podcast will perform well on on social media or something and he was struggling with that.
2:53:01
And I'm wondering like we hear about the spiky intelligence concept.
2:53:05
Do you think that the the problems that underly uh Arc Prize's robustness are related to the same continual learning problem that um that Doresh was highlighting or are these two separate problems where we could see us solving one and not the other?
2:53:24
>> I think that's happening, right?
2:53:24
The scores are much higher in V1 than V2.
2:53:25
I I so if I kind of lay out and tease here kind of the the version we're doing a public preview for today, um you know, we've got V Arc V1, V2, and V3.
2:53:33
V1 was introduced back in 2019.
2:53:36
It was designed to challenge deep learning as a paradigm.
2:53:40
Remember this is before language models years before sort of language models really hit any sort of stride in terms of the research.
2:53:44
Um and it uh sort of was robust through that uh advancement and that's because language models sort inherit some of the same fundamental limitations that pure deep learning do.
2:53:55
>> V2 was designed to challenge this new paradigm of AI reasoning systems.
2:53:58
Um it's still a static benchmark.
2:54:01
So the puzzles look very similar to V1.
2:54:03
It it might actually be kind of surprising that like you can't beat V2 if you can beat V1 because they look like they're in domain from each other.
2:54:10
But totally >> the the intuition here is generally the V2 puzzles require um longer reasoning chains generally to solve them.
2:54:17
And so that gets harder to do.
2:54:20
>> Um one of the things that we started to see though this year is really the emergence of a lot of these agent systems that are being placed into dynamic open-ended environments.
2:54:27
And while static reasoning, I think benchmarks are useful and will continue to be useful.
2:54:32
Um, this is one of the motivations for starting to build um, B3 in defining what we're calling an interactive reasoning benchmark to help evaluate and really challenge some of these frontier AI agent systems we're starting to see emerge. >> Okay.
2:54:45
So, should we do the live demo?
2:54:47
>> I think we can giant like chair launch today, I guess. Uh, >> yeah.
2:54:51
So, so I can I can kind of read through this, too. Yeah.
2:54:54
Give us the overview and then and then we'll play.
2:54:56
It's a little It's a little unique for us because you might be surprised like, "Oh, wait. V3 is launching.
2:54:58
Didn't V2 launch like 3 months ago?"
2:55:00
Um, so today is a public preview.
2:55:02
We're we're showing off the first three public games from the eventual data set.
2:55:06
We're building them this year.
2:55:07
We intend to launch the full version in early 2026.
2:55:10
>> We're we're going about the launch a little differently than we did with V1 and V2 because V3 is such a big upgrade over V1 and V2.
2:55:15
We want to get >> And by upgrade you mean by upgrade you mean significantly more challenging.
2:55:20
the gap between what's easy for humans and hard for AI is getting wider again with V3 compared to what the earlier versions were, which I think is one of our other design principles we have.
2:55:32
Um, and so it's and it's also just like a very different domain than V1 and D2.
2:55:36
They look like arts, what they're dynamic, and there's a lot we don't know about them quite yet.
2:55:40
Both what humans can do, what they actually find easy, um, what AI agents can do, how much can you create, you know, custom harnesses and scaffolds to be able to maybe make progress.
2:55:49
I think we're going to learn a lot.
2:55:50
Um, and this is why we're sort of launching the first three games early to make contact with reality here and increase already learning over the next maybe month or so.
2:55:56
On our game design, our API design, we actually have we launched our first piece of infrastructure uh today as well, an API that you can actually build agents and go run against these first three games.
2:56:08
And we launched a $10,000 agent contest that's running for the next month for whoever can build the best agent that gets the best uh top score on the the games that were.
2:56:17
>> So even if they get 1% but they are the top even if it's low whatever it is money is going out the door.
2:56:22
>> One important thing like V1 and V2 ARC V3 has a public and private data set.
2:56:25
So we've got three public games.
2:56:27
There's also three private games.
2:56:28
Those are actually what we're going to be awarding on top score performance on.
2:56:32
So, if you're thinking like, "Oh, I'll just make a really good, you know, harness for the three public games."
2:56:37
It won't work cuz they won't translate into the three hidden games.
2:56:40
So, >> smart, clever, clever. I love it.
2:56:40
Uh, so, uh, I'm I'm sharing my screen to the stream.
2:56:45
I don't know if you'll be able to see it, but I will read through this.
2:56:50
So, uh, >> my suggestion is you guys should we should play Locksmith.
2:56:52
This is LS20, our first game.
2:56:54
I think you guys should just collaborate on this one game together.
2:56:57
It'll take about 5 to 10 minutes to probably play through.
2:56:58
And I think seeing both of you like work together on it live, I think, will be a fun, entertaining.
2:57:02
I should be able to see the screen.
2:57:04
>> Okay, I'll I'll I'll read it out.
2:57:04
So, um, human instructions.
2:57:06
You are playing a game ID.
2:57:08
There are no instructions intentionally.
2:57:10
You must play the game to discover controls, rules, and goal. Press start to play. Choose your controls. WD or arrow keys.
2:57:16
Uh, play to learn the rules of the game. Win the game. Profit. Just kidding. Just kidding. No prizes here.
2:57:23
So, I click start and I'm presented with a large grid of squares.
2:57:29
It looks like the most intense Arc AGI puzzle possible because the original Arc AGI puzzles were something like a 3x3 grid and now I'm seeing >> they're all 64x 64. >> 64x 64. Okay.
2:57:43
So, um I do have the ability to use um arrows to move this blue and orange block.
2:57:51
And Jordy, can you see me moving it around? Yes. Okay.
2:57:55
So, if I go on top of this, >> the bottom left hand corner updated slightly.
2:58:00
Let me see if I can zoom this out a little bit so that >> All right, you've learned one important thing.
2:58:05
>> Um, and if I keep moving, click seems to do nothing.
2:58:08
Spacebar seems to do nothing.
2:58:10
>> I'm going to give a little commentary while you're going here, too, John.
2:58:11
So, so, you know, the data set, the three games that we launched today, all of them are completely different.
2:58:16
None of them look similar.
2:58:17
In fact, this is the only game in the set that looks like a 2D agent game from kind of a tops down camera view that we're launching. uh right now. Okay.
2:58:25
All the other ones are quite different.
2:58:27
And this is actually a design goal of the benchmark is for um all the you know eventual hundreds of games we're going to have for them to be entirely different and very novel and diverse from each other.
2:58:35
>> So I I appear to have lost a life.
2:58:35
If I look in the top right I now have three I I had three red dots. Now I have two reset. I got a red flash.
2:58:46
>> And so uh I think I died.
2:58:46
Um but if I move forward I get a green circle which I imagine means I won.
2:58:52
And now I'm on a new level and there's >> Do you know why you like got to the new level?
2:58:58
Can you articulate that yet? >> Yes.
2:59:00
So I I believe that I stepped on a button that rotated my or or kind of rearranged the the icon the goal icon in the bottom left of the screen.
2:59:13
Um, at first I thought that the that the in the bottom left I see a little like like blue and white line and I thought the line was like a map that I have to take, but it appears to be a puzzle shape that I have to match up with a puzzle shape that's on the grid somewhere.
2:59:30
And I'm stepping on a button to change it.
2:59:33
So when I step on this rotate or change button, I'm getting kind of like a different Tetris piece.
2:59:42
And if I keep doing this, I might land on this. Okay, that matches now.
2:59:48
So the bottom left matches the the little goal. You see the goal? I go over to it.
2:59:53
But >> if I go over to it, I die.
2:59:56
And so I think I ran out of purple steps.
2:59:58
So I have a set of >> There we go.
3:00:02
>> Purple like like uh energy. I have energy. Yeah. Yeah.
3:00:05
I I I have I have energy.
3:00:07
So now I'm going to do the same thing.
3:00:10
>> But this >> is another very important uh design goal.
3:00:12
many of the games in ARV3 inherit which is there's intentional efficiency limits on actions that you can take.
3:00:19
>> So now green score I did it >> which is smart because the idea of like you know Scott Woo can probably like oneshot a math problem and it's very efficient for him and then someone else like is going to be able to solve the same problem but but like over weeks and is that really the same level of of intelligence?
3:00:35
intelligence? This is really, you know, you guys have probably like for for for a long time actually, I think games were considered a solve problem in AI, you know, with Alph Go and all the chess games and yeah and one of the reasons for this was like they they didn't the only thing that stopped them from being
3:00:51
like totally superhuman in those is that they just didn't scale the current algorithms enough and so they didn't solve you know it completely but they all were basically most of them use RL and so they're trying to take reward signal and understand you know what actions I took to produce the reward signal. This is one of the things that
3:01:05
This is one of the things that efficiency helps with is it limits the ability for an agent to just naively be able to go ex gather a reward signal by spamming and playing the games hundreds of thousands of times.
3:01:14
This is something humans don't need to do, right?
3:01:15
You know, you already beat level two uh in what less than 5 minutes here with a very limited number of efficient, you know, actions that you took.
3:01:22
And this is something we don't see from the frontier like LM state or or other agents we've been testing.
3:01:29
>> Okay, I got my next level.
3:01:29
But >> all right, the next level we we started introducing some new concepts here.
3:01:32
So, so I'm seeing they're different colors.
3:01:35
So, I I have blue and white and I need blue and y or blue and orange.
3:01:37
And so, I'm going to step on this >> uh this color block.
3:01:41
Now, I have blue and teal.
3:01:44
Now, I have blue and red.
3:01:44
I'm doing okay on efficiency. Have about half my life. Okay, that seems good.
3:01:48
But I think if I make a run for it, I won't make it.
3:01:52
So, I'm going to pick up this purple cube to refresh my energy. Run over here. Am I going to make it? Am I going to make it? I made it. There we go. >> Yes, there we go. Victory.
3:02:03
Okay, now that there's a different one, I'll start by changing the color.
3:02:08
Okay, I nailed the color. Blue and teal.
3:02:10
That's the end gold and the black cube.
3:02:12
Then I need to switch my icon from this one Tetris piece to a different one. Okay, I got that. No, that's not it.
3:02:20
I need the the little like chair block.
3:02:23
Okay, that's >> You're running out of lives, John. >> Roughly correct. Let's see. Okay, I refreshed. I refreshed. >> Go over.
3:02:30
I think this this other block.
3:02:33
I think this is a button.
3:02:33
I think I'm gonna run out of lives though. I need to go refresh.
3:02:39
>> I refresh just in time. I one away. >> One away. Okay. Let's keep rotating. >> Keep rotating. Okay. Now it matches. >> There we go. >> Okay.
3:02:48
And I'll just pick up the free energy. Boom. Green.
3:02:52
>> You might have noticed scaffolding new uh new things you have to learn, right?
3:02:56
There's a progression system here.
3:02:56
It's not just learn one rule in level one and apply it for the entire game.
3:02:59
But we found a really an element one design goal is that all the games are fun. >> Yes.
3:03:05
>> And one of the things we found where we're doing early design game design was that um folks did not find the games fun if they just took one rule they learned and did that just repeated it. Right.
3:03:12
So introducing new things you have to continually learn throughout the game is >> a big function of whether humans can find these things actually entertaining and fun.
3:03:20
>> Well, we we should figure out the infrastructure to have pure PvP speedruns of the entire prize. the whole team.
3:03:28
>> Um, real real quick while we have you, we have our next guest in the uh in the waiting room, but uh I wanted to ask you because it is top of mind for us this week.
3:03:37
I I think like the broader tech community went from sort of like not like like taking like AI safety and alignment, you know, super super seriously or kind of making jokes about it until this week.
3:03:47
I think AI psychosis has been top of mind for a lot of people.
3:03:52
If you were running one of these scaled labs today, how how would you be trying to sort of quickly react to some of the um some of the different kind of stories that seem to be bubbling up around um people just getting like too, you know, too deep in this sort of recursive prompting.
3:04:11
>> You know, my my sort of view generally on a safety stuff is you want to be empirical about it.
3:04:15
Um, you know, this was sort of my big issue when we were going through all the 1047 legislation last year in California is trying to make predictions about what future harm might happen by being able to predict the future and in some cases poorly predict the future.
3:04:30
I think even ARC, you know, was de sort of clear demonstration of an eval that suggested we were not just scaling up pre-training.
3:04:37
AGI is not just going to emerge from scaling up this pre-training regime.
3:04:40
I mean that was sort of the predicated thesis on why we needed something like 1047 at the time to like stop this imminent urgent potentially dangerous scaling.
3:04:48
>> Um so now I I actually think uh you know my my sort of view probably aligns quite closely with opens.
3:04:52
I think you actually need to deploy the technology into the environment into the world in order to make contact with reality and learn what are the actual issues that you care about what does society care about.
3:04:59
Um my my sort of view is that like uh society is actually better at dealing with um fast change than slow change.
3:05:09
Uh, you know, this is another kind of counterpoint that I think if you go look at the safety community would argue, oh, slow slow takeoff is better than fast takeoff.
3:05:15
And I think in a lot of capabilities actually fast takeoff is slight is is perhaps desirable because humans notice change.
3:05:20
This is like literally what we were evolved to do is in our environment, we notice when things change fast.
3:05:25
We don't notice when things change slow, right?
3:05:26
The >> frog boiling in the pot is sort of the classic here, right?
3:05:29
Um, and so I think fast change stories like this, issues with psychosis are good in in a weird way because they uh they um are sort of societal antibodies of like, oh, hey, something changed here. We should react to it. >> Yep.
3:05:42
>> Yep. No, that's really that's kind of my sort of broad framework of what what what I'm what what I think how we should sort of like run these >> in in some ways the the the negative externalities of social media like let's say like somebody's developing body dysmorphia like the the the ways the
3:05:58
social media spectrum changed from like sending like group messages to suddenly like you're sharing your entire life like it actually was very slow and so I think that this development with uh with >> that creeps in right you wake up 10 years And you're like, hm, are we happy with like where we got to? >> Brain rot is a meme that took years to
3:06:14
>> Brain rot is a meme that took years to develop. Fascinating.
3:06:16
You have to come back on soon. We could go way deeper.
3:06:20
We're going to be playing this all day.
3:06:21
>> Hope you guys had fun playing the first game. >> Fantastic.
3:06:24
I was not expecting a game.
3:06:24
I was expecting a puzzle and we are clearly in game territory. Very fun. Uh, very exciting work. All right, guys. Thanks for having me on.
3:06:32
>> We'll talk soon, Mike. Bye.
3:06:34
>> Uh, up next, we'll stop and waiting is Kamani. Welcome to the stream. Sorry for the wait.
3:06:39
We were we were proving our humanity with Arc AGI V3. Welcome to the stream. >> In the suit. >> In the suit.
3:06:47
You look fantastic on a special day.
3:06:49
>> You didn't have to wear the suit just for the show. Come on.
3:06:54
>> And look, I wanted to kind of outdo y'all a little bit, too. So, >> fantastic. >> Look looking sharp. >> Looking sharp.
3:07:00
>> Are you in Are you in DC right now? >> I am in DC.
3:07:02
I'm in a little phone booth about two blocks from the White House. >> Fantastic. Give us the update. What's happening in DC? Um, yeah.
3:07:08
Let me turn my phone here real quick. Um, >> awesome guys.
3:07:13
Thanks for having me back on. Pleasure to be here.
3:07:14
Um, today's a big day for for Cryptoland.
3:07:17
Uh, President Trump just signed the Genius uh bill into law.
3:07:20
Um, yes, huge huge day for industry.
3:07:24
This is the uh probably most consequential piece of financial legislation since Frank DoddFrank.
3:07:31
>> Um, it's the first crypto legislation um we've ever had.
3:07:34
we've ever had. Um, and just an incredible amount of >> which just crazy to say out loud this this you know over how however far we are into this uh >> it's been like crypto has been a thing for like almost 20 years now like what Bitcoin >> it's a thing that you've been basically
3:07:52
begging for just like give us some regulatory finally here it's great >> finally here >> yeah I mean industry has been begging for regulation under the Biden administration and they said nope no clarity we're just going to shoot randomly uh and you know finally like we actually can operate. It's pretty It's pretty amazing. >> Fantastic.
3:08:11
Um uh what does this mean for uh non-stable coin chains?
3:08:15
Like what does this mean for Salana?
3:08:18
What does this mean for Bitcoin?
3:08:19
Is there any implication there?
3:08:20
Um or is this purely >> priced in?
3:08:26
>> Uh yeah, it's definitely not priced in.
3:08:29
So first let me touch quickly on like what I think it means for the US and then can get into kind of like the the chains and stuff. Yeah.
3:08:35
>> So I think the correct way to think about this this is um you know there's 8 billion people in the world and um I have a theory that if you were to go to each of those eight billion people and ask them hey if you can denominate your wealth in any asset it could be Apple
3:08:49
stock S&P 500 bars of gold euros yen yuan whatever you want if you could denominate your wealth in any asset without fear of political persecution in your local jurisdiction what asset would you choose and my suspicion is that somewhere between 60 and 80% of the world would say US dollars. >> Mhm. >> Mhm.
3:09:06
>> Um and for people who don't live inside the United States, having a US dollar bank account is somewhere between very difficult and impossible um outside of like the top 1% or so of the of the global global wealth.
3:09:18
And stable coins make it trivial for anyone in the world to host to have dollars in their pocket.
3:09:26
Um you don't need to sign up for permission.
3:09:28
You just you take your phone, your phone literally picks a random number, that's your private key, and you're you're good to go to receive dollars.
3:09:34
Um, I think this will probably represent one of the like largest movements in capital in human history as you enable, call it, five to six billion people to like start holding more dollars more easily than they otherwise could hold them.
3:09:47
Um, and the scale of what it means is actually really profound and I think really underappreciated.
3:09:51
uh which is why uh I think it's not priced in because it's just hard to grasp what this means for everyone in the world to be able to hold one currency.
3:10:00
This has never happened in human history before. >> Yeah.
3:10:03
>> Um >> this is fascinating.
3:10:04
I mean I'm I'm I'm like I I'm almost like zooming out like I'm sure it's good for a lot of different projects and companies and tokens and chains, but it feels just incredibly bullish for America.
3:10:15
like we've been because there was this narrative of like oh the tariffs and stuff like we're going to lose dollar dominance where like we're going to lose global reserve status and this just feels like coming from the the halfcourt shot. >> Yeah.
3:10:28
It was like an artificial you know it was like a supply demand imbalance but the supply constraint was just coming from >> like you don't have regulators in in other places that said you can't set up a dollar US dollar bank account in our country.
3:10:43
We're just not going to allow that. >> Exactly. Yeah.
3:10:44
You don't have to be like even cryp pro crypto or crypto native or anything like that for America.
3:10:50
>> And this is like you know going back sort of the core tenants of crypto like permissionless finance.
3:10:53
It's like okay like permissionless access to US dollars where somebody says I want to get paid in dollars.
3:11:00
They set up a wallet and it's software and they get paid and it's it is >> it's very cool.
3:11:06
I mean it's very much capitalism like just applied to money like if you identify as a capitalist like you should pause and think oh my god like actually the world as a whole was actually shockingly anti- capitalist in terms of just like currency access nations. >> Yeah. Yeah. It's wild.
3:11:21
Uh what else should we talk about today?
3:11:24
Uh is there any other downstream things?
3:11:26
Uh can you give us a a temperature check on the rest of the regulation? What happens next?
3:11:31
Uh anything else that's going on in Washington that we should be following?
3:11:35
Yeah, I mean downstream implications, there's a ton.
3:11:37
So, first a quick regulation.
3:11:40
So, the Genius Act passed, that's the stable coin bill.
3:11:42
Um, the next major act the industry is focused on is now called the Clarity Act.
3:11:48
Y >> um this is being uh primarily sponsored by uh Tar French Hill in the house.
3:11:50
Um and this is otherwise known as kind of the market structure act.
3:11:55
And the basic kind of outline of of clarity is to answer the question which regulator should regulate what what should they they regulate?
3:12:02
This really kind of defines the lines between Treasury, SEC, and CFCC.
3:12:05
Who has kind of, you know, perview over which domains?
3:12:08
Uh, answer the question, what is a security, what it's not?
3:12:12
It deals with DeFi to some extent.
3:12:14
Um, and so that's kind of the other really big meaty bill.
3:12:17
>> Um, and then the third one is actually one that kind of got added in recently and industry is fine with is just basically this anti-CBDC bill.
3:12:22
Basically just saying like America won't make a CBDC.
3:12:25
Um, so that that's all in in flight now.
3:12:28
You know, I'm optimistic clarity will pass in the next few months.
3:12:31
Um, it passed the House earlier this week.
3:12:33
Um, it's now in the Senate and there's going to be some revisions and stuff in the Senate. We'll see where it goes.
3:12:38
>> Anything anything that's that's in there so far that you're that you think that that that the Senate should should focus on correcting?
3:12:47
>> Yeah, there there's definitely more disagreement in industry about like, you know, the the substance of clarity.
3:12:50
I can't get into the specifics, but like certainly we're we're right now in dialogue with a lot of our peers and um lawmakers and their staff on on these issues. >> Makes sense. Makes sense.
3:13:02
>> The other question you asked was like what's the implications for like industry?
3:13:04
I really want to emphasize that I mean >> by making it it's now legal stable coins were illegal for all practical purposes for like any me major regulated company kind of anywhere in the world.
3:13:17
>> Um and it is now legal and blessed for them to interact with stable coins.
3:13:20
And so what I think yeah you're going to see every bank you know Jamie Diamond Bank of America CEOs are talking about this now.
3:13:27
I actually don't think that's the super relevant angle to me.
3:13:29
The much more important angle is it is now legal and okay for iOS, Android, Facebook, WhatsApp, Instagram, Tik Tok, every piece of major consumer software in the world now can legally embed stable coins. >> Wow.
3:13:42
>> Um, and it's going to be a matter matter of time before they do.
3:13:44
And what that's going to mean for global commerce, for global payments, for people accessing crypto is profound.
3:13:49
And I think you're and and as all those people get their private keys and onboard to these wallets um that's then going to obviously provide um the the foundation that that that money is going to be transacting over Ethereum, Salana, etc.
3:14:02
And it's going to be just a massive boon for all of crypto as you on board a few billion people with this stuff. >> It's fascinating. >> Absolutely wild. >> Yeah, I'm excited.
3:14:10
I imagine that Yeah, the the big tech angle is super fascinating.
3:14:14
I I imagine that some of them will take different paths, different shapes.
3:14:17
It's like we're going to test different structures. >> Yeah.
3:14:20
What I hope is like, hey, stable coins are already there there's large market caps.
3:14:25
There's lots of liquidity.
3:14:27
Let's not have everybody launch their own stable coin.
3:14:29
Let's just like focus on actually integrating them into and making them, you know, more valuable in these different ecosystems.
3:14:34
It's also funny because I feel like for a while there was a little bit of a uh there's a little bit of like a chattering class critique of crypto being like well if it was so great like why why why doesn't Apple just integrate stable coins and it's like well because it was illegal and like they're a very riskaverse company.
3:14:51
There was no chance they were going to do it but now they actually have it as an option and then there's a competitive dynamic and they'll probably run experiments and yeah maybe it'll take a few years for them to build some stuff.
3:15:00
Maybe some stuff will be aqua hires or acquisitions.
3:15:01
There's a bunch of different thing ways it can play out, but it'll be really interesting to follow like how this actually gets in the hands of like billions and billions of people. Fascinating.
3:15:10
>> Yeah, it's an incredible opportunity and I mean you just look at like now where's the liquidity, where are people trading today and you know um >> and like which chains have to scale to support all these users and it's just it bodess incredibly well for for industry as a whole. >> That's fantastic.
3:15:26
>> Well, as there's more news, come back on and uh I'm sure you'll be celebrating responsibly in uh in DC. So have fun out there.
3:15:34
>> Hey guys, thanks so much for having me back on. >> Have a great one. We'll talk to you soon. Bye Kyle. >> Talk to you soon.
3:15:39
>> What a way to end the week.
3:15:40
>> Fantastic little news hits, little chopping it up with friends. Uh impro fun. I I enjoyed that.
3:15:50
>> One of the most entertaining >> Yeah. >> people on earth. >> Indeed. Indeed.
3:15:54
We should close with uh this this wonderful article in the mansion section. >> Which one are we?
3:16:01
We we had a few queued up.
3:16:03
So >> So this is I wanted to I wanted to do this because it is essentially written by Arnold Schwarzenegger.
3:16:08
So it is it's his words as told to Mark Meyers who just transcribed them and then they published it in the Wall Street Journal.
3:16:18
And it's one it's Arnold Schwarzenegger the moment he fell in love with America. >> Uh beautiful.
3:16:24
>> My early years in a in Austria were challenging.
3:16:27
After World War II, the economy in the Graz suburb where I grew up was shattered. Everyone suffered.
3:16:31
My father Gustav was very very was a very very sweet man.
3:16:35
But when he drank, his personality changed.
3:16:37
He became more violent and demanding.
3:16:40
His behavior and drinking were influenced by the war's remnants.
3:16:44
Shrapnel in his legs that caused him pain, the len the lingering effects of malaria, and what we now call post-traumatic stress disorder.
3:16:51
My mother, Aurelia, was a homemaker.
3:16:53
Our family lived on the second floor of a three-story building.
3:16:58
The first floor was occupied by the local forest ranger and the second wars uh and for and the second war is for my father.
3:17:05
Uh the second floor was for my father.
3:17:07
The town's police chief um he was the town's police chief.
3:17:12
The forest ranger had a phone but we didn't and there was no running water.
3:17:16
I wasn't happy with the life I saw unfolding.
3:17:18
I always had a feeling deep down that I needed to look for something else, something outside the box.
3:17:24
At age 10, I fell in love with America.
3:17:26
That came from watching film roles in school.
3:17:29
The teacher would advance the strips by turning a knob, showing one image at a time. I was blown away.
3:17:36
They were about things like the Golden uh like the Golden Gate Bridge in the Empire State Building and cars with huge fins driving on US highways with six lanes on each side.
3:17:44
At some point, there was a role on Hollywood.
3:17:47
I'd never seen anything like it.
3:17:49
The glamour, the lights, and the houses.
3:17:51
I said to myself, "What am I doing here?"
3:17:53
I wanted to be in America to become famous and rich.
3:17:57
As I got older, the question shifted to, "How do I get there?"
3:18:00
At 15, I stumbled into bodybuilding at our local lake and said to myself, "Well, that's in America."
3:18:07
The lifeguard always had other top athletes around, including weightlifters.
3:18:12
As a teen, my testosterone was kicking in, and I wanted to look like a He-Man.
3:18:15
I read about Roy Reg Reg Park, an English bodybuilder who played Hercules in Italian movies. He won Mr.
3:18:23
Universe three times and became an actor. My dream was possible.
3:18:27
All of my time was spent in this world of physical fitness, building up muscles to complete to compete in contests and fantasizing about movie stardom.
3:18:36
At one point, I was in school looking out the window and daydreaming when a piece of chalk hit me in the head.
3:18:43
My teacher said, "Arnold, what do you think you're doing in here?" talking to myself.
3:18:50
I had no interest in what he was saying.
3:18:52
He's not interested in the school.
3:18:52
In 1967, I won the amateur Mr.
3:18:54
Universe title in London when I was 20.
3:18:56
Then I trained in Munich for another year and won my first professional Mr.
3:19:01
Universe title after I won in 1960.
3:19:05
>> It's so funny this the the thinking of the 60s and and and just like hippie hippie culture and then Arnold's just like rising the ranks.
3:19:14
>> Yes, it's bodybuilding. It's amazing.
3:19:17
Uh, so he said in 19687 I won the amateur Mr.
3:19:19
Universe title and then he won the first professional universe title.
3:19:23
After I won in 1968, Joe Wider, a bodybuilding enthusiast who published Muscle magazines, brought me to Los Angeles.
3:19:29
He put me up in an apartment in Venice near Gold's gym where I trained.
3:19:34
When I arrived, let's hear for Gold's production team.
3:19:36
You guys love that place.
3:19:38
>> That's where I work at. It's a great spot. >> It's fantastic.
3:19:39
Um, when I arrived in LA, I was totally dis I was totally disappointed.
3:19:45
The city looked nothing like New York City with its tall buildings.
3:19:49
Venice's sidewalks were dirty and no one cleaned them, and drugs were sold in back alleys. Two more Mr.
3:19:55
Universe titles followed.
3:19:55
In 1969 and 1970, from the start, I knew bodybuilding was going to lead to acting.
3:20:01
In 1970, I was cast in Hercules in New York and then in the TV series and films, including Stay Hungry in 1976, for which I won the Golden Globe Award for best acting debut.
3:20:12
Then came The Streets of San Francisco and Pumping Iron in 1977.
3:20:18
>> Pumping Iron is fantastic.
3:20:18
If you haven't seen it, I know you haven't.
3:20:22
Conan the Barbarian in the big one in 1982.
3:20:24
I was quoting Conan the Barbarian to our team.
3:20:27
I don't think anyone got the reference, but we'll play the YouTube video. >> This is wild.
3:20:31
So, he says, "In the middle of all this, I studied remotely for a bachelor's degree in business."
3:20:37
Let's give it up for all the the business majors out there.
3:20:39
Most important major clear >> at the University of Wisconsin.
3:20:43
>> He was doing that remotely.
3:20:45
>> How were you doing 82? >> College in ' 82.
3:20:47
I guess mail it mail it back.
3:20:50
>> Oh, you maybe get mailed an assignment back.
3:20:53
That is Yeah, that is crazy. How do you do that? I don't I have no idea.
3:20:57
From the time I arrived in America, I went to >> He invented remote work.
3:21:00
>> Arnold invented remote work. >> He really did.
3:21:02
Yeah, that is a crazy crazy thing.
3:21:04
U we'll have to have him on the show and ask him about his his time as a remote business major at University of Wisconsin Superior.
3:21:10
Uh I went to community college to take English classes and get smart about business.
3:21:13
In 1983, I became a US citizen.
3:21:15
That's Aaron for Arnold Schwarzenegger.
3:21:18
One of the best to ever do it.
3:21:19
Today, I live in contemp in a contemporary house in Brentwood, Los Angeles.
3:21:24
It's the perfect place where I can see the foliage and the mountains. I'm close to town.
3:21:28
It's private and I have my pet I have all my pet animals.
3:21:32
Uh which means not just dogs.
3:21:35
He definitely has more than that.
3:21:38
Uh we'd love to know what he has.
3:21:38
Uh in the 1980s, I took my mother to the white to a White House dinner to meet President Reagan.
3:21:44
At the table, I went to scratch my nose and she hit my hand in front of everyone.
3:21:48
She said, "Don't pick your nose in the White House, please, Arnold." I wasn't.
3:21:52
But that didn't stop her.
3:21:54
No matter how big you get, your mother knows how to shrink you just a little. What a great story.
3:22:01
>> I uh am sad that I was not super conscious when Arnold was the governor of California. >> Yeah. Great. >> Like I I remember it. >> Yeah.
3:22:13
>> But I I I wasn't there for probably all the incredible uh moments.
3:22:16
Um, and uh, I don't even have a strong take on if he was like a great governor, but uh, I think he's awesome as a as an individual. >> Totally.
3:22:29
>> And uh, >> that's a great place to be.
3:22:31
>> I hope Sam Sam Sulick I hope he makes a run in politics as well. >> Me too. I would love it.
3:22:36
>> I think that would be powerful. >> And Hollywood first.
3:22:38
>> Can you imagine daily vlog Sam Sulick vlogs from the campaign trail? >> Incredible.
3:22:44
>> He's just getting all of his constituents.
3:22:45
Hey, just come for a lift. Come for a lift.
3:22:47
I think we're I think we're manifesting it. >> We can talk policy.
3:22:50
>> We get him in a Jason Carmen directed reboot of the Terminator. >> Yeah.
3:22:54
>> Get him into Hollywood. Get him trained up.
3:22:56
Make him a household name more than he is already.
3:22:58
Then get him on the campaign trail.
3:23:00
I think we have a winner in Soule.
3:23:03
>> Well, this was a crazy, wild, often fun week and I can't wait for next week. >> Yeah. or we'll see you Monday.