0:03
[Music] [Music] Scary. [Music] Wow. [Music] [Music] Heat. [Music] Heat. Heat. Heat. N.
Search this video
0:52
[Applause] [Music] [Applause] [Music] Heat. [Music] Heat. [Music] Heat. [Music] Heat. Heat. Heat. N.
1:59
[Music] [Music] [Applause] [Music] Heat. Heat.
2:55
[Music] [Applause] [Music] We came to this [Music] world to reach the stars.
3:18
We came to this [Music] world to see our future.
3:28
[Music] We came to this [Music] world to shape our future.
3:45
We came to feel the new way. Heat. Heat. [Music] N. Heat. Heat. [Applause] [Music] N.
4:16
[Music] [Applause] [Music] Feeling freaking [Music] [Applause] [Music] You're watching BPB, brother.
4:54
Today is Friday, May 16th, 2025.
4:57
We are live from the Temple of Technology, the fortress of finance, the capital of capital.
5:03
This thing going in my mouth is really bad.
5:08
Yeah, the amount of microplastics going into our mouths right now, it's terrible.
5:11
Anyway, we got a great show.
5:14
We got Matt Grim coming on to the show.
5:16
Uh Andrew just did the Murf challenge.
5:19
He's going to break it down for us.
5:19
We got Kevin Wheel from Open AI, Blake Scho from Boom Supersonic, Tim Fist from IFP, Chris Best from Substack, and Sean Henry from Stored.
5:28
Some legendary founders, some legendary folks, some yappers, some commentators. We got a great show. We're on the news.
5:35
But first, let's bring in Matt Grim if he's in the studio.
5:38
If not, we can show you some photos. Oh, here he is. How you doing? Hey guys, how are you? Fantastic. Looking great.
5:45
Uh, can you break it down for us what's going on today? How did it go?
5:50
It went uh it was a great day.
5:52
Uh, I'm here with uh Chris Wy.
5:55
Chris Wy is the executive director of the Lieutenant Michael P. Murphy Navy Seal Museum.
5:59
So, I wanted to give it a second to Chris to talk about uh what the Murf Challenge is, why we do it, kind of the story behind it, and a little bit of Lieutenant Murphy's history. So, sounds fantastic.
6:07
Uh thank you for having me on.
6:10
So, uh what we just accomplished today was a onem run, 100 pull-ups, 200 push-ups, 300 air squats, and another onem run.
6:17
Uh all wearing a nice weighted vest, as you can see. There we go. Yep.
6:22
But the whole uh history behind this was that it was Michael's favorite workout when he was deployed.
6:28
We don't deploy with gym equipment, so you have to kind of improvise and find what you can uh can use in your area so you can have a good workout.
6:37
Uh if everybody doesn't know the story of Lieutenant Murphy, the blockbuster movie, The Lone Survivor with uh M uh Wahberg and uh Taylor Kit, that was the portrayal of Operation Red Wings and how Michael stepped out into a hail of bullets to make a phone call to try to save everybody.
6:56
So ultimately receiving the Medal of Honor uh after he died that day and just a great workout in a great way to remember all that have s sac sacrifice I'm sorry and suffered uh our own little discomfort today during the the Murf challenge. It's awesome.
7:15
Uh how long has this been going on?
7:17
Uh Andrew's been a part of this for a couple years, but can you break down some of the history of this particular event?
7:22
So, what happened was after uh the helicopter pilot recovered Michael and the other guys on that mountain side, uh when he got back home, he uh heard of Michael's workout and then started doing it on his own, calling it the Murf and it spread so quickly through the local CrossFit gyms and then it became uh too much for him to handle and then he involved a family.
7:42
So, this has been going on as the Murf Challenge for about 12 to 13 years now and it's just growing every single year.
7:50
Uh we're hoping to uh continue having this grow and have the partnership with Andrew because they're amazing amazing partners and uh and people here.
7:58
It's it's been such a touching event today. That's amazing.
8:03
Can you talk about Ander's role here?
8:05
Is this just a bonding event for you guys? Is this for charity?
8:07
Uh how are you thinking about Anderrol's involvement in your How many people from Andre came out for this?
8:13
Well, today we had about 150 Anderillians come out and it's uh for us it's not just about the team bonding part.
8:20
That is certainly a part of it.
8:21
You know, kind of team morale, team bonding and the team that uh works together, sweats together, stays together.
8:26
So, that's certainly a part of it.
8:27
But more importantly than that, like we started Andrew to bring the best possible technology to those who serve in defense of our freedoms and our nation and our allies.
8:34
And uh and a part of that extends to supporting the veterans community.
8:38
So, we've been pretty active in uh supporting the veterans community through a couple different ways that I can talk about in a couple minutes.
8:43
So today was uh less for us about the morale piece and more about a um sort of a signal or a um a moment to celebrate our veterans both that work at Andre and in the community at large and um our partnership with the Murf Foundation and with Chris himself and and all of that is just a a symbol of that.
9:02
How'd you wind up doing Matt?
9:04
I tried to do it with you over Christmas and I got crushed.
9:07
Well, you guys don't even look like you broke a sweat. What's going on?
9:11
Yeah, I I appreciate that. Do another one now.
9:13
We we had a we there were 150 of us out there and um I I noticed there were there were two empty vests out there with technology on the front.
9:21
So if you guys uh want to want to step up next year, we'd be happy to have you and you can come out with us. We'll do it live. Yeah.
9:30
I did like three pull-ups today, so I'm getting there. You were repping.
9:34
I think I did a couple sets of five, but not quite at the hundred in a row yet.
9:38
Um, talk to me about Chris for for your time and thanks for the partnership and your service and everything you do for veterans. Thank you so much. Yeah, thank you. Um, I got all right. Cheers. Awesome.
9:50
Matt, uh, talk to me about uh, the role that veterans play at anderol.
9:56
Uh people think tech company, military, obviously there's discipline overlaps, but a lot of times there's not so many immediate skill overlaps or do people have that wrong?
10:09
Are you hiring software engineers from the military?
10:12
Are you hiring hardware engineers?
10:13
Are you hiring operators and finance people?
10:15
What are the different types of roles uh that veterans are filling at Andreal today?
10:21
Yeah, happy to talk about uh talk about all of that.
10:23
So, we've been um involved in the veterans community since day one of the organization.
10:26
one of our coh one of our co-founders in fact is a veteran himself and has helped kind of uh lay the foundations for those partnerships.
10:34
Right now about 12 13% of our employee base is veterans which is wildly over the national average and uh sign certainly higher than that in in the Silicon Valley kind of technology community.
10:44
We have a we have a partnership with a great organization called Skillbridge that's a program run by the DoD for helping um veterans transition.
10:51
We have a couple of uh recruiting relationships with uh targeted to that kind of transitioning veteran kind of community.
10:56
And we hire across all roles.
10:58
Everything from kind of field op field ops technicians who are installing our products and training users in the field, some of our maintenance and repair technicians, some of our production technicians on the factory floor building products all the way through some design engineers, uh mechies and ees doing work.
11:13
Yes, we have some coders, some CS folks who are uh from from the veterans community and all the way up into the leadership ranks, you know, of our of our leadership ranks.
11:22
We've got um a pretty high v veteran representation.
11:24
Everything from flag officers to retired colonels through uh through the whole ranks.
11:28
So, it's a big part of our part of our culture.
11:31
It's a big part of our uh kind of the mission of the company and something that we're proud to support.
11:36
And beyond that, I would say that uh we frequently raise some money for veterans charities.
11:41
We've got a couple in particular that we support.
11:42
I've got my my my notes here in front of me with that.
11:45
Uh so on uh Veterans Day last year, we had our first big uh public launch of the swag store and we had a swag drop that uh raised about 60,000 some odd dollars uh that went to a Blueest Star Families Foundation.
11:56
Um a good teaser for you, breaking news here on the Technology Brothers podcast network.
12:02
We're launching our next swag drop on Memorial Day in about 10 days. Let's go. Oh.
12:06
Uh, we'll be raising a whole bunch more money exactly for the foundation that that that Chris runs in, uh, in Lieutenant Murphy's honor.
12:13
So, uh, everybody be be be on the lookout for that.
12:16
Uh, we've given some money to a great charity called Warriors of Field that is, uh, actually run by one of the Ander executives that specializes in, uh, helping veterans kind of in in distress, kind of going through some bad times by doing some mentorship and some outdoor activities and wilderness adventures and hunting and that sort of thing.
12:33
So, trying to build some bonds there to help veterans who are in a tough spot.
12:36
So, uh, we've been we've been very active through all the years on all this and something that we're looking forward to, uh, continuing doing and scaling as the company grows. Okay.
12:44
Give us a take about how to get a job at Anderoll.
12:46
Well, we've got a a great website, you know, and.
12:48
com/careers, but more importantly than than the obvious, just apply on the website line is like we look for people with a real mission drive.
12:56
We look for people who are really driven by the call to service, who are really uh interested in protecting uh American values and our allies values and um and and and the escalating arms race that is happening across the world.
13:08
Whether that's artificial intelligence, whether that's drone technology, whether that's any sort of the the kind of the next generation of what the war war fighters are going to face face in the battlefield.
13:16
like we look for people who are just really motivated and really driven and look uh you know want to come honestly grind and work hard on really challenging and rewarding problems to to help defend our country.
13:25
So um the the number one thing you could do if you're interested in a job at Android uh obviously apply but more importantly is um you know build up that skill set, build up that engineering skill set, that problem solving skill set and um and that mission drive and uh that's who we look for. That's fantastic.
13:39
Jordan, you got anything else? No, this is great.
13:42
We got to come out for the next one. We got to start working.
13:43
I mean I think it's doable.
13:45
I don't I think I would have died this year, but I think I could do it next year if I start training today.
13:49
John finished the winners today finished in our in our women's division.
13:55
She finished in about 45 minutes.
13:56
Our in our men's division, he finished in 35 minutes doing a mile run, 100 pull-ups, 200 push-ups, 300 squats, and another mile run in 35 minutes, which is uh totally ridiculous and and very impressive.
14:09
Um and then me myself, I finished in about about an hour and 11 hour and 12 minutes. Uh, not bad.
14:15
But you had you had three vests on, but but finished it with the full vest, full reps, all of that.
14:20
So, pretty pretty pretty happy day. That's amazing. Absolute dog. Absolute dog. Well, congratulations. Thanks for coming by.
14:28
Uh, love to have you back and to talk more, but enjoy the rest of the day. For sure.
14:32
Thanks for putting this on. Bye.
14:36
I think it's time we take off the mustaches and get down to business. Oh. Uh, yeah.
14:42
a little I I I thought it might be fun to do a little bit of the news in a mustache.
14:45
Uh well, you can stay in a mustache.
14:48
I'm personally going to keep mine on a little bit, John.
14:50
Well, first up, um we got Sam Alman with a big announcement.
14:54
Uh OpenAI is introducing Codeex.
14:56
It's a software engineering agent that runs in the cloud and does tasks for you like writing a new feature or fixing a bug.
15:00
You can run many tasks in parallel.
15:03
Uh starting to roll out to Chat GBT Pro, Enterprise, and Teams users.
15:08
And he sends some more info.
15:10
Um, I think this is interesting because I had this crazy experience with 03 the other day where I wanted to know how uh how high a desk was actually on the Pat McAfee studio.
15:21
So, I was looking at so I took a screenshot of Pat McAfee standing and I and I said, "Okay, look up how tall Pat McAfee is and tell me how tall is his desk because he seems to have a good knowing where he stands." Yeah. Yeah.
15:34
When you were doing that when when Oh, when I was thinking just thinking Yes.
15:40
Um, it was struck my thinking.
15:40
But, but this is what what was crazy is that I thought I thought it would just like look at the image and just kind of guess.
15:47
It wound up writing like hundreds of lines of code, interpreting the pixels in the image and looking at different elements of the image to decide, okay, well, there's a he has a Coca-Cola there.
16:00
This is how tall a Coca-Cola is. Let me expand this.
16:01
All this just to find that it's just a normal size desk.
16:05
It's just a standard desk which is like 36 inches.
16:07
Can you imagine giving that task to a human to say like can you can you kind of figure out how high this desk is? Yeah.
16:13
And they go and they write hundreds of lines of code and you're like like I guess like good job for going above and beyond, but like a simple estimate would have been fine. Yeah.
16:24
And I also and I also had an interesting experience where uh today I kicked off a deep research report asking for some you know top news and and put together some some news of the day like what should I be thinking about and the deep research report asked me hey do you want me to run this once or do you want me to run this just like every day or every week?
16:42
I'd be happy to do that for you.
16:44
And I was like oh wow it just it just volunteered to do it on a cron job which normally you would have to write about.
16:49
Uh anyway any other takeaways from the OpenAI launch?
16:53
talking to Kevin Wheel uh in a little bit.
16:54
But no, I I don't think this should be a huge surprise to anyone.
16:56
Uh we, you know, we talked with Sarah uh probably a few weeks ago at this point and she just said, "Look, to understand, you know, where the labs are going, understand what they value, what they find important.
17:09
Codegen has always been one of those things."
17:11
And so, uh this launch, uh should not be a surprise to anyone, but I'm excited to get in deeper with, uh Kevin a little bit. Yeah.
17:19
I mean so many questions for him but uh I like this idea of the front door to AI as the foundation models commoditize.
17:27
It becomes more about the being the front door to the internet.
17:33
Are you just laughing at your own mustache?
17:34
Are you laughing at your own mustache?
17:36
Laughing at your own jokes.
17:40
Normally you think I'd be laughing at you.
17:41
Every once in a while I look over at the while we're talking and I just crack up or uh Yeah, there there's other stuff that I'm just Well, I didn't really get it.
17:48
We went right into this like formal interview which Yeah. Yeah.
17:53
Uh normally we would have had a bit longer to laugh at ourselves. So anyways, sorry.
17:58
I think you look I think you look great.
18:00
In fact, I'm not laughing because it's so convincing. It looks natural. It's growing on me. It's just normal.
18:04
Anyway, um the the front door to AI like they are building a new Google where it's this front front door to knowledge engine stuff.
18:12
Uh but they need the front door to codegen as well.
18:17
And so this new paradigm of you know yeah and the interesting thing that runs the cloud valable I highly doubt Kevin can comment on this today but the immediate question I have is how does this integrate with a potential windsurf application.
18:31
This makes me think that windsurf might end up continuing to be a standalone Yep. app. Yep.
18:37
Um, and uh, so yeah, I mean it's really blurring the lines.
18:44
Like I didn't know that I wanted I didn't know that I needed code to decide how tall a desk is.
18:48
It decided that for me and I really like that as a product evolution where eventually you go to a you go you open the OpenAI app. Yeah.
18:58
And you ask for something and it decides do you do we need to do code for this?
19:04
Do we need to do an image model? Do we need to use 40?
19:05
Do we need to use three 03?
19:07
Do we need to use deep research?
19:09
Like it should decide for me.
19:11
That's a really cool one.
19:11
This is why I've always said I don't think OpenAI cares about naming because they're always in the long enough time horizon.
19:19
They're just going to do all the routing based on understanding, you know, what is required.
19:22
So anyway, on the flip side of AI, we have another post from DD Doss.
19:27
Uh Meta will delay its biggest AI model launch, Llama 4 Behemoth.
19:30
Uh four reasons highlighted.
19:33
Doesn't perform well internally.
19:35
huge reorg in AI leadership.
19:38
11 out of 14 researchers on Llama have left.
19:41
And we saw that funny post where someone was saying like, "I was on the Llama team.
19:44
I worked on Llama one, two, three, but not four."
19:46
And they put that on their resume on on LinkedIn. Brutal.
19:49
Uh all this after admitting that they game some of the benchmarks.
19:52
Uh glad glad meta can afford to light billions of dollars on fire for open source.
19:57
And so this is an interesting like discussion for me because I think that I don't think open source will win.
20:05
I think you need to build a product around it.
20:06
I But at the same time, when we talk to people like Aaron Gin, I think we it would be amazing if America had a rock solid open-source offering that was hardcoded, you know, or baked into the weights with American values and that was an option for uh for countries that are maybe deciding between American China. I want them on my team.
20:27
It's so hard to understand what Llama's real enterprise adoption looks like. Yeah. Right.
20:31
Yeah, we don't hear a lot about companies uh at least to date.
20:39
There was a big boom for a long time.
20:39
I think I I I think the answer lies in that open router data. Yeah.
20:43
Um and there are a lot of open source models running there. Kind of unclear.
20:47
Uh the Wall Street Journal has more deep dive here.
20:51
Uh Meta is contemplating significant management changes to its AI product group as a result.
20:55
Uh its performance has been hobbled by training challenges.
20:59
And so there is a take here that's like they believed almost too much in scaling laws, right?
21:04
Because Behemoth is this massive model.
21:06
I think it's a biggest context window, even bigger context window than what Google's offering.
21:12
It's a huge parameter model. Uh uh what is it?
21:14
Two billions of active 288 billion active parameters.
21:18
Two trillion total parameters.
21:22
We used to be in like Falcon 9B was like exciting. 9 billion parameters.
21:27
We're now up in 288 billion active parameters, two trillion parameters.
21:31
The circle has gotten so big from the GPT3 circle to the GPT4 circle.
21:36
We are now at the massive circle that's covering everything.
21:38
But we're not getting better results just from pure scaling because you have to add that RL layer on top, that post training.
21:44
And that's what uh OpenAI has been really really good at with the O models.
21:47
And that's what Deep Seek got right as well with their R1 model, the reasoning on top.
21:53
And it seems like they yeah I mean the main thing here is despite having effectively infinity resources they're struggling on the team side there isn't cohesion there there's a lot of churn there's former employees that that clearly were talented that are like going out and saying like you said you know I'm not I wasn't involved in this thing like feeling the need to sort of publicly state that they didn't play a role in that.
22:20
Y and so it doesn't matter how much money they have to spend um and what their capex looks like if there's not uh real cohesion. Yeah.
22:27
I've been thinking about this because the whole meme with pre-training was scale is all you need.
22:37
Just keep scaling the number of tokens and energy and parameters and and anything that go data that goes into the model, right? Just 10x everything.
22:47
10x that and then 10x it again and then 10x it again.
22:50
You're only a few 10xes away.
22:52
You're only a few 10x from from AGI.
22:54
And that was that was very clearly not the case.
22:56
It that the scale scale was not all you needed in a singular context.
23:01
The the uh exponential chart was secretly a sigmoid curve, right?
23:05
It would it would go exponential and then it would flatten out and see diminishing marginal returns.
23:10
And I think that that's true in life.
23:11
I think that's true in in so much technology development.
23:14
uh you need to be on the sigmoid grind set.
23:20
You need to be understanding that you're going to go through bursts of explosive exponential growth, but those exponential growth periods will stop.
23:25
I mean, we've seen this with the show, like when we first started quote tweeting, uh just printing out posts and quote tweeting.
23:32
We saw exponential growth in the follower count of the show and the listenership.
23:35
And then eventually we played that out and then we got to a plateau and then we said, "Let's do guests."
23:40
And then we saw another one and now we're going to come up with the third act, the fourth act, the fifth act.
23:45
You have to be constantly reinventing yourself.
23:47
And if you're in artificial intelligence, you can't just say, I'm going to rely on the pre-training scaling holding forever.
23:53
You need to go into okay, how do we scale up reinforcement learning?
23:57
How do we scale up tool use?
23:59
How do we scale up product use and product functionality to make this a really, really great product?
24:05
You're not going to get there purely on one single curve.
24:06
The arc of technology throughout history has always been a series of S-curves.
24:11
There's been the semiconductor boom, the internet boom, the mobile boom, the AI boom.
24:14
And if you want to make money or participate or understand the the the role of technology as it's evolved over the past 60 years or hundred years, you can't just see it as one linear exponential like one exponential graph.
24:28
It looks smooth when you zoom out.
24:30
When you zoom in, it's actually a bunch of S-curves. Yep.
24:33
There's the there's a boom where everyone's like mobile is going to the moon.
24:37
If the trend continues, there will be 10 trillion mobile devices and then what what happens?
24:42
Oh, it's like a couple billion because there's only a couple billion people, right? That's right.
24:46
Uh and that's just the that's just the way it goes.
24:48
Anyway, uh staying in mobile, let's go over to Tim Sweeney.
24:52
He says Apple's app review team should We have to we have to flag this flag. Okay. Which is just wild.
24:57
So, uh Z says, "Did they seriously name it Behemoth, the quintessential creature only God can tame from the book of Job?"
25:05
and then give it the demon branding.
25:07
And I gotta say that the the branding here is this a real photo that they used?
25:12
I think this is in the journal, too. Yeah, it is. What a crazy photo. It is a crazy asset.
25:18
I mean, you got to respect the development of the delts of the demon.
25:19
I mean, but it's still We were talking about that with with Orion, right?
25:26
Like like uh like the like the story of Orion, right?
25:29
It's actually the most crazy brand asset uh that you could choose for something like this.
25:38
It's like how do we how do we make this as scary as possible for the average Meta Platforms user? Yeah.
25:42
So, so uh so Facebook's Meta's latest virtual reality augmented reality headset, the glasses that everyone's raving about is called Orion.
25:52
And the Greek myth of Orion tells the story of a mighty hunter, sometimes portrayed as a giant, who is eventually killed and placed in the sky as a constellation.
26:01
It's like the worst possible metaphor you would want for a giant tech company trying to do something new and ambitious, right?
26:06
It's a very, very weird pick.
26:08
But he thinks about the Roman Empire all the time.
26:10
So maybe he's interpreting that myth differently and there's a different reading.
26:14
But I would love to know how they wound up with that.
26:16
But different versions of the myth uh detail his life and death, including how he was blinded, restored by his by his sight, his sight by the sun, and possibly killed by either a scorpion or Artemis, the goddess of the hunt due to jealousy or a trick by Apollo.
26:29
And so like if you read the metaphor of Orion, it's like it's not good for behemoth in Wikipedia is a beast from the biblical book of Job and is a form of the chaos monster created by God at the beginning of creation.
26:44
It's like a very weird choice.
26:44
You should actually you should just call it 67B2. That's fine. We like those names.
26:52
Let's just stick with those names. 40 is a great name. Orion is a great name.
26:56
Orion and behemoth sound cool.
26:56
Yeah, they sound cool until you peel back like the true metaphor is like bizarre.
27:02
Uh and so yeah, be behemoth has not only been a very large model, but it has been a very large problem for and it has been hard to tame and hard to get to produce uh fantastic results over there.
27:13
But good luck to the team over there.
27:16
Hopefully there's some new blood.
27:16
Some new folks come in and they produce a great, fantastic product for the Meta team.
27:21
Uh we're rooting for you.
27:21
Uh anyway, let's go over to Tim Sweeney.
27:25
Apple's app review team should be free to review all submitted apps promptly and accept or reject accordingly to according to the plain language of their guidelines.
27:34
App review shouldn't be weaponized by senior management as a tool to delay or obstruct competition, due process, or free speech.
27:39
And so then Naval said Apple continues to mock the court.
27:45
So this is of course Apple should need to approve Fortnite because Fortnite has a third-party checkout where you can go and buy Fortnite V-Bucks uh without paying the 30% app store fee.
27:56
Of course, you have to go off the app and there's a popup warning that comes up and Apple's being very aggressive about that.
28:02
But Apple is pushing the limits.
28:04
They are they are going one mile an hour under the speed limit.
28:08
They It's certainly not the It's certainly not the spirit of the law, but they are following the letter of the law.
28:13
At least they would argue that.
28:13
And who knows, maybe there'll be another court, another another battle in court.
28:18
But anyway, no, this feels like they're setting themselves up for another potentially separate lawsuit around the approval process.
28:25
Yeah, the approval process because Epic can can actually say this is costing us.
28:30
Yeah, this might have cost them already nine figures, right?
28:36
You can imagine when it goes back in the app store, the amount of the flood of demand and new revenue and so delaying that by even a you know a few days would be would be damaging.
28:43
So yeah, the Wall Street Journal says Epic Games Fortnite claims Apple block submission now unavailable on iOS.
28:51
The claim is the latest in a long-running feud between Epic Games and Apple. The legal began. So I agree with you.
28:57
There could be another case, but do you know how long it's been for the first case?
29:04
This is we're coming up on the 5year anniversary.
29:07
It started in August 13th of 2020.
29:10
Epic implemented a dra a direct payment system within Fortnite's iOS version circumventing Apple's 30% commission fee.
29:17
This action violated app Apple's App Store guidelines leading to Fortnite's removal from the App Store.
29:24
In response, Apple uh Epic fi filed an antitrust lawsuit against Apple in the United States District Court for the Northern District of California challenging Apple's restrictions on alternative inapp payment methods. It took five years.
29:36
And so, yes, they might have taken a couple pennies or millions of dollars out of Fortnite's pocketbook by delaying a few days, but if they can turn this into a fiveyear lawsuit, that's five years of of 30% fees, right? Yeah.
29:51
So that could be the calculus is like, yeah, let's just uh let's just let's just play the court game. It's fine. Crazy.
29:58
Anyway, uh other news, hostile, Novo Nordisk has uh their CEO has stepped down.
30:05
This is the maker of uh Ompic and the stock has been on an absolute tear.
30:11
I would love for you to pull it up on public.
30:12
Let me know what it's doing.
30:14
Um but we have an we have a post from an I'm an anonisk addict.
30:19
No, you're you're saying a tear downward, right?
30:21
Well, it was on a tear upwards for a long time during the GLP1 boom, but it's down 50% over the last year. O, yeah, not good.
30:27
So, yeah, may maybe make sense.
30:30
But, uh, a non-risk addict has a take here.
30:32
No idea if he has done a good job as CEO or if the change makes sense based on expectations going forward, but firing a CEO for highlighted reasons is obscene.
30:40
Novo traded at 50 XP, the market going crazy and then being a little less so, is not the CEO's fault.
30:48
So, if you're running a company and it becomes kind of a meme stock and it runs up to an insane multiple and then it pulls back a little bit, but you're still way above where you started as CEO, can you really be hit with, hey, the stock's down 50%.
31:03
We've seen this with Palunteer.
31:04
Palanteer was extremely hot.
31:04
And then, of course, it pulled back a little bit.
31:08
We see this with Tesla, too.
31:08
Tesla pulled back a little bit, but it's still like a really solid company.
31:12
really really incredible market caps and and great private uh great great price to equity uh ratios.
31:18
And so um the highlighted reasons here are during his eight-year tenure as CEO, Novo Nordisk's sales profits and share price have almost tripled.
31:27
Novo Nordisk has clear strategy, a strong portfol product portfolio, and an experienced leadership team.
31:32
The changes are however made in the light of recent market challenges Novo Nordisk has been facing and they have there are because this is an older technology.
31:39
As we talked to some of those biotech folks, um we've seen that uh there has been more competition than usual.
31:45
Normally, you create some breakthrough, you're able to lock it down for years.
31:48
That hasn't been the case in the GLP1 uh scenario.
31:50
hyper competitive from not only other scaled pharma companies but the you know the compounding and and yep considering the recent market challenges the share price decline and the wish from Novo Nordis Foundation the Novo Nordis board and Lars Foyard Fergard Jorgensson have jointly concluded that initiating a CEO succession is in the best interest of the company and its shareholders.
32:15
Yeah, very rare that you see a CEO uh go out while the stock is still trading at 50 XP, doing really well overall, even if it is down um 50% recently. I mean, that's not good.
32:26
But anyway, uh you know what Nova should do?
32:28
They should get a ramp, obviously.
32:29
Uh that should be the first thing that they do with the new CEO.
32:34
They should bring in ramp ramp. Time is money. Save both.
32:38
Easy to use corporate cards, bill payment, accounting, and a whole lot more all in one place.
32:42
They should also be using Figma.
32:44
We should talk about Figma.
32:44
Think faster, build faster.
32:46
Figma helps design and development teams build great products together.
32:51
You can get started for free. The best Figma is Figma.
32:53
For Figma, you can do a lot with it.
32:55
You can generate marketing assets, you can generate apps, you can create websites, and a whole lot more.
33:01
Yeah, thank you to Figma for supporting the show.
33:03
They should also get on Vanta.
33:05
Automate compliance, manage risk, improve trust continuously.
33:08
Vanta's trust management back to back.
33:10
takes the manual work out of your security compliance process and replaces it with continuous automation whether you're pursuing your first framework or managing a complex program.
33:18
Uh anyway, uh the other news is uh semi analysis is talking about the AI deals that have been happening in the Middle East.
33:25
Uh Dylan Patel coming on the show in a couple weeks.
33:27
Uh US strikes a deal with the United Arab Emirates and the Kingdom of Saudi Arabia. A 5 gawatt data center.
33:34
The humane G42 diversion uh and American AI wins.
33:38
Dylan Patel breaks it down.
33:40
He says the US Saudi out UAE AI deals are complete wins.
33:44
Everything is there to win over the Middle East and accelerate infrastructure spend.
33:48
So American companies, AI companies are going to be making more money from this because we have a new customer.
33:54
The main subtext here is that China is locked out of Middle East AI infrastructure investments.
33:59
There's a different world.
34:01
There's a different path where yes, there were some controversial moments in Trump's tour of the Middle East, but there is a different world where we are watching Cining do that tour and we're watching from the sidelines.
34:12
That's not what happened.
34:14
What happened is is our president went there and we would have struck deals for the first time on the show.
34:20
Instead of instead of Sam Alman and Elon Musk and Alex Wang, it could have been the founders of Huawei and Deepseek and uh Highfly.
34:26
Yeah, you can imag there's alternative timeline where you know the QIA is investing y you know hundred billion dollars not actually 100 but massive amount of money into Manis right or or Deepseek or some of these other players.
34:43
So uh Manus about Manis open source AI agents with Manus with Manis you're gonna make me put my mustache back on John. No, no, no.
34:54
We have a we have a very important person coming on the show.
34:57
Jordy, put that mustache down. Put that mustache down. Put that mustache down.
35:01
We have the chief operating officer of or chief product officer of OpenAI coming into the studio.
35:07
Um, and so, uh, the UAE ruling family used headline grabbing investment numbers, deals with Trump's family businesses, and ties with tech executives to score an NVIDIA deal.
35:15
And it's all looking good.
35:17
Um, let's see if we have Kevin in the studio. Let's bring him in. He's here. Fantastic. How you doing? This is wild.
35:25
I just like joined a Zoom and I'm live with you guys. We're live.
35:29
Yeah, it's great to have you. Welcome.
35:31
I was watching you on X and thinking how weird it was that in 30 seconds I was going to be on the screen. This is awesome.
35:37
Uh yeah, we're leveraging this thing.
35:39
Uh it's called the internet and technology.
35:41
Um but we we'd love to get you up to speed on it since uh it's funny a funny story.
35:45
Uh, so Tyler Cowan called into the show before one of your guys' earlier releases and he was like, "AGI's here."
35:55
Like he was basically saying like, "It's a couple days out, but he couldn't get his camera working."
35:59
And it was this like funny funny dichotomy of like AGI's here, but like the internet is still actually There was actually another day when we had to basically take the whole show down because Zoom and both both Zoom and Google Hangouts just went completely down like nationwide.
36:11
And so we were like, well, we can't do our show now, I guess.
36:18
The running joke is that audio video is like ASI complete. Yes.
36:23
Well, everything else in the world is solved. Yeah.
36:25
Wait, so I mean I think uh long term for the show, we need to go proprietary.
36:29
We need to build our own video streaming stack.
36:31
Can you can you help with that? Codeex. Codeex. You know what?
36:36
You know what can help with that?
36:37
Yeah, codeex can help with that. Break it down. How would I use codecs?
36:39
And by the way, you get to talk about AI and then you're going to have Blake from Boom here.
36:45
So, you get to talk about supersonic flight afterwards.
36:47
You guys have the coolest lives ever. It's full stack. Full stack technology.
36:52
Um, but yeah, c can can you can you take us through the announcement?
36:55
Uh, break down exactly what launched, when it's available, to who, what you're excited about, and then we'll go into some of the trade-offs in the product design and development. Yeah, let's do it.
37:05
So, we just launched this morning uh Codeex, which is a cloud-based software engineering agent that can work on many tasks in parallel.
37:14
So a lot of folks are used to the kind of cursor winds surf style, you know, GitHub copilot style of AI development, which is really more about augmenting a single engineer, right?
37:26
You're you're writing code in your IDE and you can press tab tab tab and it'll autocomplete and you're, you know, 10 20 40% faster. Yep.
37:32
With Codeex, you actually have a software agent that runs in the cloud that can do entire tasks for you.
37:40
So, you give it a task, it goes off and does it, and suddenly you have a PR and you didn't work on it at all.
37:45
Um, and it's powered by a version of 03 that we fine-tuned specifically to be really good at these kind of hard software tasks.
37:53
Uh, we're super excited about it.
37:56
It launched today inside chat GPT.
37:57
So, you can use it if you're it's rolling out now to pro, to enterprise, to teams.
38:02
Well, it'll make its way to plus in the coming weeks. Mh.
38:07
But the cool thing is it's a it's an agent.
38:09
It's a software agent in the cloud.
38:11
So, you know, you're using it today from chat GPT.
38:13
You can imagine using it in the future from, you know, your terminal, your IDE, all kinds of other places and even, you know, hooking it up via API to your bug cues and just having this software agent churn through every single bug that you have.
38:28
you know, for each one, look at the context of the bug, understand your codebase, and then suggest proactively how you would fix that bug and give you a PR that you can review.
38:40
So, the the world of software engineering is changing.
38:44
We're super excited about it. It's a research preview.
38:46
It's not perfect yet, but I think this is the future. Yeah.
38:50
So, uh I mean, I've already noticed that uh Chat GPT has been writing code for me for a while.
38:55
Uh it seems to be writing more and more code.
38:58
I was telling Jordy about how I wanted to know the height of a desk in an image and I knew how tall a person next to it was.
39:04
And I thought that it would just use uh images in chat GPT to kind of oneshot this or guess.
39:09
Uh but it wound up spinning up and looking at the individual pixels with a bunch of Python code.
39:16
I think it wound up writing like 5,000 lines of code or 500 lines of code and it got it really right.
39:20
Uh unsatisfactory because it was just an average size table.
39:24
It was like literally the standard size, but it really knew it.
39:27
Um but but so I'm wondering like like is this something that will in that will feel like a deep research project where or deep research functionality where I click a button to say hey let's use codeex for this I'm giving you the hint or is this something that can be automatically triggered just from a text interaction like images and chatbt yeah so it's a I mean by the way you know you never know if that desk was actually like 38. 2 two inches.
39:51
So yeah, it's worth it totally worth it.
39:54
Write the 500 lines of code. Yeah, why not?
39:58
You know, it's too cheap to meter.
40:00
Actually, that's been one of the big the coolest thing about 03.
40:02
03 personally for me has been a kind of feel the AGI sort of moment using that model, the things that it can do and it's a lot of it comes from the fact that it can use tools while it reasons. Yeah.
40:13
So it's thinking and in the process of thinking it can do some web searches and then it can take what it learned from a web search and write some code and then after that code it can do image analysis and then it can write some more code and do another web search and then finally with all of that context that will output the answer.
40:30
It's like it it's really been an unlock for a huge number of use cases and that's the kind of thing that enables codecs here because you're right chatbt has been able to write code for a while right you can you can just go to chat GBT and type in like here's a you know write me code to sort this array of integers that I have and it can give you the code and it's going to be great at it.
40:52
The difference here is codeex is built to work on big complex code bases.
40:57
So you're not just saying like do this little task for me, write this function.
41:01
You're saying I have a bug and I don't know where it is.
41:06
There's a, you know, 100,000line codebase.
41:08
Can you please go understand my codebase and and try and fix this bug?
41:14
Or I'm a new engineer at a new job.
41:17
I'm trying to understand what the heck is going on in this codebase.
41:18
Can you explain to me where the code does X or Y or Z?
41:22
And very quickly, it'll look through the code and give you an explanation of how something works.
41:26
Um, you know, it's funny.
41:28
I've even seen examples where people go, "Hey, codeex, find a bug in this codebase and fix it." Just find one randomly. Just go. That's amazing.
41:42
Complex stuff and do do hard work on a huge amount of existing context that differs from what you've been able to do in chatbt for a long time. Yeah. Um, wild.
41:52
From a personal perspective, I I used to write Python pretty regularly.
41:57
I haven't written much code, so I haven't really gotten into the cursor winds surf world.
42:00
Uh, obviously I've been writing code via chatbt now.
42:06
Uh, I noticed recently I kicked off a deep research report and it's and it prompted me to put it on effectively a cron job.
42:13
It was like, do you want me to just run this for you every week?
42:17
And I said, yeah, that sounds awesome. Uh, that's great.
42:18
Uh, but I if I'm like a I don't have a repo, but I could set one up.
42:24
Is there a world where me as kind of like a proumer non-technical user should set up a repo to house the custom code that Codeex writes for me to make a better experience for all the little custom software and tools and random stuff I use?
42:41
Or should I just live in the 03 world where the code is pretty much ephemeral?
42:48
I think it depends what you're looking to do.
42:50
You know, for simple things where you just want to like quickly put together a script or something, writing it inside chat GPT and not using version control and all of that is fine.
42:58
But if you're if you're going to do something, you know, that you expect to be longer lived, like it's the basis for something you actually want to build for yourself and maintain, then I think setting up a quick GitHub repo and using codecs on it makes a ton of sense.
43:12
By the way, I I so I used to be an engineer.
43:15
I haven't uh you know I still dabble and screw around on the side and write code but nothing major and I hadn't written any code at OpenAI.
43:23
I've been there about a year um and haven't checked in a thing. Yeah.
43:28
like Tuesday Tuesday Tuesday night I think.
43:30
Um I was I was doing you know the rest of my work and I was like I want to fix a couple bugs and so I went and found a couple really basic bugs uh because I didn't want to screw anything up and sent codeex off to work on both of them in parallel.
43:47
Check back in a few minutes and I had two PRs. They looked right.
43:52
So I submitted them, got them code reviewed by somebody, you know, who's actually a good engineer, and they were submitted, and now I've got a couple commits in the codebase.
43:59
So it's just it's like this is stolen valor, though. This is stolen valor.
44:03
You didn't write that code.
44:08
But it's like I I actually just kept, you know, other than just like getting to play around with the product and and offering a little bit of feedback on on a couple things, I was off doing the rest of my work.
44:19
And I had this software agent working for me in the cloud writing code. Yeah.
44:23
Talk about I I'm I'm so curious to hear about the kind of internal testing process and when when you guys decided the right time to actually roll this out as a research preview because I imagine you've been using codeex in one form of another internally for like a very long time.
44:42
maybe didn't have a name or anything like that, but um I'm sure that chat GPT has been, you know, contributing to the effectively the chat GPT codebase um you know, almost since the beginning in some form or another. Yeah, for sure.
44:59
I mean, we're we're big users of our own tools.
45:01
There was a there was a version for a while that um that was mostly about the kind of how do you come up to speed quickly in a big codebase that uh was good at at understanding our codebase and answering questions about it that was really popular with new engineers on the team and then once you kind of you know get to understand it you might not use that tool as often unless you're exploring a new area of the codebase.
45:25
So that was like a proto version of this. Yeah.
45:27
But we've been working a lot over the last six months at improving the ability of our models to code. Like you've got GPT 4.
45:36
1 which we released a little while ago which is kind of um which is very quickly become a a really popular model.
45:44
It's now I think default and wind surf.
45:46
uh it's increasingly uh a large percentage of of cursor users coding and you know that's that came from focusing on the things that matter in creating a really good coding model that you can rely on means really good instruction following longer context you know the
46:02
ability to like not just make the changes but to make the changes a way a developer would so don't add a bunch of extraneous stuff don't add weird comments you make surgical precise changes that accomplish the job um So there's a style element to writing good code, not just a correctness element. And we've been focusing on all of this.
46:20
And we've been focusing on all of this.
46:22
And then you you kind of bring that together with 03 and all of the the things that we were you know that 03's ability to tool call and to reason and um and and suddenly you can put together a really good coding model.
46:36
So we've been thinking about this for a long time.
46:40
This is the first time when we're like, okay, this is now good enough that we think it it deserves being a product for the rest of the world and we're excited to see how people use it.
46:49
Can you talk a little bit about product design, product like inspiration in product design?
46:56
I I noticed like the very first iOS app had these incredible haptics when the tokens were streaming through that I hadn't really seen anyone do.
47:06
I just opened the app last night and saw that when you're using voice mode to dictate to it, it has a different modal now.
47:14
It feels like there's a very strong design language evolving.
47:16
At the same time, there was, you know, people complaining, oh, I can't even I'm using Chacht.
47:21
I can't even stay logged in.
47:23
That that bug obviously got crushed pretty quickly.
47:25
Um, but but what what is the actual product design inspiration?
47:30
Are there people that are pulling from certain uh schools of thought or anything or like is there someone driving that internally or is it just like baked into the culture? Yeah, for sure.
47:40
Ian Silber leads our design.
47:40
I was fortunate to work with him at Instagram.
47:44
He's an incredible designer.
47:44
It's building for for Chat GPT is a really interesting thing because we have, you know, well over 500 million weekly active users at this point.
47:54
So, it's a it's a big scaled product. Yeah.
47:59
And so for products of that size, you one of the most important things is to simplify, right?
48:04
You're not just serving power users at that point.
48:06
And so you're serving people that are just trying to get something done in their day.
48:11
They don't want they don't care about the complexity.
48:13
They don't care what the models are called.
48:15
They just have a task and they want to complete it and you want to help them.
48:18
But then on the other hand, we have folks who are super deep AI enthusiasts and want to, you know, digest and every single new model that we use and try it out in all these different ways.
48:30
And so we want to both and we want them to we don't want to sort of uh soften the edges of their experience.
48:37
We want them to to be able to do everything they possibly can to experience all of you know the power of AI.
48:44
And so we both want to like simplify the experience for for a lot of our users and we want to provide the people that want it all of the bells and whistles.
48:53
And so we try and we try and balance that.
48:56
Um you know so we we try and make it so that you don't need to worry about things like the model picker as much.
49:03
You don't need to like have a bunch of AI knowledge in the background to do what you want to do in OpenAI or in chat GBT.
49:09
But if you have that, you should be able to expose the sharp edges and like test the different new features and stuff.
49:15
And so we we really actually try and get both of those things right.
49:19
And it's a delicate balance.
49:19
How do you is that why is that why you guys don't seem to put too much emphasis into perfectly naming products just because in the on a long enough time horizon it doesn't really matter.
49:31
I just come to, you know, chatbt and I work with it to get the outputs and the results that I want and I'm not regardless of my experience level, I don't necessarily care the underlying sort of models doing the actual work.
49:45
Wait, are you saying are you naming isn't great?
49:53
I was I was alluding to No, to be clear, we think that you have the second worst naming after Behemoth, which is a very very untameable, you know, demon demon monster, potentially a disaster.
50:04
So, we are now long 40 and 03 and we just I just, you know, if you look only six months ago, a new model would get announced and people are like, "Oh, it's so confusing, blah, blah, blah."
50:19
But it just it seemed to me, you know, as an observer that you guys like it wasn't like, oh, this is a problem and we need to fix it.
50:27
It was just more so like let's just keep making really great models and make them easier to access in really intuitive ways. Yeah.
50:33
I mean, in all in all seriousness, it comes from our focus.
50:38
We we have this principle of iterative deployment that we really believe in, which is that these are new these models are new systems, right?
50:46
Each one, each model has capabilities that we understand somewhat and and also we discover new things about it.
50:52
And we believe that no matter how many smart people we have inside of our walls, there are way more smart people outside our walls.
51:00
And the best thing we can do in a world of AI evolving so quickly is to kind of co-evolve with society to ship stuff early and ship often and, you know, learn together.
51:11
And so one of the reasons that we have this explosion of models is we're trying to build new capabilities rapidly.
51:18
And sometimes the easiest way to do that is to kind of build it into a new model that's really good at one specific thing or a handful of specific things but can't do everything.
51:29
And so you end up with this like profusion of models that do different things well. Like 4.
51:33
1 is really good at coding and instruction following but it's like not as chatty.
51:41
And so if you're asking about other things, you might prefer 40 for some things and 4. 1 for others. Seems totally natural.
51:46
You'd go like, well, why don't you just build one that's, you know, good at coding when you're coding and good at chatting when you're chatting. And we will do that.
51:53
That's what we're trying to get to with GPT5, where we're trying to bring more of these things together.
51:58
But if we tried to do that from the beginning, we wouldn't have been able to launch as fast.
52:02
And so we've we've opted for like launching fast having a little bit of you know confusion that comes with it but we learn faster and then over time you sort of integrate and simplify.
52:12
I guess like the the meta question though is like is the future just you're already using mixture of experts within the models.
52:19
Is the future like a mixture of mixture of experts models?
52:22
And so I I I go to one there's one command line text is the universal interface not drop down model pickers and and and it routes me it says hey this person doesn't want to chat they want to write code. Okay we're using 401.
52:35
Uh yeah, that seems kind of logical and already this is kind of happening with the model picker becoming like just the UI is getting less and less in your face and it's a little bit easier just to have a natural uh uh interaction.
52:48
But how do you see it evolving?
52:49
Yeah, I think over time the the capabilities that that have existed for a little while, you sort of learn how to bring them into a general model. Yeah.
52:57
And then but you're always going to have these new frontier capabilities. Yeah.
53:02
that you're going to want to be able to iterate on really quickly and you might want to do specialized things to to make the model really great at some new frontier capability.
53:09
And so I think yes, ideally you have a you have a a sort of model, you know, a layer above the models that's doing the choosing for you. Yeah.
53:20
But that is, you know, it's a hard problem, especially it goes back to the the building for simplicity versus building to enable power users. Sure.
53:26
As a power user, you might be the only one that knows for a particular question you're asking.
53:33
Whether you want a 80% good answer immediately or you'll wait a minute for a 95% good answer or whether you want to do deep research and wait 20 minutes and get an amazing answer. Yeah.
53:45
I mean, you could theoretically like train the user on that a little bit like like it's also like how you work with the comp that I or like my personal framework is like when you're working with people on your team or teammates, there's certain people you'd work with that you would have to explain effectively like the exact tool set that they should use to accomplish the task.
54:06
You should get the CRM and you should go in the CRM and do this and that.
54:07
And then there's people that are maybe have greater intelligence or experience or context and you just sort of like discuss the task with them and you're not even thinking about the underlying sort of like toolkit to accomplish the task.
54:23
It's just sort of this higher level yeah you know conversation.
54:24
I I I want to talk about AB testing versus personalization.
54:29
Um when you choose like a default model or the default prompts when you open up chat GBT it says create an image, write a Python script, make up a story. what's in the news.
54:39
Um there's a couple options there.
54:41
Uh there's a lot of personalization going on, but you could also imagine doing AB testing to understand what will drive turn down or retention up.
54:50
Uh how are you thinking about the balancing act between those two techniques of product development?
54:56
Uh I don't think they're really in at odds in any way. We do both. Yeah.
54:58
So, we we definitely AB test a lot of things uh because we're trying to learn what works and and you know how how we can help people understand this new kind of strange world of AI.
55:11
It's a funny it's a funny product, right?
55:14
You're we're used to products where uh you you have a UI like computers before AI needed very specific inputs like this button does this specific thing and that button does this other thing.
55:27
And if you wanted to do a third thing and there wasn't a button for it, you probably just couldn't do that thing, right?
55:31
But then every time you hit the button, you got the same output. It was very consistent. Yep.
55:36
LLMs are basically the opposite, right?
55:39
You can give them input that is the that has the full complexity and nuance of the human language and you have no limits on what you ask.
55:48
And then also what you get out is not the same from one thing to the next.
55:53
They might be substantially the same, but the words are not identical, right?
55:57
And so it's just a it's a totally different way of building product.
55:58
And when someone comes to chat GPT for the first time, if they just hear from their friends, hey, this this AI thing is super cool.
56:06
It can do all this stuff for me.
56:07
And they show up at the front door of chat GPT, they're a new user, like what's the mental model?
56:13
Because it it flies in the face of almost every thing that you've learned using computers over the last however long.
56:18
So we we really think a lot about how we get people going uh and how we teach them all of the different capabilities which you know by the way the capabilities are changing every month or two too.
56:30
So it's a it's a really challenging problem um but something that we care a lot about because that's you know if you go from uh being a novice chat GPT user to being a power user it can really change your life.
56:46
It can save you a ton of time.
56:46
can accomplish a lot of tasks for you and that's only increasing.
56:49
So the the upside of us being able to teach people well is also increasing. Yeah.
56:53
I feel like a decade from now people are going to look back at this moment and realize that the people that fully understood the capabil like the full capability set of the models just had this ridiculous sort of extreme advantage.
57:10
It was the same thing with social media.
57:12
like the people that really understood and took social media seriously early on are like famous now, like actually famous. Uh interesting.
57:20
Uh what what is your post-mortem on the sycopency thing?
57:22
I feel like that like that made news because it was kind of blanketly a uh uh like it was kind of like everyone was experiencing or at least all the power users were experiencing it.
57:33
Um, but I could imagine a situation where some some people really like that type of interaction and it was beneficial and it made improved their lives.
57:44
And so, uh, if you go to the YouTube algorithm right now and you only search and click on positive content that reassures you, you can have a sick authentic experience and that can be good for everyone involved.
57:58
Uh so how do you what is your postmortem on it and how do you think that uh the personalization uh will play out in the future?
58:06
Yeah, it's a this was a really important issue.
58:09
I mean we so we the the story for people who don't know we we rolled out a new version of GPT40 which is something we do pretty regularly.
58:16
There's we're always you know to your point about AB testing we're testing new versions of GPT40 that are incremental improvements over previous versions.
58:25
So we rolled out a new one and you know we had AB tested in the past.
58:31
So we um you know the metrics looked good.
58:34
that it looked like a really solid model, had some new stuff around personalization.
58:37
And then as it got out there, we saw that a number of of use cases, not like super widespread, but enough use cases where we saw the model sort of um overly like like just being some of it was like glazing, what people call glazing. Yeah, we called it that.
58:59
I think we used that term. Yeah.
59:03
But then but then there were other cases that were more uh more serious totally where someone had real problems and you know maybe they they were having mental issues and the model was sort of validating them in ways that didn't really comport with reality and that's like a that's a real thing and we took that super seriously.
59:22
So, we rolled the model back and then um basically have spent the last few weeks diving into uh where this is coming from and what we need to do to um to to make sure it doesn't happen again.
59:35
And we've tried to be super transparent about it.
59:37
So, you know, we tweeted as soon as we were rolling it back and then immediately put out uh a postmortem like, you know, within a day or so and then put out a second after we had done a bunch of deep dives.
59:50
And so we've gone through, we've like the the team uh did some great work.
59:55
We've got evals now that measure this.
59:57
Um we understand a bunch of the root causes from where this came from.
1:00:02
You know, as always with these things, they're not it's never just one thing.
1:00:06
It's like a little bit of this combined with a little bit of that and then this unexpected thing happened and together they created something that that you know wasn't up to the standards that we set for ourselves. Yeah. So uh I Yeah, sorry.
1:00:19
Uh I I I just have an interesting uh realization with the product.
1:00:22
So uh we've been hearing this this like this request for feature on social media for a while of like I wish I could just reset my algorithm, start fresh because I feel like it's funneled me in some sort of echo chamber and I don't like that echo chamber and I want to start fresh.
1:00:39
Uh and I don't know if social media feeds actually have that feature. It might just be buried.
1:00:43
But I've noticed that with the chat memories uh early on I was really aggressive about prompt engineering and and basically like prompt hacking.
1:00:51
And so to get the best responses if I was trying to learn about trains I would say like I am a world expert in trains.
1:00:57
I own multiple train lines and railroads.
1:01:00
Give me a breakdown of the market map of trains. Basically lying to it.
1:01:05
Uh and then it remembered that.
1:01:07
And so now it's like well as a train conductor you'll probably want to eat this for dinner.
1:01:13
And I'm like, okay, I I h I have to back up.
1:01:15
I wasn't being completely honest with you, chatbt.
1:01:17
But the good news is that the saved memories are there and I can delete them all and so I can kind of reset my experience.
1:01:23
Was but was that a was that a was that a learning from the demand that people are seeing and the and the uh re and the not the stated preference on social media for resetting or do you think that that's important or what other lessons are you learning from from how social media has played out because you obviously have a lot of experience there and a lot of people at the team have experience in social media.
1:01:45
Yeah, it's personalization is a is a really powerful thing that I think we're just at the very beginning of like you want I mean in the same way that we we know each other a little bit. Yeah.
1:01:57
You have you have best friends that you know super well who you're really comfortable with and then you have strangers and you're you know your level of comfort in interacting with them is very different. Yeah.
1:02:06
You want your if you have a super assistant in your life in chat GBT you want it to know you really well.
1:02:12
really well. you wanted to know your habits and how you like to do certain things and you know even down to like do you want uh you know more flowery supportive language or do you want crisp analytical tur language things like that um I've I was uh messing around last night with chat GBT trying to do um
1:02:32
trying to give my son some math homework and uh he I just said hey can you design 10 math problems for Matthew and the model knew knew Matthew was my son, knew he was 10 years old and developed a bunch of like grade level appropriate escalating, you know, and it was like that was super cool. And it even it was
1:02:51
And it even it was like, oh, he likes Legos and Kevin, you're a runner and so a bunch of these program a bunch of the questions were like Lego themed and you're going on a run with your dad and this and so like, okay, this is really cool.
1:03:04
It's truly making me emotional.
1:03:06
It's like the coolest thing as like a parent to be able as a parent to be able to give your to give your children like a truly bespoke magical experience is like almost priceless even in the context of home like something like homework, right?
1:03:23
And so to that and and even knowing that like kids today like both of our all of you know we we've got five kids between the two of us and um knowing that they'll grow up only knowing the sort of world in which this kind of technology exists where a parent can just like generate magic for them in like seconds is just unbelievable.
1:03:45
I mean, and and think of the world.
1:03:48
This thing designed 10 math problems. So, I used 03.
1:03:54
It designed 10 math problems for my 10-year-old that were escalating in difficulty across a range of different things.
1:04:01
It could easily, if he was entering in the answers, could easily realize over time where, you know, what what concepts he understood and what concepts he didn't and just become a personalized tutor for every single kid.
1:04:15
And remember, I mean, this is free, right?
1:04:18
We don't charge for Chat GBT.
1:04:19
You can you can get a subscription, but you can also use it for free basically.
1:04:22
You you don't need you don't even need an account.
1:04:24
You need an Android phone anywhere in the world for free and you can, you know, get this thing that's starting to be more and more of a personalized tutor.
1:04:32
I just like I think it's incredibly powerful. Totally.
1:04:36
Every every study I've ever seen says that when you pair, you know, traditional learning with personalized tutoring, the it's like a standard deviations of improvement. So, I I I'm with you.
1:04:51
I think the future is going to be very different and there's a lot of reasons to be optimistic about what the next generation is going to be able to do with AI.
1:04:57
I need to be more honest with Jess.
1:05:00
How do you how do you explain the rate of AI progress to let's say like a family member of yours that's not in tech? That's a good question.
1:05:13
Um I think the only way honestly because you can talk about it but it's like you can't you can't get fit by reading about going to the gym. Like you just use it.
1:05:25
Um, and so I've been trying to get any of my family members who aren't using it, just just try it.
1:05:29
Start ask, you know, for everything that you're doing, ask why couldn't I use chat GPT for this?
1:05:36
And you start to realize that there are more things that you can say yes to there and you can start using chat GBT and then, you know, the more you use it, the more you realize the value and and off you go.
1:05:45
It's really hard to explain in the abstract, right?
1:05:49
People go, "Ah, agents, I'm hearing so much about agents. What is it?"
1:05:52
And then you use deep research or you use codeex and you're like, "Oh, wow.
1:05:56
That just saved me a ton of time or did something I couldn't have even done." Yeah. Yeah.
1:06:02
You know, that's the promise of the future.
1:06:04
I think we're all going to be much more productive.
1:06:05
We get to not focus as much on the doing of particular things.
1:06:10
We get to focus on the outcomes and, you know, what we do once the once some of the labor itself is is taken care of.
1:06:16
And that's that's a super exciting future for me. Couple more.
1:06:18
Uh, talk to us about healthbench. Yeah.
1:06:25
So, uh, a huge amount I mean people are increasingly using chat GPT for health.
1:06:30
I've done it any number of times.
1:06:32
My son had a had a small surgery that was supposed to be, you know, 99. 9% innocuous, 0. 1% bad.
1:06:38
And we got the results back from the doctor before I could talk to the doctor. And they look scary.
1:06:48
look scary. uh and uh it it's full of a bunch of medical jargon that even as a you know former scientist I didn't understand and I put it in chat GBT and said this looks weird what is this is this should I be worried and it said oh no no no don't don't worry it's fine and
1:07:04
I was like okay explain it like I'm five and it did and you know I couldn't get a hold of the doctor for another 72 hours that would have been a bad 72 hours for me if I was sitting there like stressed out about my son and ChatGBT gave me the piece of mind. So we always try and say
1:07:19
So we always try and say with anybody anybody asks about anything medical, you know, this isn't a substitute for actually seeing a doctor.
1:07:27
JBT is not a doctor, but here it is.
1:07:30
Here's here's here's the understanding of what's going on there.
1:07:31
And it's really valuable.
1:07:34
You know, for all of us, you know, for me, it saved me 72 hours of of anxiousness.
1:07:38
For somebody else who doesn't have access to a doctor, it might be a totally different thing.
1:07:43
So, um, anyways, we care a lot about this.
1:07:47
We want to make if people are using chat GBT for this, we want to make sure that the answers that it gives are really good answers.
1:07:53
And so we're putting a lot of effort into improving uh ChatGpt's ability to act as or to to answer medical questions.
1:08:02
And the only way that you really know if you're doing it right is if you have a benchmark, right?
1:08:07
You've got to have something to test against to show that you're getting better.
1:08:10
and we figured if we done a lot of work to put this benchmark together then you know others could benefit from it outside of us and so we we we open sourced it.
1:08:20
Last question can you give us the 30 seconds on why you were excited to join Cisco's board of directors. Oh yeah totally.
1:08:27
Um so I mean Cisco is an is an incredible company like iconic Silicon Valley software and hardware company.
1:08:36
uh we all use it every day in in a hundred different ways and they're at this really interesting point because I think AI can transform their business and they can either be transformed like it can either happen to them or they can get ahead of it and build some really amazing software and tools and become you know a sort of an even powerful leader for the next in the next generation and I think that's not unique to Cisco.
1:09:07
I think a lot of companies are at that turning point, but Cisco really realizes it.
1:09:09
By the way, they're actually a launch partner today for us with codeex.
1:09:12
So, they're one of our early partners.
1:09:16
They're looking a lot at how AI can can help them get more done, you know, faster, more cheaply, etc.
1:09:21
So, it's just AI is going to really impact their business over the next uh you know, 3 to 5 years.
1:09:28
And I'm excited to be a part of that and hopefully help them navigate uh this transition gracefully. Amazing.
1:09:35
We'll have like five more hours of questions, but we will let you go. You go.
1:09:40
We'll come up to most questions can be answered by chat GPT, but there's certain questions you got to go to the source organic farm farm to table. Yeah. Yeah.
1:09:49
Well, thanks so much for having me on.
1:09:51
It was good to see you guys and talking about supersonic jets that I wish I could stay for that one. Yeah. Yeah. We'll talk to you soon. Great to see you. Cheers. See you. Bye. Um, we got is the man.
1:10:01
Uh but you know what else is the man? Linear is the man.
1:10:07
Linear is a purpose-built tool for planning and building products.
1:10:09
Meet the system for modern software development.
1:10:13
Streamline issues, projects, and product road maps. Linear.
1:10:15
Uh and also numeral is the man. Sales tax on autopilot.
1:10:20
Spend less than five minutes per month on sales tax compliance.
1:10:22
Put your sales tax on autopilot.
1:10:24
There was numeral was getting picked up by an anal.
1:10:28
Oh yeah, we saw the news.
1:10:30
There may be out there in the market, but Sam said too low. Too low.
1:10:35
You're not bullish enough on sales tax.
1:10:38
You tell him it's on record or not. Well, it's live.
1:10:40
He posted No, he No, he posted this. He just said too low.
1:10:43
Okay, so we're not we're not scooping anyone here.
1:10:46
We're not scooping anybody.
1:10:46
Uh we have Blake from Boom Supersonic in the studio.
1:10:51
Welcome to the show, Blake. It's been too long.
1:10:55
We've been wanting to do this for a while.
1:10:56
Yeah, ideally from a supersonic plane.
1:11:00
We're we're we're aviation afficionados. We like traveling.
1:11:02
Uh but the hard thing with the show is that it takes us six hours to get across the country.
1:11:10
We podcast for three hours a day.
1:11:12
The math just doesn't make sense.
1:11:14
So, we're happy that you're working on a solution for us. Indeed.
1:11:15
Well, thank you for Thank you for having me.
1:11:17
Yeah, it will be fun to do the first ever Supersonic webcast.
1:11:21
Maybe we should just agree to do that. I'm 100% on board. Let's do it. count me in.
1:11:25
Uh can you give us just the general update?
1:11:29
Obviously explain like what you're building, but what's uh what's the latest news in the world of Boom Supersonic? Yeah.
1:11:34
Well, I think for anybody who's not been following the story, you the goal is to really pick up where Concord left off, fermented supersonic renaissance, ultimately deliver uh faser travel for everybody from the president on down to every family.
1:11:48
And so it's a it's a decadesl long uh mission, you know, ultimately to replace subsonic with supersonic for every passenger on every route.
1:11:56
Uh but uh but building building companies like Boom is it's like building an iceberg from the bottom up.
1:12:03
And I feel like this year is the year that the iceberg is like really emerged from the the surface of the water.
1:12:09
In January, we broke the sound barrier uh with our test airplane, the XV1.
1:12:16
Breaking the sound barrier.
1:12:16
We love to see it, right?
1:12:18
And then and then in February we did it again and arguably we broke it permanently. Double kill. Fantastic. That's great news. I don't know.
1:12:27
You said you broke it permanently. Is that what you said? Yes.
1:12:29
Because we proved we could do it reliably with no audible sonic boom. Oh, there you go. Okay.
1:12:36
They said it couldn't be done.
1:12:38
They said it couldn't be done. Yeah.
1:12:40
People People, by the way, it turns out that like people have been telling me this for decades and mostly the claims are it's like really hard and you have to like change the aerodynamics of the airplane and D it. No, it's a software fix. Really? How? Well, explain that.
1:12:55
Everything is computer, John. That's it. Everything is computer.
1:12:57
If you fly the airplane at the right altitude, at the right speed for the current atmosphere, the boom makes a Uturn in the sky and never touches the ground. No way.
1:13:08
But you need to be able to calculate that in real time.
1:13:12
You need like decently good weather data and and and ironically algorithms that were developed for computer gaming. Wow. Oh, that makes sense. Yeah.
1:13:20
So, it's a it's a physics simulation.
1:13:22
It's it's it's ray tracing.
1:13:25
You're running Unreal Engine almost.
1:13:25
I'm sure you have proprietary system.
1:13:28
Um what uh does that need to go through FAA approval to use that or is that separate from the rest of the aircraft uh authentic what's it called qualification or like approval? Yeah.
1:13:41
So you need to certify an airplane carry passengers.
1:13:45
You know basically you got to prove you meet all the safety standards. Yeah.
1:13:46
Uh and this is this is separate because we we have um one of the dumbest regulations ever created.
1:13:53
In 1973 we banned supersonic flight in the US.
1:13:56
Like literally there's a regulation that says thou shalt not exceed Mach 1. Yeah.
1:14:00
And it's speed limit, not sound limit. Right. Right. It's really stupid. It's really stupid.
1:14:06
And um uh and so no, you know, we could like make it play Mozart when it flies over and you're still not allowed to do it.
1:14:13
And that's why that's why those coast to coast flights are still stuck at six hours.
1:14:16
That's why no one's done this already.
1:14:18
And uh and so we we're working really hard to get that uh changed.
1:14:24
We had it feels like something that like the EO would just drop and it would be like, "Oh my god, he just tweeted it out or put it on Truth Social and it happened."
1:14:31
Like, love him or hate him, but like the guy definitely like likes to rip a crazy idea on short notice for everyone.
1:14:36
Uh, are you optimistic that there will be change?
1:14:40
Does this need to go through the House and Senate or is this something that could just happen in EO?
1:14:44
So, we've had good conversations at FAA uh this week.
1:14:49
a bipartisan bill dropped in the House and the Senate. Uh that does it. Cool. And uh Elon endorsed it.
1:14:55
Uh Jared Isaacman endorsed it.
1:14:58
Uh we've got bipartisan support. Great.
1:15:00
Uh some people think this could get through the Senate unanimously.
1:15:04
Uh I think that's probably a little ambitious.
1:15:07
But uh but even even the idea that that's possible, I think is pretty cool. Yeah.
1:15:11
Some lawmakers like traveling slower because it inspires them to grind harder.
1:15:18
more time on the plane, more time to prep their filibusters.
1:15:20
They don't want to go fast.
1:15:22
I guess, you know, the bare case for supersonic I get so much email done.
1:15:26
I do get a lot of email done on a six hour flight.
1:15:30
If I got there faster, I'd do less email.
1:15:32
I was texting Jordy when we were flying to DC.
1:15:33
I was like, "This is incredible.
1:15:35
I've never been more productive."
1:15:36
Uh, but I think faster flights will just force us all to be more efficient with our time. Right. That's right. Yeah.
1:15:42
Uh but but I mean uh break down the actual uh scope of the problem here. How much is engineering? How much is regulatory?
1:15:49
How are you staffing against that?
1:15:52
Do you have a massive government affairs team and lobbyists or I imagine there's still a ton of engineers at the on the you're not just taking some like white labeled plane and then doing a bunch of lobbying and that's the endgame, right? Yeah.
1:16:06
I mean, the most surprising thing is that we don't have to invent anything fundamentally new.
1:16:12
This is not a science project.
1:16:14
It's not a technology project.
1:16:16
It's really just an engineering project.
1:16:18
In fact, we're taking 20-year-old 787 technology, basically reshaping the airplane, making it long and skinny, putting twice as many engines.
1:16:25
And so, there's a lot of engineering and testing that goes into that.
1:16:28
But there's no science and there's no new technology.
1:16:32
And there's only one regulation that needs to change. Mhm.
1:16:33
Um, so, uh, so we've got small numbers of great engineers.
1:16:38
I'm a big believer in like tiny teams.
1:16:40
Uh, that are very focused.
1:16:43
Like we did, we we built XB1, our test supersonic jet, which is 50 people. 50 people. Wow.
1:16:48
And the overall company's not that much bigger. Right. Right. We're like 115 now.
1:16:51
And like we're growing very slowly.
1:16:53
Like if a team does not complain about being understaffed, I know they're overstaffed. Yeah. Yeah.
1:16:57
Uh I mean we we were joking about like white labeling a plane.
1:17:02
Obviously you are standing on the shoulders of giants using some uh off-the-shelf technology using some technology licensed from other firms.
1:17:11
Can you take me through the journey of engine development the decisions that you made?
1:17:15
Have you changed course on any of those decisions?
1:17:18
Uh what would you recommend to the next generation of aircraft builders? Yeah.
1:17:21
Um well I'll go broader than that.
1:17:24
Like if if you're thinking about doing hardware at all, my my my advice, frankly, any startup, any startup, all startups are hard.
1:17:30
I don't think any are easy.
1:17:32
F I think the difficulty level is set by founders because we tend to run at our own red line.
1:17:37
And if somehow it gets easier, we'll just make the job harder again like like like Bri, you like Brian just decided to like, you know, double what Airbnb is, you know, because I guess the old thing got too easy. Yeah.
1:17:46
Um and and so and so what I what I've found along the way is I'm far more successful if I pick a mission that deeply inspires me and makes it worth being at my red line.
1:17:58
So I never get up in the morning and think is it worth it? Okay, great.
1:18:00
So now now probably building supersonic jets is like you know the ultimate hard mode.
1:18:05
Um and but what I have found along the way is uh people around the company from the legacy industry will have all these stories about all these things that are impossible to do.
1:18:15
you know, you can't build your own jet engine.
1:18:17
Like, only a big company can do that.
1:18:18
There's tons of proprietary technology, blah blah blah blah blah blah.
1:18:22
It's all [ __ ] It's all [ __ ] Like, I'll tell you one story.
1:18:26
Like, we were trying to get um high temperature super alloys for our engine and you know, we thought we needed to go license it from one of the big three.
1:18:33
We get into this licensing conversation and they're like, "Oh, we can't give you this thing because there's a trade secret."
1:18:38
I'm like, "Don't tell me the trade secret. I'll give you the part.
1:18:42
just make me the part and hand me the parts back and don't tell me the secret.
1:18:45
And they said, "No, no, no, no. You'll find the secret." So, we can't do that.
1:18:48
And for for like two weeks, we're like, "Oh, shit."
1:18:52
Like, "How do we ever get this engine built?"
1:18:53
But eventually, we just went into the supply chain and we found the trade secret. We found it.
1:18:59
And I'll tell you the trade secret. Okay.
1:19:02
There is no trade secret. Oh, interesting.
1:19:05
The thing that was theoretically proprietary is an open- source material developed by NASA where all the specs are public.
1:19:11
Well, the trade secret is that they're using open source.
1:19:14
That was the trade secret. That's the trade secret.
1:19:16
The trade secret was there is no trade secret.
1:19:18
Trade secret because there's all this fake proprietary and you know everybody's telling you why you have to work with them, why you have to use their stuff, why you couldn't create it yourself.
1:19:26
And you know, and it just nine times out of 10 it's just not true.
1:19:29
And so we we we found you know we've probably heard Elon talk about the idiot index which is like how much a finished part cost uh is divided by the raw materials cost.
1:19:39
We found a thing that's more important.
1:19:41
It's called the slacker index.
1:19:44
Slacker index is how long it takes to get something divided by how long it takes to actually make it. Mhm. Wow.
1:19:48
And and so we've got these turbine blades.
1:19:52
We're 3D printing turbine blades for our engine.
1:19:53
And we go quote it out of the traditional aerospace supply chain.
1:19:56
It's going to cost a million dollars for one engine's worth of parts and it's going to take six months.
1:20:01
I was like, well, how long does it take to print a blade?
1:20:05
Well, it's like actually about 24 hours.
1:20:07
So, why does it take six months to get one?
1:20:08
Well, they were like printing them like one at a time and then you got to wait for your turn on the machine. All this not okay. What's the machine cost? $2 million.
1:20:16
How long does it take to get a machine?
1:20:18
Oh, actually they've got them in inventory.
1:20:19
You get them in a couple weeks.
1:20:21
So, so for the price of two engines worth of blades, we got the 3D printers and the blades and we can and and we beat the lead time of just outsourcing it and that and that pattern exists everywhere in this business.
1:20:33
Uh, can you talk about the current fundraising market for hard tech companies?
1:20:39
There's a bunch of companies raising a ton of money.
1:20:40
There's other, it's feast and famine out there.
1:20:42
Uh, yeah, the F-35 cost a trillion dollars somehow.
1:20:47
I feel like trillion dollar raise is on the table soon based on the current market.
1:20:53
Uh, yeah, we'd like we'd personally like to see you do one T on 5T. I'd love to see that.
1:20:57
I'd love to see that on 5T.
1:20:59
That should be the new the new goal.
1:21:01
But but I mean there is there is a world where a plane company comes out and raises just massive money and just dumps it all into subcontractors and takes a very different approach. Why wouldn't that work?
1:21:12
Uh and and and what advice would you give to the next generation of hard techch founders in aviation or otherwise? Yeah.
1:21:19
So the the key difference so there's this mythology that hardware companies are more capital intensive and if you go look at like how much money did Uber raise, how much did Lyft raise, how much did Stripe raise, how much did Airbnb raise, like these like theoretically capital-like businesses consumed billions in venture capital. Yep. Before IPO. So wait what?
1:21:39
And if you go look at like SpaceX or Andril like the you know the theoretically hard techch companies like oftentimes they raise less money. Mhm.
1:21:46
So, you know, WTF and I I think that the difference is uh if you are building say Uber or Airbnb, you have an idea that sounds really counterintuitive like invite strangers from the internet to sleep on your couch. Mhm. Right.
1:21:59
It doesn't sound like a good idea, but it's actually really cheap to test. Yeah.
1:22:03
And um and and so what happens in most internet businesses is for very small amounts of capital allow you to test whether you have product market fit and and you and you test it by means of building and shipping the product.
1:22:16
Uh we can't build a supersonic airliner as a means to testing whether anybody wants a supersonic airliner, right?
1:22:25
Uh so so what you have to do in a hardware business is is find a capital light way to demonstrate that you're a product market fit.
1:22:32
And the answer can't be shipping the product.
1:22:36
So, so, so you know, our strategy was pre-orders. Yeah.
1:22:38
Um, so we, you know, we got United American to make like deposits, like non-refundable deposits on airplanes against a specific design.
1:22:47
So, it's not just like, okay, here's, you know, here's a dollar that says, you know, supersonic is cool.
1:22:51
It's like, nope, here's a significant multi-million dollar deposit, a whole further deposit schedule against a very specific airplane with very specific specs.
1:22:59
And and then so we can go to investors and say, um, if we build this, there's obviously a gigantic market and this is going to be like a getting paid for it.
1:23:12
You guys are already getting paid for it. That's right. Yeah.
1:23:13
So, all you all you have to all you have to believe is we actually ship the product. Mhm.
1:23:20
What was the uh I want to ask you what was the worst week building boom and what was the best week?
1:23:25
I'm guessing the best week was this year.
1:23:27
Yeah, I think the uh I I give you the best moment.
1:23:33
Um the best moment was the day that we broke the sound barrier for the first time, but it was it wasn't actually the moment of breaking the sound barrier because we' done so much testing at that point.
1:23:42
I knew if we flew that day, we were going to break the sound barrier.
1:23:45
Uh what I didn't know is whether we'd have the weather to do it and uh and like I'd gotten up that morning and it was extra cloudy and like we can't you know we can't do it on a cloudy day.
1:23:56
Um and uh and so the team was kind of debriefing checking the weather da da and uh and one of our safety culture rules is like no top management are in the safety briefs.
1:24:06
It's the team making an independent go no-go decision with no nobody senior putting their thumb in the scale and telling them they need to go and pressurizing it.
1:24:14
Uh but so the team so I'm like in the hanger waiting and the the team the team walks out and I could just tell from the energy like it was go time.
1:24:24
Like no no nobody had to say anything.
1:24:26
It just like I see him walking over the airplane.
1:24:27
They're like hooking up hooking it up to the to hooking the toe bars up like we're going to go.
1:24:31
And at that moment I I really teared up uh because I I knew I knew it was going to happen.
1:24:35
And then then we were then like when it actually happened and everybody was jumping up and down and screaming and I was sort of like, "Yeah, of course."
1:24:43
So that was uh I think that was the best moment.
1:24:45
The worst moment um we have near-death experiences like basically every year.
1:24:49
Um and at this point I just expected I'm like, "Okay, uh you know, every year we're going to get the equivalent of a cancer diagnosis."
1:24:59
And you know, if I'm lucky it's stage two, not stage four.
1:25:01
Um, but you know, we we we've had like some like we like survived stage four startup cancer.
1:25:07
Uh, uh, we had an extremely difficult fund raise.
1:25:10
I think we got down to a week of cash. Wow.
1:25:11
Uh, like the lawyers had a plan to shut the company down.
1:25:15
I remember calling one of our investors when we were sort of three weeks away.
1:25:19
Um, and he's like, "Usually when people tell me you're three weeks away, you're telling me you're going to shut the company down."
1:25:24
And I was like, "Of course we're not shutting the company down.
1:25:27
We're talking about how we're get through this little knot hole here."
1:25:31
And uh, people who are like who have like 12 months of cash and they're like I'm out of this.
1:25:35
I've been doing this for a year.
1:25:36
I'm going to go be a PM in big tech.
1:25:40
I've done the PM at big tech.
1:25:40
I don't want to do that again. I'll never go back.
1:25:43
You could never make me go back. I can't go back.
1:25:46
The the thing the thing with your with for you and your team and and I'm sure the entire cap table, it's like if we don't do this, we have to be if we're not successful here, we're going to be reminded multiple times a year. multiple times. Yeah. Like Yeah.
1:26:04
Every time you get on a plane, every time you you know, it like is is one of the very few companies on earth that can bend reality into into an, you know, bend time, right?
1:26:14
And I and I just think uh I feel like humanity is fortunate that you are running this company because so many people would have gotten the stage four diagnosis and just said, "All right, like we we took a good shot.
1:26:29
We did our best and that's it." Yeah.
1:26:29
I mean, it's my biggest wish for other founders is pick the startup where if you got the stage four diagnosis, you say screw it, we're beating this.
1:26:39
And that and that's that, you know, for me that's supersonic flight.
1:26:43
But every for every founder it's something different.
1:26:47
Uh but I think there's this founder market thing and if or founder mission thing and if you get that really in alignment then uh like like then then you can run through brick walls. Yeah.
1:26:57
Um, in many ways I would run through brick walls for a notetaking app. I would too.
1:27:04
Um, the in in many ways you could think of Boom as being a very logical, obvious next step in commercial aviation, longhaul travel, that type of thing. Yeah.
1:27:19
Um, what about the flying car market?
1:27:22
I've often said we have flying cars.
1:27:26
They're just helicopters.
1:27:28
But the problem with helicopters is that they're not evenly distributed.
1:27:30
Not everyone has a helicopter.
1:27:32
But if everyone had a helicopter on their house, they could fly to work.
1:27:36
And yeah, you need air traffic control and stuff.
1:27:39
We all need to be good pilots.
1:27:39
But if a company could drop the cost of a helicopter by 100x and make it 100x safer, uh, yeah, we'd probably basically say, "Yeah, we solve flying cars."
1:27:48
And yet that hasn't happened.
1:27:50
So what is your take on the boom of helicopters whether or the boom of flying cars or anything else in aviation?
1:27:59
There are great people working on that actually and it's actually way harder than supersonic flight. Way harder.
1:28:04
But Jo uh- because um oh zillion reasons like one is you need a whole new set of regulations right because these the electric vertical takeoff and landing so Joby and Archer who are really the two leaders in this. Yeah.
1:28:18
Uh it's not technically or regulatory a helicopter.
1:28:24
So there's a whole new set of rules.
1:28:24
uh they need a whole new set of infrastructure because isn't that a mistake?
1:28:29
Well, what I'm saying is like is like what if what if I started a flying car company that was perfectly regulated as a helicopter just like uh there these companies now that are regulated as sea gliders.
1:28:39
Billy Fowler and region are doing great work. Yeah. Yeah. I'm sure you love it. Right.
1:28:47
It's like legally a boat but it's like a plane but he has but he's designated in one way.
1:28:50
So it's a lot easier and I feel like there's almost like this regulatory arbitrage or regulatory hack that the hard tech founders might need to think through.
1:28:57
So it's this is if the regulations were done well, your idea would be exactly on point.
1:29:03
The problem is we have we have too many regulations that are prescriptive about exactly how something has to be done rather than setting a safety bar or a noise bar. Right.
1:29:12
And so and so like if you look at the regulations for helicopters, they tell you how you must build your helicopter. Okay. Yeah.
1:29:20
So, if you want to build an electric helicopter with distributed rotors, well, it doesn't have the parts that the regulations tell you how to design.
1:29:28
Um, uh, you know, or like how do you, you know, or that you're allowed to have like a wing on the thing, too.
1:29:34
So, so, so that like the better helicopter is that doesn't fit the regulations for helicopters. Yep. Yep.
1:29:40
So, yeah, it' almost be better bug in the regulations. Yeah.
1:29:44
almost just to regulate against like like if if flying things are crashing at a a rate above more than like one in a billion, it's no go.
1:29:54
Doesn't matter how you build it instead of saying you have to have this this screw in this place.
1:29:58
That's that that that's right.
1:30:00
And for actually for large transport aircraft, it's actually a bit better. Yeah.
1:30:03
One one in a billion is actually the regulatory safety bar. Yeah.
1:30:07
And um there are and yet there also still some things that are like prescripted but there's a me there's a mechanism to get around the prescription uh called equivalent level of safety.
1:30:15
So we can go to FAA and say um uh and say like you know you you say that our throttle has to be built in such and such and such a way.
1:30:26
Uh but the real thing is a reliable throttle.
1:30:28
Let me show you my reliable throttle that's done differently and you can get an equivalent level of safety finding. You can go forwards.
1:30:34
Uh but the if you look at like going from a helicopter to an electric vertical takeoff and landing, you know, flying car, uh like like the whole thing is different.
1:30:42
What's your what's your timeline to EV talls being in American skies? Who wins? Who goes first?
1:30:53
Will I be on a boom supersonic or a or a EV tall?
1:30:55
Yeah, I mean just a just kind of a wild guess. Is it like before 2030? Seems unlikely to me. Is it um is it 2040?
1:31:06
Is it I I I I don't I feel uncomfortable speaking for other people's timelines.
1:31:11
I think I think the technology is going to be ready uh but before the infrastructure and the regulatory environment are laid flat. Yeah.
1:31:19
Um and uh that the the stated timelines for EV toll are like super near-term. Yeah.
1:31:25
And but I I suspect it's optimistic. I don't know. And uh I don't know.
1:31:30
Maybe we've been made the mistake of being more realistic and and now it just sounds worse, you know?
1:31:33
So, our our timeline is 2029 for being ready for the first passenger.
1:31:37
I mean, there's also like incredible things happening in EVOL that are unmanned.
1:31:40
You look at what Zipline's doing.
1:31:43
We talked to the founder of Zipline and we were blown away by the progress there.
1:31:47
And that was one of the things where they had to go to another country to get favorable regulatory treatment and kind of figure things out.
1:31:54
Um, but then have really been able to accelerate. Yeah.
1:31:56
I mean, Keller's crushing it.
1:31:57
Like, like I love Zipline.
1:31:59
Uh, great amazing founder, amazing company, great success story for how you do these things.
1:32:03
They're they're brilliant.
1:32:06
Keller was brilliant at like finding the first market.
1:32:08
Uh, that was like the the soft target.
1:32:10
Y um the the the small and uncrrewed are radically easier than large and crude.
1:32:17
And the um and the re the reason for both is safety.
1:32:23
If you go big, um, even if it's uncrrewed, you have a problem because if the thing crashes, it can hurt somebody on the ground. Y, right.
1:32:30
And so there's a threshold, you know, somewhere around, I think, 55 pounds like above above which like it's radically harder.
1:32:37
Y um, and then if you put a human on board, like obviously, you know, you don't want to hurt anybody.
1:32:42
And so and so in building XB1 like the the the most surprising things were about um the sort of the second order effect of our our our choice uh to put a human on board from day one and and not to have an ejection seat like we we basically put everything on hard mode but we learned a lot more from that. Yeah.
1:33:02
Is there is there an easy mode version of the boom story where you go to some country with lax regulations?
1:33:09
You say we're going uncrrewed, we're going smaller.
1:33:12
Maybe it's like a ISR reconnaissance application.
1:33:17
You're only flying over the water counting on fish or whales or something and and it's just much easier to actually get flying on a routine start the flywheel like Zipline did. Have you thought?
1:33:27
I mean, you could go, you know, like there are people doing kind of what you're describing that I think are smart like Hermes. Hermes is smart. Yeah, totally.
1:33:33
And you know, but what is what what is Hermes doing?
1:33:37
They're they're basically building a hypersonic bomber.
1:33:38
Uh and uh you know, and it's kind of okay if the trips are one way. Sure.
1:33:43
Um you know, and and yet and this I think they'll have a great business and they're great people working on that and I you know, I'm I'm cheering for them.
1:33:51
They'll probably get to Mach Five long before I get to Mach Five. Yeah.
1:33:53
Um but the uh but doing that and then and then later putting a human on board is extremely difficult.
1:33:59
Like if you look at the history of development of aircraft platforms, you find that defense technology makes it into commercial technology, but there are basically zero cases of a defense product becoming a commercial product.
1:34:14
But there but the reverse is not true.
1:34:16
There are many cases of commercial products becoming defense products.
1:34:18
So the the Boeing 707 became the KC135.
1:34:19
The 767 uh became the KC46.
1:34:27
uh uh the the uh the 737 has turned into a whole bunch of command and control airplanes and anti-ubmarine airplanes.
1:34:34
Uh so uh uh the I and I think the reason is if when you're commercial you have to worry about safety and you have to worry about noise and and neither of those things is retrofittable. Mhm.
1:34:46
They're actually foundational to product architecture.
1:34:49
And so technology can go in either direction, but products go in one direction because safety and noise are fundamental to product architecture.
1:34:58
You mentioned AJ at Hermes.
1:34:58
When are we going to see the foot race?
1:35:00
I want the two founders of the supersonic companies to race each other to see who's faster in the real world on the ground and then we'll race the planes in the sky.
1:35:09
I mean, I think AJ AJ and I have been telling each other the other one's going to win the race. So, I don't know.
1:35:13
No, I I I think I think it'd be like the most at least for me, it would be like the most pathetic race ever.
1:35:18
I'm like I I am like not a runner at all.
1:35:20
Well, well, we'll do it as a pay-per-view.
1:35:23
We'll raise some non-dilutive capital.
1:35:25
You guys can maybe split it based on Let's actually make it a little bit less about the running. Let's just do the Murf.
1:35:32
We'll just do the Ander Murf together.
1:35:34
Uh well, we're going to hold you to being the first people to podcast Supersonic.
1:35:39
And I would I would love to have you back on the show many times between now and then because this conversation has been extremely insightful and uh we are just grateful for the work that you and the team do. It's great.
1:35:53
Thank you so much, Blake. That was so much fun. Thank you. We'll talk to you soon. Cheers.
1:35:55
And in the meantime, we will tell you about public investing for those who take it seriously.
1:36:00
Multiasset investing, industryleading yields, trusted by millions. Trusted by millions.
1:36:04
We'll also tell you about Aid Sleep.
1:36:05
Five-year warranty, 30- night risk-f free trial, free returns, free shipping.
1:36:09
How'd you sleep last night, Jordy? I got an 89. Did you smoke me again? 100.
1:36:14
Clinically backed sleep fitness.
1:36:17
He's got a 100 sleep score.
1:36:20
Put They got the new Pod 4 Ultra. Pod Five Ultra.
1:36:22
Pod Five Ultra with a with a blanket with the same heating and cooling technology. Got to pick it up. Can't wait.
1:36:29
Anyway, we got Tim Fist from IFP in the studio.
1:36:31
Uh we remember he was here when we were under attack and all of Zoom uh that was devastating. That was devastating.
1:36:40
Such an interesting conversation but we were technically Yeah. Yeah. It was awful.
1:36:45
So he's at the Institute for Progress.
1:36:48
We'll bring him into the studio and talk uh to him. How you doing Tim? Welcome back. Good.
1:36:52
Thanks for having me guys. We made it. We made it.
1:36:54
Thank you so much for for bearing with us.
1:36:56
Um, that was such a weird, you know, it's not the first time a nation state has to take down TVP.
1:37:03
We start talking and we get spicy and all of a sudden the connection gets fuzzy. It might happen again. We don't know.
1:37:10
Yeah, we had uh 10 cent was pretty upset at the uh H20 allegations.
1:37:15
So, yeah, prime suspect number one.
1:37:15
Take us through the allegations again, break down what you guys published, and then we'll go through uh some of the the reaction and the fallout from the piece. Yeah.
1:37:27
So, u god, so much has happened since uh this was in the news.
1:37:29
It feels like there's five different things in export controls that have gone on since.
1:37:34
Um but yeah, basically this was the uh H20 um inference chip from Nvidia, which some might remember was um designed to be compliant with the export controls.
1:37:44
Um so is this the chip they sold about um reportedly 1 million of them into China in 2024.
1:37:48
China in 2024. And yeah, there was a lot of outcry at the time for the Bureau of Industry and Security who's you know the part of government that administers um and enforces export controls to do something about this because you know 2024 was the period where everyone
1:38:01
realized that inference compute was perhaps you know the most important strategic input to frontier AI development because of you know test time comput scaling, reinforcement learning and synthetic data generation is the key things and um yeah uh you
1:38:15
know uh earlier this year uh there was reporting that you know Nvidia had a huge number of additional sales plan to like big Chinese companies and a bunch of people including us sort of said hey is this really what we want to be doing do we want to sort of allow you know
1:38:28
China to get access to millions more of these chips and yet the government ended up taking action on this and issued some guidance basically saying hey no you can't make these sales um Nvidia reportedly was left holding the bag to the tune of about 5.5 billion um and now
1:38:40
5 billion um and now you know since then we've seen a bag for ants for Nvidia it's barely fleshwood.
1:38:48
Well, they made up for it with deals in Saudi Arabia, right? Yeah, indeed. Yeah.
1:38:52
So, that's the new thing, right?
1:38:54
It's the administration walking back on the diffusion rule, the speaking framework, and then the series of deals that have been announced um over in the Gul and Yeah.
1:39:01
And what's your reaction to the diffusion rule and the deals in Saudi Arabia? Is this a step forward?
1:39:05
Are you excited about this? Is this positive news? Yeah.
1:39:09
So, it really depends on the details of all this.
1:39:12
So, I think um on the deals um I guess fundamentally, you know, what do we want?
1:39:17
We need the US AI tech stack to win the global competition against China.
1:39:22
And I think a big part of that is locking in these early adopters and big spenders, especially like the United Arab Emirates.
1:39:28
Um, but I think we need to be thinking about how to structure deals like this to sort of get the US the outcomes it wants, which is, you know, US tech diffusion, the kind of like Chinese tech stack locked out in a way that, you know, we couldn't handle with 5G, like China won sort of like the 5G battle globally.
1:39:43
um and then appropriate national security guard rails in place.
1:39:47
And it's pretty unclear like what the trade-offs that have been made for this deal are.
1:39:51
made for this deal are. So I think like the high level specs that we've gotten is um a 5 gawatt AI data center campus to be deployed over some time period in Abu Dhabi and then reports of 500,000 chips per year to be exported with maybe fourth of those for US firms who are building data centers over there and one
1:40:08
for G42 this big tech conglomerate in the UE and kind of depending on how quickly that all happens and whether they're referring to you know 100,000 of today's chips or 100,000 of like chips in uh you know 5 years time this could be the difference between you know 1x to 100x in terms of like different differential in compute. So yeah I think
1:40:27
So yeah I think that really makes the difference and I think the key question here is you know uh do we want an AI lab in you know an authoritarian country to have the biggest clusters or close to the biggest clusters in the world.
1:40:38
Um and you know this is a country that does collaborate with China in areas like drones and 5G and military technologies and so you need to sort of be careful with giving them access to sort of frontier scale compute here. Yeah. Yeah.
1:40:48
So there's kind of like you talk about I'm curious if you have insight on capital flows from the UAE and Saudi into Chinese AI is how how much investment activity is there? Do you have any insight?
1:41:03
I don't have stats on this.
1:41:05
I suppose the interesting to say on this you guys might remember there was this $ 1.
1:41:09
5 billion deal between Microsoft and G42 a couple of years ago that sparked a lot of this stuff.
1:41:16
of this stuff. Um so this was essentially you know Microsoft making investment into G42 and starting to build data centers in collaboration with them and the department of commerce got really involved in this and part one of their requirements was to divest from Chinese companies in the kind of AI stack um for G42 which is you know uh
1:41:35
this huge tech conglomerate uh that is uh funded through the nation sovereign wealth fund which is the second largest in the world so pretty serious money we're talking trillions of dollars um and yeah reportedly what they did is they had a bunch of passive investments in Chinese companies including you know
1:41:50
hyperscalers and AI companies and reportedly G42 did divest from those companies but then basically moved the investments over to another fund called Lunate that was sort of also owned by this you know big sprawling tech conglomerate that's ultimately funded by you know the sovereign wealth fund so big question marks about how much are
1:42:05
they actually decoupling from China and like how costly is this to them like are they actually burning bridges that are hard to reverse yeah there's some element of like we have this smooth gradient of friendship with different countries obviously ly there's the five eyes like the closest allies. We sell
1:42:17
We sell nuclear submarines to some countries.
1:42:20
Uh but then there's countries that are more jump balls and could go either way.
1:42:25
uh as the diffusion rule goes away, do we need uh firmer rules around, hey, we're willing to sell to you, but don't immediately set up a reseller and start selling, just passing these on to China because you could imagine if the natural economic forces take hold.
1:42:41
Uh and I was in some country that was just about to get 500,000 chips.
1:42:46
Um it's pretty easy to just immediately start reselling these if I and and just print money basically. Yeah.
1:42:54
And there's there kind of two ways to do that, right?
1:42:56
One is you can just on sell the chips, so kind of smuggle them into China.
1:42:59
And the other way that's pretty straightforward is set up your own data center and rent it out as cloud computing. Exactly.
1:43:03
So yeah, I think that's what you're referring to.
1:43:06
And yeah, currently there's not great guard rails around either of those things.
1:43:09
And I think the administration wants to set up better versions of this, but yeah, I think this needs to be part where there's sort of like countries that are really willing to, you know, buy hundreds of thousands to millions of chips.
1:43:18
You kind of want these structured deals that have these guardrails in place.
1:43:21
And I think the Trump administration is really really well positioned to strike these kind of smart bespoke deals that get us the right possible outcome across each of these dimensions.
1:43:29
Um and hopefully that's what they're pursuing. Yeah.
1:43:30
Uh can we talk about the news today or maybe it came out late yesterday that Nvidia to set up research center in Shanghai maintaining foothold in China.
1:43:42
Basically they're they're opening this R&D center uh in a as almost like an olive branch to China is kind of how I would describe it.
1:43:49
They they say they're going to use it to understand Chinese customer demands and design US compliant products.
1:43:56
Uh and they're basically doing this to just kind of navigate uh domest uh sorry navigate export controls and compete with companies like Huawei.
1:44:04
This uh to me I mean there's so much to unpack here.
1:44:11
I love your kind of initial take and then I want to kind of maybe move more high level.
1:44:16
And to me, this signals that that um maybe Jensen doesn't take national the national security concerns about like an AI war as seriously as even some of the US Foundation uh model labs uh talk about it.
1:44:32
Yeah, I'd say that's an accurate assessment.
1:44:38
And you know, I think it's really hard to design, you know, sanctions, export controls, like these kinds of things in a way that it's not that sort of can't easily be escaped from, but it's kind of like the spirit of the rule in that like, hey, we're worried about China, uh, you know, beating the US in like the most strategically important technology of this century.
1:44:55
Uh, and we don't want you selling to them.
1:44:56
But then, you know, you can do all these additional, you can do all these things to sort of like actually be cooperating around the scenes like design comps or whatever.
1:45:03
And to be fair to Nvidia, you know, like I think they say, look, look, if the if the speed limit is 60 and we're going 55, like we're not breaking the law.
1:45:09
Like we should be allowed to do this.
1:45:11
And if you think like the strategic perspective for them is also a lot of their customers are like these big US hyperscalers, right?
1:45:17
Um and all of these hyperscalers are developing their own custom silicon.
1:45:22
So you know, we know Google has their TPU um you know, that they use for both training and inference.
1:45:28
We know Amazon has like traium and inferentia that they're also using for training and inference.
1:45:33
Microsoft is developing their own stack and so you know huge source of revenue for um Nvidia is potentially at risk.
1:45:37
So it kind of makes sense for them to want to be uh you know diversifying to other parts of the world and not just be selling to us hyperscalers.
1:45:43
So they're certainly in like a difficult position overall.
1:45:47
And yeah, what you think is right here depends really on how much you buy this kind of argument that hey, over the next 5 years, AI as a technology that will like reshape the global balance of economic and military power is a thing and will be a really big deal. Yeah.
1:46:00
Um it seems like the speed limit is getting lower and lower though.
1:46:05
Uh we went from the H100 to the H20.
1:46:08
Now, Nvidia is preparing to release a modified version of the H20 chip for the Chinese market after your piece and the changes to the H20 restrictions.
1:46:16
Um, is there a point where the restrictions are so are so ownorous that no one wants to buy a car that goes 9 miles an hour?
1:46:28
You know, and a certain point Nvidia will just lose the market share because um Huawei Ascend chips will just be outperforming um are you tracking any of that?
1:46:41
Yeah, so here it becomes an interesting conversation.
1:46:43
conversation. I think the kind of crux of the matter is uh if you know Huawei has better chips than Nvidia in the Chinese market um and is able to sort of capture that market how bad is that like should we sort of set export controls such that Nvidia is always slightly ahead of Huawei for example and sort of
1:46:59
like raise those limits over time and here I think there's two sort of dimensions to it one is that you know in AI the quality of the chips matter so how sort of individually performant each chip is but also the quantity really matters so you know you can have a million chips uh each that are uh you know like half as good and sort of like
1:47:16
substitute for like 500,000 chips for example um and so you know this is a crude approximation but you both want the sort of best chips and as many of them as possible and where we're trying to squeeze China like we being like the US government is across this whole supply chain so not just on you know being able to procure chips directly but
1:47:33
also being able to manufacture their own so having like the semiconductor manufacturing equipment and the fabs and everything there and so I think the goal of the US government has to be really to restrict the quantity of AI chips that China, so like SMIC and Huawei as like the key firms here are able to produce. And so by this logic,
1:47:48
And so by this logic, you know, even if you know, you Nvidia has a chip that's uh only slightly better or slightly worse than Huawei's, you still might want to restrict it because you would prefer them not to have access to 10 times as many chips than they otherwise would have.
1:48:01
Like you don't want them to have access to essentially like, you know, TSMC's production capacity of like being able to do many, many millions more chips than you could otherwise produce.
1:48:10
So I think yeah these are like hard trade-offs to make.
1:48:13
I think where it really matters is in foreign availability.
1:48:16
I think you know where Huawei is accessing foreign markets and outperforming US chips that's really bad and we should sort of make sure that that is not the case and is you know it is US chips that are being used globally in countries that aren't China. Yeah.
1:48:30
Do you think Huawei will ever go public or do they not want people to know uh they want people to know as little about their business as possible? Yeah, I'm clear.
1:48:39
Depends what the CCP wants.
1:48:42
It's notoriously opaque organization. Yeah. Interesting.
1:48:44
Um, what is your take on open-source AI?
1:48:47
Um, we were talking to Erin Gin about this idea that uh, if America does not provide a state-of-the-art um, open-source stack to countries that want to build their own AI products off on top of fine-tuned or post-trained LLMs that meet their definition of uh, free speech or or ideals or morals.
1:49:09
Um the stack by default will be Huawei Ascend, Deepseek, Manis.
1:49:20
Yeah, I totally buy this.
1:49:20
I think that um being able to sort of have the best open source models in the world be American is really important for a similar reason as like the chip stack and the data centers and the cloud services.
1:49:31
Um I think there's a question about how sticky is this ecosystem actually.
1:49:37
So you talk to people who are like yeah we really need to like lock in the tech stack globally like American open source models need to be sort of like the rails that the world runs on.
1:49:44
But then if you look at sort of how AI developers work they are very happy to switch between different base models for their application depending on which happens to be the best.
1:49:51
You know you look at like the revenue of Frontier Labs and you know when they have the best model it's up here and when they don't it's down here like everyone is switching every day depending on like who who has the best model.
1:50:00
So it's yeah like you want to diffuse American open source as widely as possible but like what about it makes it sticky and my hypothesis would be that it's probably the secure security and reliability side of things.
1:50:12
So you know everyone's worried about you know the fact that you can insert like back doors to create slipper agents into open source models and there's no way to actually detect these.
1:50:20
Uh I think you know the US wins if it can prove that it has the most trustworthy and reliable models over China.
1:50:26
Similar to how I think US cloud computing companies compete over you know Huawei and Alibaba cloud like often Huawei and Alibaba are coming to the market with a cheaper option but the US is just more trustworthy in terms of you know data privacy security etc.
1:50:37
Um, so I think trying to figure out those technical problems around security, reliability, interpretability and sort of proving that US models win across those dimensions is probably the way to make the kind of open source models from the US more sticky.
1:50:49
Um, or just sort of be way better and you know this is uh where sort of most of the effort is currently going and definitely support uh those efforts as well like you know building out more domestic compute for example and like finding more training data sets. Yeah.
1:51:02
How do you think about the dynamic between uh the importance of the application layer versus the foundation model layer?
1:51:09
Uh we've seen efforts on the foundation model layer at the national level all over the place, but let's just use Mistral as an example.
1:51:18
Mistral has a consumer app called Lehat.
1:51:21
called Lehat. uh it is a direct competitor to chatpt and uh I think that's great for the French and they could potentially have a fine-tuned model that meets their sty standards and guidelines but if at the end of the day the French consumers 90%
1:51:38
of them are using Google and 90% of them use uh open AI well then all of that is kind of worthless uh and sure maybe mistall will be cheaper in the enterprise and be implemented in French businesses or European businesses. But
1:51:52
But in terms of like control of the population, that feels like the diffusion of American ideals in Europe, which doesn't seem extremely controversial because America and European ideals are ideals are pretty similar, but you can see how this would play out in other countries. Yeah, totally.
1:52:10
And the economics are pretty brutal here, right?
1:52:12
Like we're in a regime where the amount of compute being used to train a model goes up 5x every single year.
1:52:18
We're moving from, you know, hundred million to train a model, you know, last year to rapidly approaching the billions.
1:52:24
If you are a company committing to this and you're only sort of just slightly better or worse than, you know, one of these huge tech companies, you can't keep sustaining this over time.
1:52:32
You have to you have to you have to check out eventually. Yeah.
1:52:35
And we've seen that with a lot of the early stage foundation model companies that have raised trained something but never been on the frontier and now they're, you know, falling out of favor more or less. Yeah, totally. Interesting.
1:52:47
Interesting. Has there been any conversation on the on the ground in DC or have you heard any chatter around uh the Manis investment that that uh definitely kicked the that benchmark made kicked the hornets nest a little bit in uh the hornets nest of American dynamis deli and aspir but u but to to give uh benchmark a
1:53:10
little bit of credit you you if you're going to be mad at Benchmark for making investment in Manis you also in some ways I think have to be mad at Jensen for going up and setting up a research and design, you know, R&D center in Shanghai right now explicitly to work on developing products for the Chinese AI ecosystem and in some way direct the CCP directly. Yeah. Yeah. I it's a bit hard Yeah. Yeah.
1:53:36
I it's a bit hard to evaluate.
1:53:40
I think you know one uh you know the one of the US advantages is like very deep capital markets, right?
1:53:46
And so being able to exert financial control over setups overseas by sort of like acquiring stakes is potentially a way to uh you know um have you sort of a better sort of fairer sort of more aligned like global system overall.
1:53:58
I think this is hard when it comes to Chinese companies.
1:54:00
But does America benefit from having American capital allocators and bite dance?
1:54:05
I don't think we have much Yeah.
1:54:07
influence or control over what bite dance does. Yeah. Exactly.
1:54:13
So I think for big companies especially those that are part of this you know um Chinese sort of state industrial military complex uh this is a pretty poor prospect.
1:54:22
This is why you know the Treasury Department has outbound um investment restrictions in a bunch of different industries which have been expanded to AI over time relatively slowly though.
1:54:33
Um, so yeah, one one thing that they've been sort of working on is trying to figure out whether to apply these outbound investment restrictions more solidly to commercial AI developers and commercial AI cluster operators.
1:54:45
Like so like investments coming from VCs actually restricting investments of those kinds into China.
1:54:50
Um, and yeah, this is complicated by the fact that just like due diligence is really hard.
1:54:54
Like let's say you're investing into an application developer who's doing something pretty innocuous like, you know, automated code agents or um, you know, search or something along these lines.
1:55:04
Uh, you don't have any control over, you know, are they going to work with the Chinese military in the future?
1:55:10
Are they going to be sort of like an instrument of state back surveillance over the local population?
1:55:15
They're not going to tell you that and they might not not have plans to do that, but they can certainly be compelled to do that in the future.
1:55:18
Uh so the due diligence question is super hard.
1:55:22
Uh do you think uh regulators or the American government needs to be thinking about the pre-training scaling law potentially not holding or reaching some sort of diminishing marginal return.
1:55:35
We've seen the data from GPT 4. 5.
1:55:37
It feels like uh GPT5 might not just be 100x bigger than GPT 4. 5.
1:55:41
It might actually be some sort of mixture of experts, different models, and more of almost like a product challenge than just a scaling and just get the more chips challenge.
1:55:55
At the same time, OpenAI is also investing in um in Stargate and and there's still a drum beat of ever larger data centers in the United States.
1:56:05
But uh the the overall tone of uh of AI research labs in America seems to have shifted away from just the ever bigger transformer.
1:56:13
And there seems to be a somewhat of a somewhat of a resignation to this idea that there might be more challenges on the path to ASI than merely scaling up the architecture that we have right now.
1:56:31
Yeah, I think there's a few things here.
1:56:34
One is that, you know, obviously these companies are still making huge investments in clusters and energy to get to the next order of magnitude of scale.
1:56:41
So there's some level of just you know uh financial buy in to the idea that you know pre-training scaling will continue but also a recognition that uh you know we've got this other scaling law this sort of test time compute scaling law and now like reinforcement learning as a paradigm that seems to really be working for language models where a lot of companies are starting to put more of their compute resources overall.
1:57:03
overall. So I think the rough balance now seems to be around sort of 8020 pre-training compute and then post- training and I expect what we might see as we see sort of um you know companies who are building out these clusters and these energy sources to support them where are they going to sort of like
1:57:19
balance the comput allocation across those calculus seems to be that um yeah pre-training comput is relatively less promising to the four and putting more of your resources in a relative sense into RL which is sort of very much at kind of like the early stages of scaling up is the better strategy. at the at the moment. Yeah.
1:57:34
moment. Yeah. Yeah, I mean the the other side of this is like as even even beyond post- training and RL as reasoning tokens and just more test time compute, more inference cost increases, maybe the real way to get a GDP boost or
1:57:52
competitive advantage out of AI is just to make sure that there is an H100 or equivalent for every member of your society or every every citizen because everyone will need to be inferencing thing at a very high level, very large model, basically constantly. And so even
1:58:06
And so even if you've trained the greatest model, if you can't have every single one of your citizens constantly inferencing it all day long, you're not going to see the benefits of I need my AI companion constantly on.
1:58:19
I mean, we do need we do need, you know, codegen and we need research and we need answers.
1:58:24
If we're if we're timing out, it's not just codegen and deep research are going to be competing with the AI companions for inference. Yeah, probably.
1:58:35
Um, but just to I guess maybe push back on maybe hypothesis underlying that um I am very um confused about where most of the compute is going to be spent and how that is going to be distributed across people.
1:58:52
So I can easily imagine a world like in two years where most people in the world still aren't really using simple tools like chat GPT like you know my grandma still like never heard of it and uh but at the same time you have some companies at the frontier who are deploying millions to billions of agents internally.
1:59:08
So you have this really like unequally distributed use of compute overall.
1:59:12
Uh so yeah I I kind of buy the idea that there'll be just like massively uneven sort of like usage of these kinds of resources and like really like located in particular countries and within particular companies. Yes.
1:59:22
Agree in the sum total of the importance of inference and compute allocated and available for inference but potentially disagree on the distribution of that.
1:59:33
And I I I think now that you now that you hash that out I think that makes a lot of sense.
1:59:36
You already see that just in the proumer versus consumer market.
1:59:40
Uh, I'm probably kicking off like three to five deep research reports using a lot of tokens.
1:59:45
I'm getting my $200 worth and there's a lot of people that are, you know, just land on it every once in a while to toy around with it.
1:59:50
That makes a ton of sense. Uh, what's next, Jordy?
1:59:53
What should we talk about?
1:59:56
How h have you guys had success explaining uh the potential for AI, you know, AI's potential progress in the next few years in in Washington broadly?
2:00:08
Do you feel like lawmakers have fully kind of uh fully understand the potential? Right.
2:00:15
Nobody has a crystal ball.
2:00:15
Uh we can't predict the future, but if you sort of extrapolate trends and even just use the products today, uh do you feel like do you feel like Washington is is pricing it in or is it still uh you know or or are people going to be you know extremely surprised in the next few years?
2:00:38
Yeah, my bet is 100% extremely surprised.
2:00:40
I am I think consistently disappointed with how the lack of kind of level of AGI pled uh people in DC are even with sort of like current capabilities and like where the sort of trend is obviously going.
2:00:50
I think like there's a lot that just isn't being priced in about kind of how weird the world is um you know in 5 years time.
2:00:58
There's also I think a sense in which you know there's a real loss here in that um if you look at technologies like the internet which you know as you probably know like came from like this with this like DARPA funded project um
2:01:11
as well as you know um early sort of genomics with the human genome project these were technologies where the government sort of saw what was coming and took sort of a really active role in shaping the development of the technology through basic R&D. So with
2:01:22
So with Arpanet for example, the focus was really on creating, you know, a resilient network system that could survive a nuclear bomb.
2:01:29
Um, but also they sort of like they were able to take that sort of secure network infrastructure and apply the notion of kind of like openness and freedom of information to create like a scaled global network that really represented American values and was like very secure.
2:01:42
Um, we don't see like an equivalent kind of level of basic R&D investment in the United States around AI.
2:01:48
And that would be really cool to see because there's basically a bunch of problems where you know right now industry is focused on where the money is which is you know B2B SAS apps and chat bots but there's like a huge space of just like if you accept that over the next few years we'll have these incredible new AI capabilities.
2:02:00
there's all these massive important societal and scientific problems in areas like, you know, materials discovery, drug discovery, etc.
2:02:07
where the government could be placing essentially huge bets that could pay off, you know, within a few years.
2:02:12
And we're kind of not doing that because we're sort of in DC at least failing to see sort of where the future is going and therefore what the role for this kind of basic R&D is.
2:02:17
Um, so yeah, I'm there's a congressional coalition that's just started called the American Science Acceleration Project, which we're really excited about and try to build hype around.
2:02:26
Um but they I think I have the right the right idea around this but you know there's um a deficit of this kind of thinking in DC at the moment. Last question for me.
2:02:32
Um how has Meta's reputation changed in DC over the last few years.
2:02:40
Famously Zuck goes to Capitol Hill.
2:02:43
They don't even understand how his business model works.
2:02:47
He says, "Senator, we sell ads."
2:02:47
He was castigated for being too left then too right.
2:02:53
and uh very a lot of political hot button issues around the Facebook app and the type of content it's servicing.
2:02:59
Obviously, a big vibe shift there, but now it seems like Meta is increasingly a very important tool in the American foreign diplomacy AI tool chest with Llama.
2:03:09
Llama is of course in their open source model that's you know there's defense llama now the DoD has partnered with Neta on this and yet uh today we learned that their behemoth model is struggling to improve capabilities they're facing setbacks um has the has the tune changed in DC to say hey Zuck like sell more ads please like do more stuff we got to we got to get you pumping because like you're a national champion now.
2:03:34
Yeah, I'd say like by and large, you know, there's different factions in DC who care about different things.
2:03:40
Obviously, sort of Meta is going through this big like antitrust case at the moment as well.
2:03:43
Uh like has been going through it.
2:03:45
Um yeah, on the AI side, I guess it's interesting as well.
2:03:49
Um like Meta is very much seen as kind of like the darling of, you know, open source for the US.
2:03:53
I think it's kind of awkward for a lot of Meta fans to see the latest batch of models and then really not be that impressive.
2:04:01
and also potentially a bit of gaming with uh leaderboards and sort of what models they're releasing there.
2:04:05
Uh obviously they've executed the strategic pivot to appeal to the current administration with you know getting rid of facteing and bringing in like the community notes type approach and that seems to being pretty effective.
2:04:14
But yeah, I think um they've certainly got a lot of backers here.
2:04:18
Um yeah, and the open source approach it's like really good to have like a really well capitalized company like really like pursuing the strategy overall.
2:04:25
Um, but yeah, I think they're copying a bit of heat for not actually delivering on the promise to some extent.
2:04:31
And also people there's sort of like another faction that's pretty worried about uh open source models with capabilities that could be significantly misused, especially in the cyber domain being like freely available to the whole world.
2:04:42
Um, but that's less of a concern while they're behind.
2:04:43
Yeah, it was very interesting in the in the in the press release around this in the in the the management of the the news that they were delaying the roll out.
2:04:52
They didn't say, "Oh, it's it's too dangerous to release."
2:04:55
They could have easily said that.
2:04:56
That's always an easy out for the AI labs to say, "It's just too good. Trust us.
2:05:00
We we we we can't trust you with it.
2:05:02
We're we're doing more safety research."
2:05:04
Instead, it seems like there's a little bit of an admission that uh it's just not at the level we want it to be.
2:05:09
Jordan, this might be out of scope for you, but do you think that AI is are at a much uh is being used in nefarious ways in the context of social engineering attacks at a greater scale than people sort of realize today?
2:05:22
This was top of mind just due to the the Coinbase news this week, the leak that they had.
2:05:29
Uh and uh I think people broadly anecdotally reported that they just felt like they had a huge uptick in like sort of inbound calls and social engineering attacks.
2:05:39
engineering attacks. My my question is is AI is already when you're using tools like Sesame and and some of these lower you know Sesame is obviously state-of-the-art but even some of these lower latency video models should already be capable and more better at
2:05:57
social engineering attacks than like you know a PhD level person globally that that even even if English wasn't their first language and and and you Uh so anyways I'm curious if you have cyber security social engineering the any sort of like insights potential regulation around it. Yeah no insights
2:06:18
Yeah no insights into the true extent.
2:06:21
I will note that it is surprising to me that we don't seem to see more of this.
2:06:25
My impression is that from sort of GBT 3.
2:06:26
5 onwards we've had LLMs that can produce better text more convincing text than the kind of median fishing email which as you know is often like pretty poorly written.
2:06:38
Um, maybe that's deliberate, but yeah, like it's kind of weird to me that we haven't seen mass sort of spear fishing campaigns of the kind that have been possible for like several years now yet, or at least they haven't been widely reported.
2:06:48
Maybe it's that yeah, existing sort of filters and defensive approaches work.
2:06:52
Maybe like we shouldn't expect actually there'd be that many people who are trying to pick this low hanging fruit.
2:06:57
They're not technically sophisticated enough.
2:06:58
Or maybe it's like it's like happening and not being reported on.
2:07:00
Yeah, I find I think it's I think there's potential that it's happening, but it's happening to a demographic that doesn't even know it's happening, right?
2:07:08
Like I don't pick up random calls, right?
2:07:10
I just don't I not going to answer the call that you know of.
2:07:14
Maybe the last time I called you, it was actually a bot. Yeah, that's true. That's true, John.
2:07:18
You do pick up non-random calls, but that's the point of them. Yeah.
2:07:22
I mean, the same thing happened with the crypto stuff.
2:07:24
like there were a lot of the victims of cryptocurrency scams like weren't profiled in the new the New York Times and so we just didn't really hear about them because like the people that with the loudest microphones didn't fall for the scams. Yeah. Tricky.
2:07:36
Anyway, thank you so much for joining Tim. This was fantastic.
2:07:40
I'm glad we survived the attacks from the nation state actors that don't want us.
2:07:44
We kept the stream up a stream together. We really appreciate it.
2:07:48
We'd love to have you back and this was fantastic. Yeah. Thanks for coming on. We'll talk to you soon. Have a great weekend.
2:07:52
Uh, next up we're bringing Chris Best from Substack.
2:07:56
But first, let me tell you about Adqu. Adqu. com.
2:07:57
Out ofome advertising made easy and measurable.
2:07:59
Say goodbye to the headaches of out of home advertising only. Adqubines technology.
2:08:02
Out of home expertise and data to enable efficient seam buying, seamless ad buying across the globe.
2:08:09
And we should also talk about wander. Find your happy place. Find your happy place.
2:08:15
Book a wander with inspiring views, hotel grade amenities, dreamy beds, top tier cleaning, and 247 concier service is a vacation home, but better.
2:08:22
Anyway, let's bring Airbnb is getting distracted. Wander's doubling down. Wander's doubling down.
2:08:27
It's a knockout dragout fight in the rental home market and we recommend Wander.
2:08:33
Anyway, let's bring in Chris Best from Substack. How you doing? Good to see you. Doing good.
2:08:37
That jingle is unstoppable. Unstoppable.
2:08:38
We got to get a jingle for Substack. Substack.
2:08:40
What's the Substack jingle?
2:08:42
What do you have a tagline?
2:08:44
What What's your What's your landing page hero hero text?
2:08:46
The app for independent voices. There we go. Same same melody.
2:08:53
I go with take take back your minds. Take back your minds. Okay. Put that in a jingle.
2:08:59
How is the process of taking back minds going? It's going pretty well. Okay. Can't complain.
2:09:03
Cross 5 million paid subscriptions. Wow, that's a lot. Awesome.
2:09:08
Apps blown ahead of I think all of the like legacy media apps combined. Amazing.
2:09:12
Uh Chris and I were in the same YC batch uh winter 18. Another one.
2:09:18
I always So, so my biggest thing with John is is actually probably good.
2:09:22
If John had just invested in in in his batch, he would have been too post economic to start a podcast and we wouldn't be here.
2:09:30
It's like Coinbase, Instacart, Zapier, Substack, uh even uh what was the other one?
2:09:36
The the the crypto exchange that absolutely ripped the NFT marketplace.
2:09:39
OpenC was in our batch, you know. Banta Repid. Yeah. Yeah. Yeah. Remarkable companies. It was good times.
2:09:47
Um, so yeah, I mean take me through the the the history and evolution of Substack and where you're going next.
2:09:55
Yeah, I mean we started we were in winter winter 2018 together.
2:09:56
Um, the basic theory is we're building a new economic engine for culture.
2:10:02
Um, you know, we want to harness the power of technology, the power of world, you know, internet scale networks combined with a business model that actually powers independence.
2:10:13
The version of that, the simple version of that that we started with is basically make it dead simple to do a paid email newsletter.
2:10:21
Um, but as the years have gone by, we've grown that into multiformat.
2:10:23
You can do you can write, you can do a podcast, you can do live video.
2:10:27
Uh, we've got this network and this app.
2:10:29
You can get the Substack app from the app store and discover this whole universe of the smartest, best independent media and culture on the internet.
2:10:37
Uh, and that's what we're building next basically.
2:10:40
I mean, I'm we're building a bunch of uh video tools.
2:10:43
We've got this live, you know, live tool that lets you do something that feels like a FaceTime call and then AI magically turns it into like a produced podcast and series of clips.
2:10:53
So, if not everybody is the brilliant, you know, YouTube sensations that you guys are if you're just someone that has something to say, uh, we make it dead easy. That's very cool. Super powerful.
2:11:03
Was Ben Thompson really an inspiration?
2:11:04
He kind of tells that story.
2:11:06
I don't know if that's true that Sir Tekker and he kind of pioneered this like independent newsletter creator really scaled the business but I mean had you been reading him at the time had you taken lessons away from cert when you built Substack? Yeah, I do.
2:11:18
There was a few people doing it, right?
2:11:20
There was like, you know, because we had this this cockami scheme.
2:11:23
We're like, "Hey, I bet you people would subscribe to, you know, things they deeply value if if if you had something great."
2:11:29
And there was a handful of people that were doing it.
2:11:33
You know, Andrew Sullivan did this with kind of like the Daily Dish back in the day.
2:11:36
Ben Thompson had strategy.
2:11:36
And here was this guy, you know, writing an email newsletter from his bedroom in Taiwan making, by our calculations millions of dollars a year.
2:11:46
We're like, you know, that that sounds pretty sweet.
2:11:47
like why does why do more people not do that basically? Yeah.
2:11:51
And you know the answer turned out to be turned out to be it's way too hard.
2:11:55
So what if we made it way easier? Yeah.
2:11:57
I mean he's he's he's invested immensely in technology to build up Passport and what he's done and now there's a few other companies that run um but u um where does this go?
2:12:06
I mean I guess the big question is like ads product you know is is tied to the written media.
2:12:14
Uh Wall Street Journal has subscriptions and ads.
2:12:17
Um New York Times has subscriptions and even after you subscribe, you still get ads.
2:12:21
We are an ad powered show.
2:12:22
Ads feel very uh they get a lot of negative attention, a lot of negative stated preferences, but the revealed preference is almost always people don't mind ads and scroll right past them on Instagram and sometimes even get value from them.
2:12:37
uh we've been very pro ads, but what has your uh take been on ads generally in uh in the Substack world?
2:12:48
The thing that's actually different about Substack is kind of creator ownership and a model that rewards quality and value.
2:12:56
So the fact on Substack that you, you know, people don't subscribe to Substack, they subscribe to you.
2:13:02
subscribe to you. um you know you can make you if you make something truly great and it a bunch of people value what they come all those things are what actually sets Substack apart you know the formats are similar to what you see elsewhere you get you can you can write a long form article you can write make a podcast you can do a video like those
2:13:19
are those are not novel things the thing that's novel is this business model that powers independence the fact that you own the you know it's your business you are getting the upside you want to invest in the quality and I think there are methods methods of doing sponsorships and advertising that are that that that work with that, right? The fact that people, you know, people
2:13:39
The fact that people, you know, people are paying premium for you guys to advertise on this program, not because you're getting like the most clicks ever that gets sold at a market like at some market rate for the demographic, but because you're making something special and you have tremendous, you know, jingle skills, but the the audience is like is is magical.
2:13:56
Anyway, there are a bunch of people today.
2:13:58
We're doing all subscription.
2:14:01
we're powering that thing that actually works really well, shockingly well, much better than people thought it was going to work.
2:14:06
People are consistently surprised by how well that business model can work for them.
2:14:10
But there are also people on Substack who are doing sponsorships and doing tremendously well.
2:14:15
And so I do think that model can and should coexist with highquality media endeavors.
2:14:20
Do you believe uh the medium is the message?
2:14:22
Uh we we we've kind of felt this where we decided to go live for somewhat of like a technical reason.
2:14:28
Uh just with the way YouTube's organized, it kind of made things easier.
2:14:33
But then once we were live, we realized, well, we're faster.
2:14:37
We're not there's no delay between us recording and uploading.
2:14:40
And so we can react to the news.
2:14:42
We could we could open up X right now and and read a headline that just dropped with you and get your reaction.
2:14:48
And it's kind of changed the nature of the show.
2:14:49
What's been your uh processing of like the McLuhan ideology?
2:14:56
I'm a huge believer in live for that exact reason.
2:14:58
You know, we're building this video product that you that is live within Substack and I think it's almost like a hack that sets the social expectation.
2:15:05
The fact that you're going to be able to sit down, make this thing, ship it off.
2:15:09
It's not necessarily the case that the majority of the audience is going to watch live. Totally.
2:15:12
It's great that those of you here, but it's, you know, it's going to be there's going to be a VOD, there's going to be clips.
2:15:17
Those might be the places that actually get the distribution.
2:15:19
But as a hack to make the thing, the fact that the, you know, the social expectation between us right now is we're sitting down having a live conversation makes it easier, makes it faster. It does feel different.
2:15:29
Like I don't know if you have this, but I feel, you know, the feeling of being live is a little bit more energizing than than otherwise. Definitely.
2:15:37
And then also it just builds I mean we we we like it because it builds trust with you.
2:15:42
Like yes, there will be clips, but we're not going to edit.
2:15:44
We can't edit out what you're saying.
2:15:46
So if you want to make your case in some way, we can't be like, "Ah, we didn't want to put that in. You're not."
2:15:50
Yeah, I could I could go back to the tape and say, "Well, here's here's the full context." Exactly. Exactly. It's always there.
2:15:55
Uh can you talk about this?
2:15:57
So So one, I want your help kind of framing something because there there's in my mind there's this stages of of Substack, right?
2:16:04
this stages of of Substack, right? like our audience at least the initial core audience is a sort of like terminally online ex you know tech enthusiast genius wealthy that that that's a core audience but average you could say terminally online bill you know 20 to
2:16:21
you know goated definitely or like in the conversation definitely in the conversation no but but to me there's been these stages of substack where like I don't know if it was like a year ago at this point that I noticed that substack had not become a household name but it had been become outside of tech. Outside of tech, Emily Auster, for
2:16:38
Outside of tech, Emily Auster, for example, but had just become something that that kids I went to college with that aren't in tech were Yep.
2:16:43
subscribe to probably three Substacks and it was like part of that.
2:16:48
It wasn't just all like it became it became like clearly part of popular culture online. Yeah. Yeah.
2:16:56
And did was there a no was there a moment is that like the wrong analysis and I just wasn't paying attention outside of our bubble or did you feel something like that too?
2:17:05
I get this question a lot and I think people because it kind of grows in pockets right like it'll be there'll be you know this bunch of crypto people that all join at the same time or a bunch of politics people or a bunch of finite like there's sort of like these you know next adjacent market pockets as it grows.
2:17:21
And so all along there's I've sort of had people ask me this be like it feels like Substack is suddenly blowing up like what does that feel like? Has your life changed?
2:17:28
And internally like it's just been growing really consistently, really steadily.
2:17:33
Like the curve just looks like a steady exponential curve.
2:17:35
And so it is kind of always true like this is this is the most exciting time.
2:17:38
Um I have had I've you know I do feel it too like I I I get less blank looks when I tell people I'm working on Substack than I used to.
2:17:45
Um but yeah, it's been a it's been a steady march.
2:17:50
How has uh kind of the war with with X maybe war is not the right word for it but obviously removing links platform changes.
2:17:59
Yeah, platform changes obviously frustrated a lot of the core creators who had been building an audience in both places.
2:18:06
But for my view, it seems like they maybe, you know, created a monster like in the sense that like it maybe made you guys react and be like uh and and it maybe it was like a short-term win, but uh will maybe be a mistake long term and that maybe it's kind of pushed you guys to think even bigger about what Substack is.
2:18:27
I mean the funny thing here is that Ben Thompson credits his early growth with LinkedIn and sharing links on LinkedIn and then LinkedIn had an algorithm change and he had said that he it wouldn't be possible to build strategy the way he did in the modern LinkedIn era.
2:18:42
So there was kind of like a LinkedIn vibe too but yeah I'm interested to hear how you've been processing just all the platform changes really.
2:18:49
So I mean this has always been my theory is if you want to build one of these businesses you need a model that supports independence.
2:18:54
You need people to be able to subscribe to you and then you also need like internet scale networks to grow on.
2:18:59
And in the old days of Substack, it would be, you know, at the very start like Trajectory, you'd have this newsletter, this website, but then you'd have to go on Twitter to promote it or go on LinkedIn or go on Instagram or what, wherever.
2:19:09
We always want you to be able to do that.
2:19:11
We want you to be able to publish.
2:19:12
It's not a walled garden, like put it everywhere.
2:19:16
Put it on the RSS feeds, put it on YouTube, whatever.
2:19:17
YouTube, whatever. But you're totally at the mercy then of these other platforms that don't necessarily you know sometimes they go to war with you like Elon got pissed off at us but a lot of the time they just they don't care right Zuck can turn around and say we're not
2:19:31
doing politics for a little while because people got mad at us and if you're somebody that writes about politics that's really bad for you and so we always knew that we need to we needed to make our own network our own place [ __ ] Zoom reactions I got to turn that off is that the OS level. I feel like I've
2:19:47
I feel like I've turned this off a thousand times and it sometimes still comes back.
2:19:51
If we could just shoot whoever built that feature, that would be perfect.
2:19:54
Um, anyway, we had to build a network.
2:19:56
We knew we had to build a network.
2:19:57
You know, it doesn't make sense.
2:19:59
It's not like, you know, Facebook or LinkedIn or X or anybody's Tik Tok's job to help you like, you know, take your audience and own it and make something independent.
2:20:08
Uh, it's never like been their main thing.
2:20:10
So, we knew we had to do that.
2:20:12
That's why we built, you know, the Substack app in the first place, which is why Elon got so mad at us.
2:20:17
he felt like we were, you know, competing, but ultimately we just knew these independent places need their own network that actually wants them to grow and thrive.
2:20:24
Um, it was a big painful thing for people that were on the platform.
2:20:28
It sucked that you're like links didn't work. Super annoying.
2:20:30
Um, but it was a it was a very small fraction of traffic.
2:20:35
Like it didn't slow down the business or the numbers at all.
2:20:39
There was a lot of sturman.
2:20:39
It was very like st you know for people that are like had a big Twitter presence and were trying to make a substack.
2:20:44
It was very painful and stressful, but it didn't slow down the growth at all. That's cool.
2:20:49
What's your your yours in the in the team's sort of decision-making process?
2:20:54
It feels like at a bunch of different points with Substack, you could make a product decision that might drive immediate m basically make the number go up, but maybe wasn't aligned with the kind of platform that you wanted to be and and maybe not even aligned with humanity.
2:21:10
And on that note, there's sort of this like interesting thing where every platform maybe other than Substack is sort of like converging on being the same app.
2:21:19
This sort of like short form sloppification of of social media where it's like, you know, doing a slot machine uh with information and and it feels like Substack is orienting around long form content.
2:21:34
We had Jason Freed on yesterday. Yeah.
2:21:37
So that's sort of rambling, but No, I I completely agree.
2:21:41
I was going to ask the same question.
2:21:42
I feel like if I land on a Substack link, it's going to be written by a human.
2:21:46
It's not going to be particularly sloppy.
2:21:47
I might not like the particular topic or something, but I feel that it's going to have like this premium uh like vibe to it, I guess.
2:21:54
premium uh like vibe to it, I guess. And I'm wondering if there's is there are we just early and there's a coming wave of slop that you're going to have to fight or or Well, I mean, here's my take on this is, you know, I think if you put yourself in
2:22:08
a the position you described where it's like, hey, we've got this business, we're trying to make the number go up and we either need to make the number go up or we need to make something that's good that we believe in and we kind of have to like make that choice. I think
2:22:17
I think as soon as you have to make that choice, that already sucks because both of those choices suck, right?
2:22:23
It sucks if you kind of give up your principles and make the number go up.
2:22:27
But it also sucks if you stick with your principles and the business like it hurts the business and the business doesn't thrive.
2:22:33
Ultimately, the business of Substack needs to grow and support the thing we're making.
2:22:37
And so the thing that we've tried to do is yoke the success of the business, set up the business model and the fundamental way that it works so that in order to make the number go up, we have to do the thing that's actually good.
2:22:50
So, an example of this is, you know, the way that we make money is it's completely free to publish on Substack to any size audience.
2:22:58
You guys should cross-publish on Substack, by the way. That'd be sick.
2:23:03
And then we only make money when you make money, right?
2:23:04
So, we're trying to help you grow.
2:23:05
We make money when you make money.
2:23:07
And so, you know, we have an algorithm the same way that Twitter or Tik Tok has an algorithm.
2:23:11
We even support short form video. You can see clips.
2:23:15
You can read long form stuff.
2:23:15
You can watch a long form podcast.
2:23:16
But it turns out that when you're tuning the algorithm to introduce thing people to things that they deeply value and might pay for, the emergent effect of that is very different than if you take the same technology and point it at the goal of get people to spend as much time here as possible.
2:23:34
Yeah, that makes sense because I I I might pay and then be satisfied, close the app, but that's a win for the algorithm.
2:23:40
Our algorithm says great, right?
2:23:43
Whereas, you know, Elon's been public about the links thing.
2:23:44
It used to be just Substack, now it's everyone.
2:23:47
He's like, "Yeah, if you're scrolling your feed, you click into a long form post, you go and read that thing and get deep value from it, you just tank the metrics." Yeah.
2:23:55
You're not seeing any ads. Like, what are we doing? Yeah. Yeah. Yeah. Yeah. It makes sense.
2:23:58
Uh, how is the health of the overall creator economy?
2:24:01
There was a big boom where venture capitalists were investing in the creator economy.
2:24:05
Uh, that's kind of died down, but uh, how healthy is the creator economy broadly?
2:24:10
We were joking yesterday that uh Aerowan in many ways in LA is a product of the creator economy.
2:24:17
There has to be double digit percentage of Arowan's revenue.
2:24:19
It's just creator revenue, you know, that that flows in in different ways and then goes into in there for sure.
2:24:24
A lot of G Wagons in the parking lot.
2:24:27
It's the LA equivalent of like the dot bubble. Yep. Um air bubble.
2:24:31
You know, I've never liked I've never liked the term creator economy.
2:24:35
Even the term content creator kind of gives me hives.
2:24:37
Here's the thing that I think is not going away is, you know, the the media landscape is shifting.
2:24:44
The legacy models have been eaten up by internet things that don't necessarily support, you know, the old businesses.
2:24:52
And there's this shift to p a shift in power to independent creators, people like you that can just set up, start a thing, make something that matters, earn the trust of people by building, going direct, building an audience.
2:25:07
I think that trend is extremely robust.
2:25:10
I think that thing is gonna is sort of inevitably going to happen at this point.
2:25:14
But I do think it's sort of undecided which version of that future we get.
2:25:18
And so we see our you know the thing we're working on at Substack is to try to bring about kind of like the best and most valuable version of that future.
2:25:26
But I think that I think that shift is not going away. Yeah.
2:25:31
I mean, you uh talking about the the uh content creator as a bad term.
2:25:37
Do you prefer journalist, writer, like more specific scalpel like terms?
2:25:45
Is that what you're getting at?
2:25:47
Or what specifically don't you like about the idea of the word content or the word content creator?
2:25:51
The phrase it's pretty rational, man.
2:25:54
It just kind of feels it it it feels like a slop version, right? Yeah.
2:25:57
You're a journalist, you're a writer, you're a podcaster, thinker, analyst, broadcaster, filmmaker, you know, comic.
2:26:06
There's a million things you can be.
2:26:06
And I listen, I'm I've made my piece with creator.
2:26:09
Like, it's it's a good generic term.
2:26:11
Um, but I think there was I think there was a moment where we kind of like cargo culted the creator economy and everybody got really sort of hyped up about it in a fluffy way and missed the the the deeper thing that is actually still happening.
2:26:24
What actually happened is that there was like a specific data point which is that VCs realized that creator content creators were the fastest growing SMB category and they were just like okay we should deploy like a billion dollars.
2:26:38
We got to make money off this.
2:26:40
We got to make we got to find a way to make money on this.
2:26:42
And the the thing that the thing that people missed is that that had been a sort of they were it was sort of like a decade into that trend.
2:26:48
And so people funded a lot of businesses that were like banks for creators, but a creator is like, you know, why would a creator not just use ramp, you know, we use ramp, right?
2:27:00
Like or why would a creator not just go to bank of America and just get a bank account, right?
2:27:04
And so I think the the sort of interesting investable opportunities uh were the Substacks which is like a new economic you know not to use your tagline but like an econom creating a new economic engine for media.
2:27:20
They should have invested in Arowan because that's like the grocery store for creators or it's all down.
2:27:28
They should have just bought stock in Mercedes because they make the the the small SUV for the for the for the creator economy.
2:27:35
Um I want to talk about uh the value of curation versus instantiation of ideas.
2:27:41
is I'm not sure if you've been following uh Ben Thompson's erosion of like the evolution of like the printing press to the internet making distribution zero marginal cost to the instantiation of ideas with uh GPT and deep research and language models.
2:27:57
It's become easier and easier to create a deep research report. The writing is good.
2:28:01
It still feels like you can tell when it's AI written, but it's at least fine.
2:28:07
Yeah, it's at least fine.
2:28:08
Um, but I guess the question is, are there any uh are there any Substackers out there that are openly using AI tools to write, but the value that they bring it the human element is just curation?
2:28:23
Um because sometimes I I feel like uh I should just share all the deep research reports that I put together because they're cool and they're I'm asking interesting questions and the and the humanness comes from the question that I'm asking and very few people would think to ask that question to deep research and the the answer is less important than the question.
2:28:41
Um versus just the general trend in in language models if you have any takes there. Yeah.
2:28:48
I mean, the way I think of it is, you know, we even before AI, we already lived in a world of infinite content. You can't get bored.
2:28:54
You can't run out of stuff to see or watch. Yeah.
2:28:59
And so the limiting factor is like your attention and your life.
2:29:01
Like what what should you pay attention to?
2:29:03
Who are you going to trust? What matters?
2:29:05
Like sort of it's it's the human alignment problem.
2:29:10
Like this is what culture is.
2:29:11
It's not just getting what you want.
2:29:13
It's figuring out what to want in the first place.
2:29:14
And that's the thing that's actually valuable.
2:29:16
Even before AI, nobody's subscribing to Substack because it's like, "Oh, I don't have enough emails to read and I want to pay money for that."
2:29:22
You're subscribing for a perspective.
2:29:24
You're subscribing for some connection, some piece of trust, some piece of curiosity.
2:29:29
And so I think all of these these technology tools that just give people superpowers kind of supercharge both sides of that.
2:29:35
There's now there's, you know, a thousand times infinity content.
2:29:39
there's more than you could even more.
2:29:40
But also the people who who have those relationships can have so much creative leverage.
2:29:45
And I I I think like literally writing for me is the is one that's not that exciting yet.
2:29:52
Although it's not impossible, but like yeah, help me do the research.
2:29:53
Help me figure this thing out.
2:29:54
Help me, you know, put the pieces of this together.
2:29:56
I think there's, you know, there's even people that have, you know, Lenny Richitzky has his Lenny bot that people subscribers can talk to.
2:30:02
Um I think all that stuff is awesome.
2:30:05
uh talk about growth hacks.
2:30:07
If uh someone out there is starting a Substack today from zero, they have no audience.
2:30:12
Uh what can they do to turbocharge their business in the short term?
2:30:17
The biggest growth hack I often have to give people is just start the the start the thing.
2:30:22
A lot of people I talk to are thinking about, oh, I should do this.
2:30:27
Should I write 10 things?
2:30:29
Should I come up with this plan? Should I do this?
2:30:30
Everybody that succeeds that I see just goes, just gets going, just starts writing, starts publishing, you know, try to make something good, try to share with people.
2:30:40
Um, and really just have a very strong bias towards action, towards thinking and moving in public and then kind of like correct based on feedback rather than trying to come up with some genius scheme.
2:30:53
Um, all that said, you know, make great stuff, share with people.
2:30:57
Have you been able to dig into any of the quantif quantitative metrics around the most successful Substacks?
2:31:05
Like is there a correlation between posting weekly and and revenue or length of post?
2:31:11
Can a deep dive ever be too long on Substack or too short or is are there any patterns that you've seen amongst the the top performers?
2:31:22
you know, there is a there is a trend that says, look, all else being equal, being consistent and publishing pretty frequently really does help. Yeah.
2:31:29
Um it's it's a lot easier.
2:31:31
You know, if you're publishing multiple times a week consistently for a long period, your chances of success really do skyrocket.
2:31:40
Other than that though, we sort of have the problem, you see this in marketing, too, where sometimes the like the uh the opposite of a good idea is a good idea. Yep.
2:31:49
where it's like one thing works but also like the opposite of that thing also works.
2:31:53
You just got to find something that's sort of good and differentiated.
2:31:55
So yeah, make something that's interesting, authentic that you actually like and then make a lot of it is pretty we found that 100%.
2:32:01
I mean 15 hours a week, three hours a day.
2:32:05
And not only that, but I mean I think we posted on X 20 times yesterday and like 10 clips and like uh some of them don't do very well, but you just set the quality bar where you set it and then you just try and get those, you know, let the winners ride basically.
2:32:21
Well, I'll tell you what, you cross stream to Substack. Easy peasy doing it. We got AI auto clips.
2:32:27
You can make your own, but also people will just find them. I love it. Let's hear it for that. Let's hear for that.
2:32:34
I like the sound of that.
2:32:34
No, I'm excited to get over there.
2:32:35
To be honest, we've been uh so, you know, the core challenge for TBPN is that we're a startup, but we're, you know, John and I are the founders, but we happen to be live for three hours a day, and then we have to spend like a couple hours prepping the show, and so there's just like, and we've built out, we're individual contributors, an amazing team now.
2:32:56
Uh, but it's just about adding these other channels.
2:32:58
But I I see a ton of opportunity on Substack and I just love how thoughtful you guys have been about, you know, building the platform and staying true to your values. It's awesome to see.
2:33:09
I have one last question and then we'll let you go.
2:33:10
Uh lessons from Lulu going direct.
2:33:13
Uh what did you learn from working with her?
2:33:16
What have you learned from the most recent uh Lulu ideology and what it means to communicate as a CEO to an audience of investors, employees, customers, etc. I'm a huge Lulu stan.
2:33:29
I've learned a lot from her.
2:33:32
Maybe one thing that is is nonobvious that I got from working with her that I wouldn't have necessarily picked up on just from the output is kind of like the in in many cases the the principles and the morals and the facts come first.
2:33:46
The most important part of the go direct comms thing is not just how do I spin this or how do I posture it or how do I say it's like are you doing the right thing?
2:33:55
Are you you know are you are you willing to stand behind the message that you're coming with?
2:34:02
Um, I think that thing matters a lot. Yeah, that's great.
2:34:05
Well, thank you so much for joining.
2:34:08
We'll let you This is awesome. Come back on again soon. We'll talk to you soon. Great. See you on Substack.
2:34:12
We'll see you over there. Bye.
2:34:14
Let's tell you about Bezel. Get bezel. com.
2:34:17
Your bezel concierge is available now to source you any watch on the planet. Seriously, any watch.
2:34:22
Also, potentially creator economy startup.
2:34:24
A lot of creators getting watches.
2:34:26
Yeah, lots of folks in tech getting watches.
2:34:31
Their cap table is absolutely stacked.
2:34:33
It is talk about numerals benchmark series A.
2:34:36
Talk about a stacked cap table.
2:34:39
Uh anyway, our next guest is here.
2:34:42
We have Sean from stored announcing a uh a major size gong moment.
2:34:47
Welcome to the stream, Sean. How you doing? Good to be here.
2:34:49
Thank you guys for having me.
2:34:52
Kick it off with the funding announcement. What's going on? What's new with you?
2:34:59
Yeah, we're excited today to announce that we've raised $200 million across our series E. Congratulations. Series E. Series E.
2:35:08
Been waiting for that gong moment. That's amazing. Let's go. We're going to hit more.
2:35:12
We're going to hit more sound effects. Give me the Ashton Hall. Success. Overnight success. Yeah.
2:35:17
How long you've been doing this?
2:35:19
I'm a few months away from my 10y year anniversary with stories. Wow. Exactly. Overnight success. overnight.
2:35:26
Classic overnight success. That's great.
2:35:28
Uh, but introduce the company. Break it down for us. What do you do? For sure.
2:35:31
So, we're building a commerce enablement platform that's entirely designed to level the shipping experience for brands of all sizes with Prime.
2:35:40
Over the last two decades, retail has changed where you don't walk in a store and swipe your credit card and walk out with the product.
2:35:45
You swipe your credit card online and you walk out with trust.
2:35:49
Trust that you're going to get the right product when the brand said you're going to have easy returns.
2:35:54
And these massive giants like Amazon have realized that is what's driving today's buying behavior.
2:35:58
And so they've invested tens of billions of dollars into building out this competitive advantage.
2:36:06
Meanwhile, every other brand is kind of in this stone age where if it's take cloud computing, they're still building their own data centers, managing their own racks in their office.
2:36:15
And so we give them a scalable platform that combines an end-to-end physical fulfillment network that ships over 30 million packages a year.
2:36:22
Last year we hit about 15% of US households.
2:36:25
We powered over 1% of Black Friday, Cyber Monday.
2:36:27
But then all of the technology not only that runs that network, but that also speaks to the consumer.
2:36:33
So there's a high probability to either yourselves or people listening actually delivered you a package before powered that tracking link.
2:36:40
You may never even have known. Wow.
2:36:42
So talk about how asset heavy or asset light the business is. Uh what do you own?
2:36:48
Do you own warehouses, trucks, planes, boats?
2:36:51
uh what are you sitting on top of?
2:36:53
Who are your key partners to make this happen? Yeah, great question.
2:36:58
Our three pillars are really a network, software, and scale.
2:37:00
We give economies of scale across a network of assets.
2:37:05
Some of those we run ourselves.
2:37:05
So, we do now operate 13 fulfillment centers. Okay.
2:37:10
Employ about 2,000 individuals across those fulfillment centers. There we go. Congratulations. That's huge. That's crazy.
2:37:17
keeps me on planes uh all the time, constantly going to a different city.
2:37:21
Um but then about 70% of our business is an entirely asset light network, existing warehouses, existing trucks, particularly existing last mile carriers where we manage a network of about three dozen of them.
2:37:33
But then all of that overlaid with our technology so the customer has one consistent experience no matter where we deploy their inventory.
2:37:40
Uh we talked to Harley at Shopify.
2:37:43
How important are uh small businesses to your business versus going after the scaled e-commerce players?
2:37:53
Obviously, not Amazon, but maybe Walmart or, you know, Macy's and like the really big players. Great question.
2:38:00
I'd say we're kind of squarely in the middle where we serve a lot of mid-market brands, about 500 of today's market leaders.
2:38:06
So we power all of the deliveries for brands like Athletic Greens, Seed Health, True Classic TE's, Proactive, the Skincare, Equip the Toothbrushes, Billy the Razors.
2:38:17
It's a lot of today's kind of multiund million revenue leading brands, the types of companies you'd see walking the aisles of a Target for example. Yeah.
2:38:25
Is that is that mostly founder?
2:38:25
I imagine that a lot of the founders of those companies kind of in your boat, been in business maybe a decade, raised a bunch of money and kind of at the same conferences.
2:38:36
Uh is that how you're doing bisdev?
2:38:39
Do you have a massive sales force? Are you doing outbound? What's working? What's not?
2:38:42
Yeah, we are one of the flattest founder sales cultures you're you'll find.
2:38:46
Uh we actually sell I mean last year we were nine figures of new sales.
2:38:51
This year we multi- nine figures of new sales in terms of how we're growing.
2:38:55
We have a seven rep sales team and so uh we we are on the plane all the time meeting with our different brands, meeting with these customers.
2:39:02
Um and I think that's actually one thing that blew away investors in this round.
2:39:07
If you look at our last four quarters of sales beats, I mean we were 3x our Q1 goal.
2:39:13
This Q1 alone, we take most of the year, uh, we have a fraction of a kind of percent of revenue on sales and marketing and R&D to a lot of peer companies, yet a lot of ROI.
2:39:23
This is such a funny interview because not only is it live, which is obviously a little higher stakes, but there's also a very chaotic soundboard potentially throwing you off.
2:39:33
Well, I mean, there's so many every single I I I don't want to throw you off. I just Good news.
2:39:36
That's my personal reaction.
2:39:38
But if you can make it through a TBPN interview like Bloomberg or CNBC is going to be a walk in the park, no sound board, really easy. You got a question?
2:39:45
What was the what was the dynamic?
2:39:47
I mean, I imagine the last, you know, five, six weeks have been intensely stressful just because of what your underlying customers have been going through.
2:39:58
Did the round get kind of done before that or were you simultaneously navigating trade?
2:40:03
I imagine you were navigating a trade war and a new financing just sounds sounds intense but but what did the timeline look like?
2:40:13
Yeah, we kind of come to the principle that everything crazy happens at once at stored so it's uh it's never a time time off.
2:40:20
Uh I'd say that we were laying the groundwork for some of this before the trade war really started.
2:40:25
Uh, and it kind of threw a big wrench into the fund raise in terms of some people realized how good it was for us and some people got really scared.
2:40:34
Uh, thankfully and it's unfortunate to say, store grew massively at other issues.
2:40:39
Take COVID, take war breaking out.
2:40:41
Take um, uh, the UPS or Canada Post strikes.
2:40:45
All those are a reason for a brand to say, you know what, I'm not going to face this on my own.
2:40:50
I don't have the economies of scale to stay flexible.
2:40:52
I don't want all the risk on my business.
2:40:54
let me go to a network like stored.
2:40:56
And so thankfully very similar here where there's really two tariff issues going on.
2:41:00
One is anything that comes into the ports from other countries in large quantities and that's kind of the standard tariff people are talking about.
2:41:08
The other one is the dimminitimus IMAX section 321.
2:41:14
all these e-commerce brands sending small packages into the US, not via a container on the water, just a small package where if it's under $800, they haven't paid taxes or tariffs on the import.
2:41:25
That was really made for people like us, individuals traveling internationally, sending products back to families, and it got exploited into this massive program where about half of Shopify's top 100 e-commerce businesses were shipping from outside the US.
2:41:39
And so when that changed, all of a sudden all these brands had to have this influx of volume into the US and these traditional providers, again back to kind of the data center analogy, are telling them, "Oh, we can get you live and set up a new fulfillment center for you in six to eight months."
2:41:54
We had a case study with that true classic t-shirt brand, multiund million revenue retailer.
2:42:00
Took them live from meeting us and signing to fully outbound shipping in 18 days.
2:42:04
And that's only possible on a techdriven network.
2:42:09
How much of the business is international versus domestic?
2:42:13
About 9% of our revenue would be either packages leaving the US or actually holding inventory uh internationally. Okay.
2:42:20
Uh I have another question. Go for it.
2:42:23
Uh I I noticed that the the raise was a mix of debt and equity.
2:42:26
Uh what are you using debt for?
2:42:29
Uh, and how do you think about is this the first time you've really included debt in a big fund raise?
2:42:36
Um, and then I have more questions that we can riff on after that.
2:42:40
Yeah, I think for storage, we're at a late stage where part of what we announced in this round is profitability, which I think is a lot rarer in a category like ours.
2:42:48
We've spent multiple quarters in a row now consistently profitable.
2:42:52
And that's compared to venture times when you're in a rapid delivery business and people are wondering are we using venture dollars to subsidize fast deliveries.
2:43:00
Well, I think the the proof is in our unit economics and in that profit.
2:43:04
But at the same time, we still had a strong balance sheet from the 300 plus million dollars we had raised prior.
2:43:09
And so when we looked at this round, we kind of said, let's raise the right amount of equity, but let's also use the scale, the profitability to complement the balance sheet with the right cost of capital and the right flexibility.
2:43:20
And part of it comes down to we actually have been inquisitive in the past as well.
2:43:24
We've acquired three businesses over the last few years.
2:43:28
All existing fulfillment centers because we've just seen if we onboard existing infrastructure to our technology, we can multiply the success of those customers, the profits of the the the acquired business and more.
2:43:39
And so, uh, plan to be on the lookout for more opportunities like that as we keep growing, which I think, funny enough, we started that, our first one in 2020.
2:43:50
Very not in vogue for venture-backed businesses to be making acquisitions now with this kind of AI wave and more.
2:43:57
There's actually some specific funds that are just being built to roll up traditional businesses and apply technology to them.
2:44:04
What are you seeing that's exciting around actual uh fulfillment automation and robotics?
2:44:12
This has been obviously a tough challenge that again the major players have invested billions of dollars at this point into uh and yet oftent times fulfillment is still a very manual sort of process.
2:44:24
It feels like if you're building a 3PL from the ground up, it's maybe easier to think robotics first if especially if you're a tech company young.
2:44:32
You're obviously aware of everything that's happening in AI, but uh at the same time, hard to replace a human.
2:44:37
They're pretty versatile.
2:44:40
Yeah, we're very excited about both AI and robotics because labor and humans are one of the biggest costs in a business like stored.
2:44:46
business like stored. and going all the way back to day one starting even back then we were seeing peer startups in in different cohorts and accelerators and more that were building drones for inventory scanning and all this robotics and uh it oftentimes shock someone when you step back and you say hey you realize over 60% of US warehouses aren't
2:45:06
even using a digital WMS they're doing pen and paperbased picking it sounds like it's made up it doesn't sound possible but it's true and so then you kind of look at this gap of where the industry is And if they can't even get to that kind of threshold, one, getting to humanoid robots or AI deployed on how to slot in a warehouse is essentially impossible. And so there's such a
2:45:25
And so there's such a fundamental advantage when we've built an entirely vertically integrated tech stack of we're really the only ones like Amazon when we're making you that promise in the cart.
2:45:36
Hey, order now, you'll get it by 5:00 p. m.
2:45:37
tomorrow or the next day.
2:45:39
We're actually going through not only the front-end consumer tech, the order management layer, the transportation management layer.
2:45:45
We're looking at we actually have this unit in Las Vegas right now.
2:45:49
We can get it to California by tomorrow if we ship it right now.
2:45:53
And so we're connecting that vertically integrated system which just gives us so much more opportunity to deploy robotics, AI, and more.
2:45:59
And so a lot of it's on the AI wave right now.
2:46:04
How do we use it to optimize demand planning, inventory placement, parcel selection, promise to c the consumer and more?
2:46:11
And the next phase is we've been using some robotics particularly around the uh conveyance slotting uh and picking in the facilities.
2:46:18
But when you can go from static with a moving arms to both moving arms and moving legs with a humanoid, it gets really interesting. Mhm.
2:46:28
Uh on other big tech trends, uh what are you most optimistic about across uh kind of the EV tall package delivery that we're seeing from zipline to uh something like automated trucking uh self-driving trucks to uh maybe even something just like really really robust language model driven AI agents uh just doing some of that paperwork for example but 100% reliably like what technologies are you the most optimistic about?
2:46:57
And if you have any timelines, I'd be interested to hear them.
2:47:03
Timelines is always the question mark because uh even some of the ones you just mentioned, there's been heavily debated and now disproven timelines over the last decade already.
2:47:11
So I I struggle to make promises, but but which one would have the biggest impact on the customer experience?
2:47:18
uh which one should we be really rooting for uh in terms of just speeding up the the time to delivery and reducing the cost?
2:47:25
I think any form of autonomous delivery whether the amazing teams over at Zipline and that kind of localized rapid last mile uh some things that failed even at Amazon like the driverless kind of sidewalk robots.
2:47:40
Anything that helps connect that last mile autonomously bends the cost and speed curve so dramatically that that's probably where we remain the most hopeful and excited.
2:47:51
But with a model like store being the network, essentially what we're doing is aggregating the demand and putting it on one form of tech.
2:47:58
But a lot of these businesses you're mentioning are actually key partners of ours where somewhere in our network we're able to kind of test this net new technology.
2:48:07
So, same thing with one of the major uh grocery and kind of food delivery platforms.
2:48:11
We have a pilot in a few cities for same day delivery for some of our brands, 2hour or less type delivery.
2:48:18
Uh but you wouldn't assume if you looked at store that we're working with that type of company because it's buried in the network.
2:48:24
So anything autonomous last mile, anything humanoid in a facility and really anything around demand planning with AI, I think that is the most critical because if you talk to brands, demand planning is the thing that really only enterprises will say they do well and I think most of them are uh a little self-impatuated when they say they're doing it well.
2:48:45
Demand planning is the biggest struggle in anything physical supply chain and speaks to uh the problem our friends at Pelaton had during co it's really hard.
2:48:56
Uh I'm looking at the bottom of our ticker.
2:48:57
It's uh poly market has the US recession in 2025 dropping like a stone.
2:49:04
Let's hear it for the US economy.
2:49:04
But my question for you is uh are you seeing data on the health of the US consumer?
2:49:12
How how is uh demand in the US economy?
2:49:18
Yeah, we have a really interesting kind of front line to the consumer across a lot of industries.
2:49:22
We've purposefully positioned to very macro resilient industries.
2:49:26
Things like health and beauty, you still the same makeup and skincare and bad times, things like nutrition and supplements, very similar.
2:49:34
A lot of subscription orders that we fulfill, almost 50% of the volume we ship.
2:49:38
And so thankfully our brands have been pretty well insulated.
2:49:40
But we actually saw an interesting trend in April which was uptick for many of them and we were questioning it saying is this consumers buying because they think prices are about to go up or what is this signal?
2:49:51
And so so far we haven't seen a kind of bull whip from that negatively in May.
2:49:55
And so year to date our our metrics and kind of markers to the consumer have actually remained pretty strong even though there's a lot of kind of fear and uncertainty when you when you watch the news and look at the macro. Let's go.
2:50:06
Let's hear it for the American consumer.
2:50:08
Endlessly relentless undefeated.
2:50:10
Uh this has been this has been fascinating. Thank you for coming on.
2:50:13
I remember the first time I heard about you and stored was from John at Strike I think in 20 21 and he was just so incredibly uh bullish on you and uh I can see why.
2:50:26
So thank you for coming on and congrats to the whole team on the milestone. Means a ton.
2:50:30
We're very proud to have strike le and uh I think uh something like 50% of the store investor base has has joined TVPN so far.
2:50:37
I saw the Kleiner chat at SUSA. So Oh yeah.
2:50:43
Thankful for not only send us more.
2:50:47
Well, come back on when you have when you have interesting data too.
2:50:48
Doesn't you don't need to come on just for fundraising news.
2:50:51
If you're seeing stuff that you think would be interesting to us and and the audience, love to have you back on. Thanks so much.
2:50:57
We keep the news flowing, so we'll reach back out soon. Thanks. We'll talk to you soon. Cheers, Sean. See you.
2:51:02
A little sound [ __ ] elevate the energy at the end of a long week.
2:51:08
It's Friday, but that doesn't mean we can't listen to the Ashton Hall sound effect, which I haven't got enough of.
2:51:13
We're going into the timeline, John. The timeline.
2:51:15
Okay, welcome to the TBPN timeline where we review the best post.
2:51:20
Many people said, "Oh, it's almost 5 on the East Coast. They're going to stop.
2:51:25
They're going to stop podcasting.
2:51:25
They can't podcast for more than 17 hours a week. We did it.
2:51:29
We have multiple four-hour streams this week. Yeah. Big big big week. Uh never podcast weekly. Always podcast strongly.
2:51:38
Always podcast strongly and daily.
2:51:40
Um scientists use crisper to rewrite DNA inside a living baby.
2:51:46
Fixing CPS1, a rare lethal liver disorder.
2:51:49
No transplant, no viruses, just three tiny L&P crisper doses.
2:51:51
Gene designed dose to design uh to dose in less than six months.
2:51:56
Personally, personalized gene therapy is finally a reality today, says Dee.
2:52:00
Uh, the world's first personalized crisper therapy given to baby with genetic disease.
2:52:05
What uh what a white pillastic story.
2:52:09
The footage from this is really sweet.
2:52:11
Uh, so there's a whole video about this.
2:52:13
You should go and watch it this weekend.
2:52:14
Uh, but Dee goes on to say, uh, right now this is applicable to single cell single gene well map mutations in organs that can safely be reached, which affect thousands of babies every year.
2:52:24
It costs under $5 million now.
2:52:26
So that's obviously expensive.
2:52:28
Uh it'll need to come down fur further and it was covered in, you know, all the mainstream media.
2:52:35
Um but what a what a fantastic uh piece of news.
2:52:38
Uh a baby's life was saved by crisper.
2:52:40
I remember learning about crisper um back when my co-founders were at Caltech and it was kind of this hot technology.
2:52:47
There'd been actually a few previous uh technologies to edit DNA.
2:52:53
zincfinger nucleases was one of them.
2:52:56
And and uh all the hot PhD research being done in in bio and bioysics uh back in this was like 2012 2013 at at Keltech in their bio division was uh all about crisper.
2:53:08
There were a couple people that um uh spun out companies around this technology.
2:53:12
Uh Jennifer Dudno won the Nobel Prize and has a fantastic um uh biography written by Walter Isacson all about that journey.
2:53:23
Um it's a it's a very interesting technology.
2:53:25
It took so long to get here but it's finally having an impact.
2:53:26
I mean it already has in many ways but very very exciting to follow. Good story.
2:53:31
Um, also out of the scientific community, wonder if we could get gene therapy to make us closer, even closer, even closer to gold ret. Absolutely.
2:53:43
Just permanently change my DNA so I can only be friendly. Yeah.
2:53:45
Just turn off the unfriendly gene entirely. Yes.
2:53:47
And also make me hotter and dumber because that's key.
2:53:52
You can't just be friendly.
2:53:54
You also have to be hot and dumb if you're going to go full golden retriever mode. Uh, anyway, I love this.
2:53:57
I love this uh headline from the Wall Street Journal. Forget humanoids at MIT.
2:54:02
Worms and turtles are inspiring a new generation of robots. Pull up this image.
2:54:08
This This is such a great image.
2:54:10
Yeah, look at this turtle robot.
2:54:10
He's so proud of his turtle robot. It's great.
2:54:14
Uh so CSIL, which is their artificial intelligence laboratory, uh envisions robots beyond humanoids, including soft, flexible, and even edible designs. Edible robots. Eat your robot.
2:54:23
You're going to be able to eat your robot. Eat the robot.
2:54:28
This is what uh Joshua was talking about, right?
2:54:30
Like let's let's explore every potential magical form factor of robot soft robots like a sea turtle.
2:54:38
Somebody acquired somebody acquired robot. com this week, too. They came out with it.
2:54:43
Edible robots for non-invasive surgery.
2:54:45
So, you eat the robot, it crawls around inside you and cleans you up and fixes you up, sews you up. I mean, yeah.
2:54:51
I mean, if you have internal bleeding or something, yeah, eat a robot, it'll sew you up, I guess.
2:54:57
Uh Russ's lab is also using new types of AI models inspired by the neural networks of worms to power robot brains. Uh what a fun story.
2:55:04
Anyway, uh and this is something that's actually gone on before.
2:55:07
I found a different video um from a different group uh that says, "We have created a robot based on a leatherback sea turtle which is alive today leisurely swimming in motion.
2:55:18
Next, we plan to modify it uh to an ancestor of the leatherback sea turtle that remains as a mezzo messoic fossil to make it swim."
2:55:25
Uh, and I think this is just like I don't know, Robus Robo Robuxi. I don't know. Um, very interesting.
2:55:33
He's just swimming around.
2:55:34
Just made a made a robot turtle. Fun.
2:55:36
People like turtles, I guess.
2:55:36
Um, we talked about a few of this.
2:55:39
Uh, what are the other funny posts?
2:55:42
Uh, I mean, more more fallout from Google.
2:55:44
The design lead, Android Auto.
2:55:46
This guy, this is somebody's LinkedIn profile if you're on audio.
2:55:51
This person says, "Spent 40% of my time arguing with the worst PM I ever worked with, 20% managing and coaching amazing designers, and 40% on the inefficient overhead of simply working at Google.
2:56:02
It's wild to see ideas we talked about in 2015 finally coming out in 2020, but they happened." Yeah.
2:56:07
And uh this is of course a quote post from Daniel says, "It's amazing how good Gemini is and how bad every product manager at Google is."
2:56:15
Is it are I feel like the PMs aren't that bad.
2:56:18
It's more just like the organizational you ship your org chart and they just they they just can't move fast enough to like position the products correctly.
2:56:26
Like the products are good, they're just not they're just not positioned correctly or or or talked about properly. I don't know.
2:56:35
Uh Cinder Pai did an interview with the All-In podcast.
2:56:37
He uh sat down with Dave Freedberg.
2:56:39
Um I'm excited to listen to that this weekend and dig into how he's thinking about product design and AI generally.
2:56:45
And I'm sure there's a lot of good insights in that interview. So, go check it out.
2:56:49
Uh uh we have a new landing page for grapile grapile cat.
2:56:56
This is not going to come up.
2:56:56
I I just thought this was a fantastic website. Okay. Okay.
2:56:59
And the AI viewer catch 3x more bugs. Merge 4x faster. 100% codebased content.
2:57:06
We got to pull up the I'm going to share I'm gonna share it. It's a fun name.
2:57:10
These AI companies, these SS companies, sometimes they get wild. Dare team. It's in the chat. Let's pull this up. Great, great sass name.
2:57:18
You got to see the animation though because the static image.
2:57:20
So the so so the the reptile the the lizard is is eating catching bugs. That's good.
2:57:28
Little on the nose at the medical. Do we got it? But we respect it. Oh, look at this. Cute. Very cute.
2:57:32
That is a fantastic website. I like it. That's fun. I'd love to see it. That's great.
2:57:39
Uh Dan Romero giving us a shout out. Little shout out.
2:57:41
I'm I'm excited to have Dan on the show.
2:57:43
Yeah, getting him locked.
2:57:45
TBPN format works because it generates daily content flowing out of the 48 hour timeline zeitgeist that can be cross-osted in video favoring algos on every major social network.
2:57:54
Weekly podcast is too slow and stale in comparison.
2:57:58
The only live competition is CNBC and they are too slow boomer downstream compared to Twitter. Interesting.
2:58:03
Um 48 hour timeline we show we do we're on a 24-hour timeline.
2:58:07
No, we sometimes catch stuff a little late. True, right?
2:58:11
Yeah, we got to be more on top of it.
2:58:12
We got to spend more time. But we're newsmaxing. We're newsmaxing folks.
2:58:15
Uh again, thank you uh for a wonderful week. Yeah.
2:58:17
Uh I hope everybody has a fantastic weekend.
2:58:19
Do you want to talk about the the public. com vibe investing?
2:58:24
Yeah, we should actually cover this. This is very cool. So, uh public.
2:58:28
com obviously sponsor of the stream has launched the ability to create synthetic portfolios around basically any investment idea you have and then back test them against the broader market, which is where it gets really interesting.
2:58:41
Uh so, we have a couple of these pulled up that we can share with you.
2:58:43
The first one is uh founders fund funded companies that have graduated uh and gone public.
2:58:48
How did they do in the public markets?
2:58:50
And you can see companies that were backed by founders fund have done 659% return over this uh test uh versus 150% in the S&P.
2:59:01
Uh, Sequoa's done similarly, little up and down here for Sequoia.
2:59:08
123% um versus 147 in the S&P 500.
2:59:11
Yeah, a little bit rockier, but you have done very well.
2:59:17
Uh, 589% versus 150 in the S&P.
2:59:20
A6Z also done well with 466% total return versus 150% in the S&P.
2:59:29
And this is where it gets really interesting, John. the F1 index.
2:59:31
These are companies that sponsor F1 teams and they have dramatically outperformed. Yes.
2:59:40
Uh the S&P 317% uh total returns versus 147% for the S&P overh the best portfolio that they tested is the size compensation size lords.
2:59:56
Uh, these are highly paid CEOs, often controversial.
2:59:59
2,000% return based on investing and CEOs that get paid a lot.
3:00:05
Uh, versus 176% in the S&P over the back test period. That's actually insane. All right.
3:00:11
It turns out when CEOs get paid well, Yeah. shareholders back out.
3:00:16
And also, who else outperforms? Bald CEOs. Bald CEOs.
3:00:19
Give it up for the bald CEOs.
3:00:25
433% versus 150 in the S&P. Uh, fascinating.
3:00:28
Bezos probably doing a lot of heavy lifting there. Lot of heavy lifting. Yeah.
3:00:32
But you got the Brian Armstrongs. Yeah. Right. Strong. Strong.
3:00:37
He's got a good I mean, this is insane.
3:00:40
I'm I'm interested to just continue tracking this over time because it really tells an interesting story. Yeah.
3:00:46
Not investment advice obviously, but uh never investment advice, always entertainment advice.
3:00:51
Steve Balmer, another bald CEO. There's a ton of them. Yep.
3:00:55
A lot of good lot of good bald CEOs out out there.
3:00:57
So, shout out to your bald friends.
3:00:58
Anyway, thank you for watching.
3:01:00
Leave us five stars on Apple Podcast and Spotify.
3:01:02
And we will see you on Monday. Thanks for watching. Cannot wait. We'll see you soon. Bye. Cheers. Have a good weekend.