0:01
[music] I see a large IP on the horizon.
Search this video
0:10
You're surrounded by journalists. Hold your position.
0:21
>> [music] >> Overnight success.
0:29
The laser 3000 double leg. That's misinformation.
0:56
[music] >> Clearing order inbound. >> Let's just roll.
1:10
We are surrounded by journal. Hold your position. Bump it up. Trust the extra five. We are extra founder. Very good.
1:43
I see multiple journalists on the horizon. Stand by. >> UAV online. >> Blaze. Double blaze. Triple glaze.
2:10
>> [music] >> Double kill. [music] Cop is wait. Team Deathmatch. We are experts. Triple blades. Let's just roll. Right.
3:08
Marky clearing order inbound. Come on, get up.
3:22
We are surrounded by journalists.
3:22
Hold [music] your position.
3:36
>> [music] >> Strike two.
3:43
[music] >> Activate golden retriever mode.
3:58
>> Mark clearing order inbound. Five.
4:19
I see multiple journalists on the horizon. Stand by.
4:26
>> Friendly IPL in triple glaze. Another one.
4:34
>> Yeah, >> you're watching TVPS.
4:36
Today's Monday, June 29th, 2026.
4:39
We are live from the CVPN Ultradome, the temple of technology, the fortress [applause] of finance, >> the capital of capital. >> Tell you about ramp. com. Time is money. Save both.
4:50
Easy to use corporate cards, bill payments, accounting, and a whole lot more. >> Lot more. >> All in one place. I got to adjust my IM.
4:59
Well, on the front of the Wall Street Journal today, this is how you know this is the whole AI 2027 Washington waking up.
5:08
The AI stories are making it to the front page, the the the world news section, not just the business and finance section more and more.
5:15
So the the very front page of the uh Wall Street Journal uh of course the picture is about the heatwave but the lead the the story with the largest text is about artificial intelligence.
5:28
China resets the AI race with the United States as security models mark gains.
5:34
We're going to get into it.
5:37
Uh this is a fascinating debate because I thought that we'd have a conclusion to the opensource AI debate by now.
5:44
by now. either they would the the frontier would have collapsed and there would be you know perfect commoditization or they would have fallen >> a conclusion it'll it'll just go it's over we're so back it's over >> if you're in open source AI that's exactly how it feels before we get into
6:02
the story completely uh Hill in the chat said did you see the US national design studio open sourced a privacy model we did >> and we got him coming on the show >> today >> in just 45 minutes >> at 11:45 I are going to be >> talking about a first iteration ondevice uh PII redaction model that is far smaller than existing models. >> It's actually it's actually tiny. It's
6:25
>> It's actually it's actually tiny.
6:25
It's 15 megs which is and you can do it in the browser.
6:31
>> And we have Chad Regetti.
6:31
He's coming on to talk about whole lot of quantum mumbo jumbo.
6:37
[laughter] >> We'll see what's going on there.
6:40
>> And Pim's coming back on from General Intuition.
6:42
And we got a bunch more founders coming on.
6:44
Jacob Deep and Brock announcing a $30 million oversubscribed fund with uh tons of TVPN guests already in the portfolio.
6:52
The rest of the portfolio soon to be on the show, I'm sure.
6:56
Uh anyway, uh open source AI.
6:56
So the uh the big the big story is centering around GLM 5. 2 from Z. AI.
7:06
Uh was officially released June 13th.
7:08
So, it's taken a couple weeks for it to really break through to the front page of the Wall Street Journal, but uh there's seen some strong performance on benchmarks, some positive reviews from developers.
7:17
I have a whole review from Tyler we can go through in a little bit, but um we're now entering another round of debates around open-source AI, uh what can the model actually do?
7:26
Is this a threat to national security?
7:28
What are the geopolitical ramifications here?
7:30
And so, I'm sure this will be an ongoing conversation throughout this week, probably next week.
7:36
We have some guests lined up to help contextualize it.
7:37
But uh laying down the facts from the journal, uh security researchers said that a new AI model released this month by China's Zoo AI, also known as Z. A. I.
7:50
can match the latest US models when it comes to finding security bugs.
7:52
A development poised to reset the global tech race and pressure the White House in its overhaul of US AI policy.
7:58
So unlike models from Anthropic or OpenAI, Zepoo's GLM 5. 2 is open weight.
8:05
You can just download it, run it anywhere.
8:08
You don't need to go to an API.
8:10
You don't need to go to a private company and pay them.
8:14
You can run it on your own server, provided you have the electricity and GPUs to do so.
8:18
It is expensive to run as we'll go into, but it is open weight.
8:24
Uh, the Wall Street Journal says that means it can be downloaded, run on hardware operated by anybody and can be modified and used without supervision. Scary stuff.
8:31
Open weight models are ideal for users who want unfettered access to systems they control.
8:36
Uh, but they're also ideal for hackers who want to run them in the shadows.
8:42
>> Unfettered intelligence. >> Unfettered.
8:44
Oh, that's >> completely out of We were completely out of names for new Neolabs.
8:49
>> That's a good Neolab name. Yeah.
8:51
>> Unfettered intelligence. Intelligence is good. Uh GLM 5.
8:53
2 has ranked as one of the top 10 most used a AI models according to data from Open Router, a company that provides access to more than 400 AI models.
9:02
And what a fantastic business.
9:05
Uh Alex Atala over there absolutely cooking at Open Router.
9:10
It's such an exciting way to plug into the AI uh the AI race without actually needing to uh play the play the benchmark game so much.
9:17
uh be be the uh the front door.
9:21
Anyway, uh in some benchmarking tests, according to cyber security company Smrap, GLM 5.
9:25
2 bested Anthropics clawed opus 4.
9:29
8 model, which was released in May.
9:32
When given further instructions, Opus 4. 8 and GLM 5.
9:34
2 can match Mythos in bug binding ability according to researchers.
9:40
So, prior to this launch, uh and there's a chart that we should pull up here about overall AI capability.
9:46
We can talk to Tyler about what this chart actually means, but there was this narrative brewing that open-source AI was slowing down relative to the closed source frontier.
9:54
And I saw a lot of uh American AI fans sort of cheer for this.
10:00
Hey, the we have the capital markets, [screaming] we have the data centers, we have the researchers.
10:06
And so we are able to push the frontier at a different rate.
10:10
And if we're actually growing at a faster rate in America within the closed source labs, that will compound and there will be a stronger takeoff in the American closed source uh AI industry.
10:20
Uh now this chart sort of goes back and forth and there's some debate over it. It's in the newsletter.
10:27
You can go sign up at tbpn. com.
10:29
While we're pulling that up, let me tell you about Codeex.
10:31
Codex is a powerful workspace for getting work done with AI agents.
10:34
Whether you're writing code, analyzing data, creating content, or automating business workflows, Codeex helps you move projects forward from start to finish.
10:42
So this chart, which we can pull up, shows progress from GPT40 to 01, 03 mini, 03, Opus 4, GPT 5, 5. 2, Opus 4. 6, GPT 5. 4, GPT 5. 5.
10:54
Uh, showing a uh, you know, linear trend in this ELO, which is a blend. >> It says GLM 5.
11:02
2 two sounds too much like a gray market peptide [laughter] taking. >> It actually does.
11:07
It does sound a lot like that.
11:09
Uh and and then you can see the red line are the uh are the Chinese models which are also improving uh over time but at a slightly lower rate.
11:18
And so the question was uh are they going to plateau while America's uh progress continues to advance?
11:25
Um and this uh this latest model GLM 5.
11:29
this latest model GLM 5.2 to seems it's very hard to apply it to this particular benchmark because this ELO was can you give us some background Tyler on where this chart came from what this is demonstrating >> yeah so this is by Casey I think it's how you pronounce it the center for AI standards and innovation um they have
11:47
they have this way to calculate like the ELOV model it's basically a uh kind of approximation of a bunch of different benchmarks um some of those like are proprietary like they're not open so it's actually hard to run these >> um also cuz I was basically trying you bench like all the recent models since this was published. It was
12:03
It was >> uh I want to say May 1st.
12:06
>> Yeah, it'd be great to throw 5.
12:06
6 Soul, mytho, and Fable.
12:09
All it would be great to just continue this chart because it's an interesting trend.
12:14
>> So, a lot of those benchmarks aren't actually public.
12:15
So, it's very hard to estimate.
12:17
Um but I I tried I got you can look at like some of the benchmarks that that are public that you can reference.
12:23
You can kind of match them up to previous models. >> Um >> 5.
12:26
2 looks like it it it is like a big step up from the like Chinese trend line, right?
12:33
>> Um but but even then I I think it's it's hard >> like I I I think that the group of benchmarks that were chosen for this ELO like definitely accentuate the the gap between US and Chinese labs.
12:48
>> Um I think there's a bunch of other like groups like uh Epoch AI has done a chart.
12:53
They basically seen a relatively stable gap between closed source and open source models. Yeah.
12:58
>> Since like 2023, like a long time. >> Yeah.
13:00
And and and perhaps at this point uh the the discussion should be more centered around cost per task more than cost per token. >> Yes. Yeah. Definitely.
13:09
Because even like you know new models a lot of times when when they come out like okay maybe the the token price is actually the exact same but the token efficiency is much better.
13:18
Then when you do a lot of these tasks, it's like it's not the the price per token, it's price per, you know, something completed and then you actually see it go.
13:27
>> And there's a lot of test time scaling laws where you can just throw a million dollars of compute at a particular problem and uh all the models do really well at it, but it's completely nonviable for any real enterprise use case and probably not even viable if you're trying to be a nefarious hacker or something.
13:43
>> Yes, most people are feeling like 5.
13:43
2 is very uh token hungry, right?
13:45
So it uses a lot of tokens.
13:48
So maybe it like it definitely is much cheaper than the frontier models.
13:50
It it's >> on a per token basis.
13:54
>> But on the per task basis it might be more expensive. >> Yeah.
13:58
I mean on that that's still it's generally not.
14:00
But on on specific tasks you can get you know if you have low thinking models, low thinking mode on the closed source ones you can see.
14:07
>> Well uh let's revisit uh John Ludig's post from 2024 May 2024.
14:10
This was pre-deseek uh talking about his uh prediction about why the future of foundation models is closed source.
14:18
He got a lot of push back from this because a lot of people like open source models but he laid out a uh a thesis around closed data closed source data flywheels exponential capex intensivity of training.
14:33
Uh and he said open source will have a home wherever smaller less capable and configurable models are needed.
14:41
enterprise workloads, for example.
14:43
But the bulk of the value creation and capture in AI will happen using frontier capabilities.
14:47
The impulse to release open-source models makes sense as a free marketing strategy and as a path to commoditizer compliments, but open source model providers will lose the capital expenditure war as open- source ROI continues to decline.
15:03
And that was the thesis around the time that uh the open source AI discussion was primarily driven by Mark Zuckerberg's work at Meta on the llama family of models.
15:12
Um the idea was that Meta would benefit from attracting talent. It was good marketing.
15:16
It told the story that Meta has an AI story and has AI talent inhouse.
15:22
has AI talent inhouse. um even if they weren't monetizing it and sharing you know a really fast takeoff in ARR around those models it showed that hey they're able to develop these models and and that might help them cut their costs in the long term very interesting that that
15:39
wound up being very different in 2026 looking at the news today which we'll go into about them spending a lot on Gemini there's been reports about them spending a lot with other closed source frontier labs that they should have commoditized with their open source plans Nonetheless, that was the idea with Meta. Uh, but then China sort of woke up
15:56
Uh, but then China sort of woke up and the Deep Seeks and Deep Seeks launch at the start of 2025.
16:01
Uh, and the game theory became way more complicated.
16:04
So, George Hots sort of summed this up nicely.
16:08
He has a take in AI will be massively deflationary, a post from just a few weeks ago as to why China benefits from investing in open source more than American firms.
16:17
He says this explains why Chinese the Chinese are giving the much more moderate resources to train models away for free.
16:24
Uh they love to see deflationary economics in the US.
16:27
It is not it is much less of a service-based economy.
16:32
And so if they can go and give away free tools that deflate the value of the service sector, that is an advantage to the Chinese economy.
16:41
economy. in his formulation uh he says even if you don't regulatory capture the US government nobody is getting a monopoly on AI we don't live in a unipolar world anymore and so he compares what's happening in uh he he he likens what's happening in DC to sort of rearranging deck chairs on the Titanic it's a very uh fun fun piece um but uh
17:04
so we're back to this discussion of what are the consequences and the impacts of open- source models particularly in the United States and there's been this clip that's resurfacing from Dario Amade when he was testifying in front of Congress in 2023 uh and it's now recirculating and it was reposted like he just said it and he did not. So be clear about that. This is So be clear about that.
17:22
This is from 3 years ago but some of his predictions were very preient as of where the frontier is today.
17:29
So he said uh I'm very concerned about where things are going.
17:33
If we talk about two to three years for the frontier models for the biorisks is sort of a bad transcription of what he was saying.
17:41
Um but he's talking about 2025 2026 remember he was saying this in 2023. Um we're there now.
17:49
I think the path that things are going in terms of the scaling of the open source models I think it's going down a very dangerous path.
17:55
And again if the path continues I think we could get to a very dangerous place.
17:59
he was worried about cyber security and bio risks being open- sourced and then uh not having a counterweight to that.
18:07
Now the good news is that uh we've talked to the CEOs of cyber security firms like Crowdstrike and Palo Alto Networks and uh they've been working with Mythos and GPT 5.
18:16
5 Cyber for months now to harden systems from LLM driven attacks and so there's still this gap between closed source and open- source models and that gap allows white hat hackers to implement fixes before black hat hackers have a chance to exploit easy bugs.
18:31
Um there still will be a bigger discussion here though in DC over the next few months as the frontier models roll out and the gap doesn't appear to be widening at the moment.
18:42
So security stances must adjust.
18:44
It's not a a close source is falling behind.
18:47
So it's never going to be an issue.
18:49
uh there will be this gap and how uh the American cyber security industry and eventually the biocurity industry implements changes and fixes before open source catches up or or commoditizes and makes that particular capability widely available is going to continue to be important.
19:05
Uh so let's go over to Tyler's quick review of GLM 5. 2.
19:07
Why don't you take me through your bullet points uh that we shared in the newsletter at tvpn.
19:13
com uh and you can tell us like what is the shape of this model? How are the reviews? >> Yeah.
19:19
So, I I think so far um one of the main things is like people are saying it's oh it's distilled, right?
19:23
This is this has been a big thing with a lot of these open source models, especially the Chinese ones.
19:27
Oh, the only reason that they're good is because they're distilled.
19:29
It's very hard to actually figure out how true this is.
19:34
>> Um but people are, you know, it certainly seems like there there's some uh you know, >> I think aspects of of anthropic models.
19:41
Didn't Anthropic openly accuse Alibaba of distilling >> a number of these these labs? >> Yeah.
19:48
And there's also been a big like professionalization of the gray market where uh uh a whole bunch of different uh sort of individual groups will uh connect a whole bunch of different entities and user accounts and script subscriptions and APIs to then create a front end to like the model that can be served at a very high rate through a VPN most likely.
20:11
Uh what's interesting is that you'd think that if you were going to do a training run, you would just find and replace on the other lab's name before you hit run.
20:22
Is that not something people can do? I don't understand. >> Yeah.
20:27
I mean, it also depends on what you're actually like maybe you're not directly distilling on the API, but you know, you're training on you know, public GitHub, you know, repos and those were all used those were all, you know, made with with with open source models.
20:41
you're kind of like distilling but it's not really count as distilling I don't know >> but so if you are like >> if you're convinced that these are like super distilled the only reason that they're good is is because they're just you know basically taking the closed sourced like labs um >> there's also this weird thing with
20:56
distilling where as more and more of the public internet and GitHub broadly and open-source repos become LLM outputs you if you train on that you are in some ways distilling because an LLM has a quirk like it's not this, it's that in text and you wind up training on a whole bunch of Amazon Kindle books. You're
21:17
You're going to wind up learning it's not this, it's that.
21:21
And the same thing applies for different code conventions in open source repos that have effectively been completely been rewritten by closed source models. >> Yeah.
21:29
And and so so I think it's safe to say that like we've generally seen that um distilled models uh generally will will generalize uh worse, right?
21:36
So you'll see really good benchmark scores.
21:40
>> Maybe they're benchmarks, maybe they're not.
21:42
But even if they're not like directly benchmarks, you still find that they generally >> Yeah.
21:45
They're kind of accidentally benchmaxing. >> Yeah. Yeah.
21:48
>> So you should always So I think initially you should just be a little bit suspicious of these super high benchmark scores. >> Yeah.
21:53
>> Um >> but they lack that big model genqua. >> Yeah.
21:56
And this is like anecdotally reinforced.
21:58
A bunch of people have been saying, you know, for coding, uh these models are really great.
22:01
GLM is it's a very good model, you know, for creative writing or or something like this where you'd imagine it's a bit harder to to kind of benchmax this.
22:09
>> Um, they'll perform a bit worse. >> Yeah.
22:11
I wonder Hey, have have people been testing it with the like TMN square bench?
22:16
Like does it reject that stuff or because it felt like that was something that was uh like widely misunderstood by American audiences that in fact that might not be the biggest deal for the CCP. >> Yeah.
22:29
So I I think you know even if that's true like the model is open source you can kind of just fine tune it to like not that maybe maybe it's a bit harder than that but I think you can kind of get around like that kind of stuff. >> Okay. Yeah.
22:39
So we talked about the token hunger and the API price.
22:41
Um and in general I mean you said I'm not convinced that there's a big market for this class of model especially as frontier models get more efficient.
22:49
And if you look at open router, the most used models are the smallest open-source models presumably being used for specific tasks that need to be repeated over and over again.
23:00
So >> yeah, so I I think like what we've seen is >> Yeah, that's the job.
23:05
>> Um like a marginal IQ point of the models is like extremely expensive.
23:10
>> Um >> frontier models are are getting very expensive. People have to cut back.
23:13
You know, they're token maxing.
23:14
This is like massive bill on their balance sheet, whatever.
23:18
Um I I I think like it it seems like there's there's now basically like two classes of models that that people really use.
23:25
There's like the frontier ones >> and they're they're using coding agents.
23:30
They need the best thing.
23:30
If you're doing cyber like you you just need the best model because you know the risk of of someone hacking you it's so great.
23:37
You just need the best thing you pay whatever it is. >> Yeah.
23:39
>> And then there's the second class which is like these very small uh very fast very cheap models that you can use for these kind of point solution things.
23:47
Maybe um you have some orchestration where using a really big model to to have these like little agents using these very cheap models. >> Yeah.
23:54
>> I think in the middle it's hard to actually figure out what is the the real use case.
23:58
>> Um maybe it's like hobbyists using these coding agents and and they don't want to pay the super expensive tokens of the closed source labs.
24:06
>> Um but generally and you see this on open router where like what are the top models by by token >> uh like usage it's these very small models.
24:14
It's like uh you know >> uh Deepseek Flash.
24:18
>> You're just spamming them for like you know every receipt that goes into ramp gets processed by an LLM at this point.
24:24
Does it need to be a Frontier model telling me that I spent $10 on a coffee? No.
24:30
It can just do standard OCR.
24:33
>> That'd be my preference.
24:35
>> Yeah, you want you want super intelligence overseeing your expenses most likely.
24:38
Um but no, you use the right tool for the job and that's clearly what's happening on >> but also I I think like it is a very good model, right?
24:46
Like we should not fully dismiss >> I I think the the idea that oh the gap is widening.
24:50
We we really don't have to worry about these these open models.
24:51
I think they are like very good. >> Yeah. Yeah.
24:55
>> And maybe if you're super worried about distillation, maybe something changes if if the models are are, you know, kept to these big partners, right?
25:01
Like what we've seen recently with with government coming in.
25:04
But >> um I I think we can't really fully dismiss these these labs. >> Yeah.
25:08
it it throws it throws a little bit of a wrench in the uh like the monetization potential like how long can you monetize a new frontier model that's more tricky and then uh the other one is just like if you're going to uh keep a model behind KYC or behind a an approval for specific companies like the government has been sort of edging towards and moving towards.
25:32
It gets a little bit tricky if all of a sudden you just wait three months and oh, I was waiting to get approved for this one for like GPT7 or whatever, but by the time I the government got back to me, my company got access to GLM6 and it's close enough.
25:53
And so, uh, that in that just throws another wrench that I think the government will have to figure out how it puzzles together with the rest of the strategy.
26:02
uh which has been yeah back and forth as always.
26:04
Anyway, let me tell you about Shopify.
26:07
Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents.
26:15
>> Google, CAPS, Meta's Gemini use as AI demand strains capacity in the financial times.
26:22
Surging appetite for advanced models is turning computing power into the tech industry's scarcest commodity.
26:30
And they have a picture here of a Google Gemini bicycle which looks fantastic.
26:36
>> What does that have to do with with meta though?
26:39
>> I think that was just the best Gemini picture. >> It is hard.
26:44
Otherwise, it's just a picture of a phone screen.
26:45
I mean, you saw in the Z. AI.
26:46
It's just a picture of the app, which is like so boring.
26:50
>> Imagine writing this >> or it's the stock image of the brain with the neurons. That's always good. >> What do you think?
26:55
think I mean this this kind of ad placement >> like on a uh what do you actually call this? It blocks the water. No, no, no.
27:03
That just the the the the part that blocks water from, you know, if if you were to ride flare. >> Yeah.
27:08
Some type of fender thing.
27:11
>> Like a Mansori kit for a city bike. >> Exactly. Exactly.
27:13
But imagine riding that that Gemini bicycle in the rain. Fantastic.
27:18
>> Uh Google has >> That's what it's for.
27:19
So the water doesn't come up and splat you. Interesting. Okay.
27:22
I I never knew what that was for. >> You never knew that. No.
27:24
learned this is >> educational.
27:27
>> This the experience of hosting the show is educational for both of us. >> It is.
27:31
>> Um Google has put limits on Meta's use of its Gemini AI models after the social media giant sought more computing capacity than the rival tech group could provide in the latest evidence of the infrastructure constraints facing even the world's largest AI providers.
27:44
Google told Meta around March that it could not provide all the Gemini capacity the company wanted to purchase.
27:49
according to three people familiar with the matter in a move that has disrupted and delayed some of Meta's internal AI projects.
27:56
So >> I I don't understand how much this is possible. >> Yeah. Yeah.
28:01
So so one >> Google spent $200 billion on capex. >> Okay.
28:06
So so so of course like this this >> around this time token maxing was becoming a thing.
28:12
a lot of every company in the world, at least every tech company in the world's kind of going a little bit crazy from a spending standpoint.
28:21
>> And so, you know, I could see Meta going and like wanting to basically buy a bunch of capacity and then being told like, hey, we we can't fulfill that.
28:30
>> Um, but I'm I'm wondering how much more we should read it like read like is it worth reading?
28:38
I mean it sounds extremely bullish for Google like if they're this tracks with what they talk about on on earnings calls. >> Yeah. Yeah.
28:47
>> Um >> yeah Google Cloud acceleration crazy.
28:50
>> You do have to wonder like could distillation be be part of this this story?
28:56
Is is that could that be a factor here? I have I have no idea. >> I don't know.
29:01
Zero hedge said Meta puts limits on Claude and Kodak fearing distillation the information.
29:06
>> But but but so this story is different.
29:08
This is Meta telling its own employees, don't use Claude and Codeex in certain parts and certain parts of our business because we don't want we don't want to accidentally do distillation is what >> is what Meta is saying.
29:22
So that's that's different.
29:24
I was wondering like is Google thinking like whoa that's a lot of >> you know cool it you know. >> Yeah.
29:31
Uh owing to the restrictions which remain in place as well as broader push to streamline AI costs.
29:35
Meta has encouraged staff to be more efficient with AI tokens.
29:38
Uh several other Google clients have been affected by the restrictions although to a lesser extent.
29:44
Meta has been particularly impacted because of its exceptionally high demand for Google's models. >> Interesting. >> Um very interesting.
29:51
I I would love to see a pie chart like breaking down what all the different ways they're using Gemini. >> Yeah.
30:01
>> In their business because >> Google has not broken out Gemini revenue. >> No. No. >> At at all to date.
30:08
So we have no idea what percentage of their AI revenue is like actually spent on Gemini >> versus other like the Gemini tokens broadly go into AI search overviews.
30:18
So that's a search product probably insane token demand there, right?
30:24
You've seen the chart of like they're in the quintilion or quadrillion tokens uh category.
30:30
And then and then you have um and then you have YouTube now has Gemini plugged in and you can chat with any video and transcribe it.
30:38
That's got to be incredibly tokenheavy.
30:40
Uh and then you have Gemini app users and free users and paid users.
30:46
So there's there's got to be a lot of just Gemini internal usage, but uh it's it's remarkable.
30:53
Yeah, I would love to see that meta pie chart of because I thought that they were spending a ton with Enthropic.
30:57
I thought they were spending a ton with Google, but I also assumed that they would be running a bunch of Llama workloads and a bunch of Muse Spark workloads because those models have performed well at various points in time.
31:10
And if you go into the meta app, you now have access to Muse Spark.
31:13
And if you go into Instagram and you search for something, it pop it populates it with a uh with a llama like llama 4 uh result.
31:24
And so I would imagine that even though that product is not broken through like crazy, I would imagine that uh it's still um generating a lot of tokens just because of the scale of Instagram.
31:36
Like Instagram has a billion, two billion users, something like that. It's huge.
31:41
And so even if it's people sort of you know accidentally winding up in an LLM powered workflow uh it probably is generating a lot of tokens just because of the scale of that of that system. On the topic of meta. Yeah.
31:54
>> Meta shared this morning. >> What they do? >> A new milestone.
31:59
>> It is a mind readader. >> Mind reader.
32:01
>> Non-invasive brain detects decoder research.
32:03
Brain to cordi v2 building on v1 which was published today.
32:06
Nature brain to cordi v2 is the highest performing endto-end pipeline capable of real-time sentence decoding from raw brain signals.
32:16
advances beyond character level performance to decoding words and semantics enabling accuracy for overall communication.
32:24
So if you thought uh you know Instagram was listening to you [laughter] >> if you thought I was listening to your uh you know conversations now you can have a you know new conspiracy uh at home which is that they might be just listening to to your thoughts.
32:43
Do you know do you know the device?
32:45
They they say this is a non-invasive device.
32:46
I just shared an image of this device and I want you to tell me do you consider this non-invasive or invasive?
32:54
[laughter] >> Look at this image of the magneto and salaf and salaf graphi device. >> No, you got to go.
33:01
You need to scroll up a little bit cuz you can't even see the whole thing here. Scroll [laughter] up.
33:09
It's not invasive >> cuz it looks like the device could actually potentially carry on for like a whole half of football.
33:15
>> It really does seem like it's a just put yourself in this in this roomsized device.
33:19
No, of course this will shrink.
33:21
>> I'm giving him I'm giving him credit here. Nonvas noninvasive. >> Okay.
33:25
>> As long as he >> You're You're putting this thing on.
33:28
You're daily driving this thing.
33:31
>> I don't know if I'm ready to daily it.
33:32
[laughter] >> I don't know if I'm ready to daily it.
33:36
Yeah, >> this will be a cool demo.
33:37
Like this this will actually when when when >> when you can just walk in, sit down in a chair and see your thoughts on a screen.
33:47
>> No, we were debating it earlier.
33:47
My buddy Rob Taves been on the show twice.
33:50
Uh dropped five predictions in Forbes recently.
33:52
We can go through them at some point.
33:53
He's going to come on the show.
33:54
But uh four of the five were very very like reasonable.
33:58
you know, Anthropic's going to be bigger and uh, you know, TSMC is going to face more comp more competition.
34:05
And then he predicts that in 2030 telepathy will be commonplace, which is a very aggressive prediction in my in my estimation.
34:15
Um, you know, it's certainly not like a uh like a straight trend line since like, you know, like TSMC has competitors right now.
34:22
The prediction is just that there will be more competition.
34:25
Uh but truthfully, uh telepathy is not really existent outside of like a few demos like this.
34:32
Uh it's not it's not really something where it's like, oh yeah, like 5% of people have the meta ray bands that take pictures.
34:41
So like face cameras are going to be bigger in five years.
34:43
And it's actually only three and a half years until 2030, which is sort of crazy to say, but we are uh getting quickly to the future. To the future.
34:52
Uh never sell your company.
34:55
Should you ever sell your company? David Center says no.
34:57
He says the best founders in the world would never sell their company.
35:00
You could never acquire Elon, Bezos, Zuck, Jobs, Ellison, Jensen, Dell, Paige, and Brin.
35:04
Scott Woo has turned down billions and keep saying no. This is a great clip. Went super viral.
35:11
Uh I don't know where did I lose internet or something? I don't know.
35:15
Anyway, um >> uh Tyler Tyler's app is in shambles.
35:21
>> I don't know about that.
35:21
Uh but uh there's some debate over this because Elon >> that's not my app.
35:26
>> Elon Yeah, this is just [laughter] a Elon did sell two companies.
35:29
He sold Zip 2 and he also sold PayPal.
35:31
Uh and then Jobs sold uh Next back to Apple. Does that count? I don't know.
35:38
He did sell Pixar to to to Disney.
35:40
That that sort of counts.
35:42
Uh and I mean Elon never sells his company.
35:46
He just sold X AI to himself, but I guess that doesn't count.
35:50
Um, but yes, it is it is uh it is a funny thing. Uh, didn't Mark Yeah.
35:56
So, what is what's the do you know the backstory here from Sasha? >> I don't know.
36:01
Uh, uh, Tyler looked it up.
36:03
Apparently, there's a Business Insider report from the time that this happened in 2007.
36:07
How Terry Seml fumbled Yahoo's Facebook deal.
36:10
Uh, how much is Facebook worth?
36:13
5 billion, 10 billion, 15 billion, whatever the number, it's probably a lot more than the 1 billion that Yahoo could have bought it for a year ago.
36:21
As Yahoo continues its soulsearching, here's an unpleasant rendition of Seml's catastrophic decision, courtesy of Wired.
36:28
When Yahoo came calling with a bid of 1 billion in cash, the pressure became too much.
36:34
Zuck relented in July of 2006.
36:37
He was just like 18 months into building the company, something like that.
36:42
verbally agreeing to sell Facebook to Yahoo. He said yes.
36:46
He said he was going to sell Facebook to Yahoo allegedly uh strategically it seemed like a good match.
36:53
Yahoo had hundreds of millions of users uh uh but its foray into social networking was struggling.
36:59
Facebook had cool tools and was looking for a mass audience.
37:03
The timing, however, could not have been worse.
37:05
uh in the days after Zuckerberg agreed to sell, Yahoo announced it was projecting slower sales and earnings growth and that it's that the launch of its new advertising platform would be delayed.
37:15
Its stock price tumbled 22% overnight.
37:17
Terry Siml, Yahoo's CEO at the time, reacted by cutting his offer from 1 billion to 800 million.
37:26
He just took 20% off, but Zuckerberg, who had been warned about Seml's reputation for last minute renegotiations, walked away.
37:32
And that's probably reasonable.
37:34
I mean, if they're cutting the price there, you have to imagine that as it gets papered, you get cut down again.
37:39
Then the earnout, you get cut down again, and all of a sudden, you're walking away with barely anything.
37:44
Uh, but two months later, Seml reissued the original $1 billion bid.
37:49
But by then, Zuckerberg had convinced his board and executive team that Yahoo wasn't a serious partner and that Facebook would be worth more on its own.
37:57
He rejected the offer and became famous as the cocky youngster who turned down $1 billion from Wired. >> Legendary. legendary.
38:03
Um, it's so interesting to imagine the road not traveled there because the the the the dynamic the way Facebook is built with this with the uh [snorts] the as a social network like could it have been successful under Yahoo's stewardship or would it have been uh less exciting, attract less talent, ultimately been disrupted and would they have had the capital and the guts to go by WhatsApp and then also by Instagram.
38:34
Um, you know, to actually maintain the the dominant position in social networking. What do you think?
38:42
>> I think Yahoo should make another offer.
38:44
We were hanging out with Jim, CEO, last week, dear friend of ours, and uh >> I I would like to see I would like to see Yahoo make another bid.
38:55
>> Hey, Meta's trading down. Just keeps going.
38:57
If it continues at this point, it >> goes down to 99.
38:58
99% might be able to pick up >> continues at this trend.
39:03
>> Anyway, let me tell you about MongoDB.
39:05
What's the only thing faster than the AI market?
39:07
Your business on MongoDB. Don't just build AI.
39:09
Own the data platform that powers it. Um, moving on.
39:11
What else is in the news?
39:16
>> Uh, chipmakers are profiting off AI at the expense of just about everyone.
39:21
>> This is on the cover of the business and finance section today.
39:23
Uh we are witnessing an extraordinary transfer of cash from the providers of AI and perhaps one day AI users to memory chip makers. Take us away John.
39:34
>> Yeah the explosive growth in Micron Technologies profit in the latest quarter is extraordinarily good news for its shareholders but it comes at the expense of the artificial intelligence companies to which it sells fast memory chips.
39:46
Micron along with the Korea with Korea's Samsung electronics and sam and skinex are to AI what oil producers are to the airlines makers of an essential input that this year suddenly became much more pricey because there is extremely limited capacity to make the high bandwidth memory that AI needs and it takes years to build production facilities.
40:09
Soaring data center demand simply jacked up prices.
40:12
Micron soaring profits are for its customers. Soaring costs.
40:16
We are witnessing an enormous transfer of cash.
40:19
They said profit shift of this scale are rare events and investors should be paying attention to where the money is coming from, where it's being spent, and how long it will keep flowing.
40:29
In the quarter ended May 28th, Micron increased prices for DRAM chips more than 60% on the previous three months while increasing shipments by a low singledigit percentage.
40:39
It said last week prices for NAND flash memory also used in data centers jumped more than 80%.
40:46
Usually memory doesn't matter that much but for Micron customers paid $18 billion more and that was just in the quarter prices quadrupled in a year and it's hurting outside AI 2.
40:56
Apple last week raised prices for MacBooks more than 15% closer to home uh closer to home for me the memory I bought on Amazon.
41:08
com a year ago to build a super quiet computer. I hate fan noise.
41:10
Good good color commentary here.
41:13
Uh has tripled in price >> and now costs more than the CPU.
41:16
For an industry in which prices usually drop every year, it's a huge turnaround in consumer electronics.
41:23
Passing on higher prices helps limit demand for chips just as higher oil prices reduce consumption.
41:30
But the AI companies aren't passing on higher prices because they are able to throw money at supply problems.
41:34
The problem in AI is that the end users aren't covering the cost of the service with big losses being recorded by AI model producers.
41:41
Everything is still priced to bring in new customers yet not yet to make money.
41:46
So higher input costs create a nasty problem.
41:49
Either losses will either be bigger or higher prices will be needed putting off potential customers.
41:55
And you can see the price of Micron's stock price has been through the roof as the company uh joins the1 trillion dollar club and becomes the first trillion dollar company in uh headquartered in Boise, Idaho.
42:07
And uh the and Idaho got a trillion dollar company before New York, I believe, and also before Florida and Austin, maybe something like that. Um it's rare. It's rare.
42:22
>> Mostly in mostly on the West Coast.
42:22
Um anyway, uh there's a whole there's a whole bunch of bull cases for Micron still.
42:29
The stock could double from here, says Barons.
42:30
I love Adam Lavine and Barons sharing the bull case.
42:32
Uh we can >> Tyler, how many trillion dollar companies are there in Europe? Out of curiosity.
42:41
>> I'm going to go with zero. >> That's true. That is uh true. You are correct.
42:48
>> The other ASML could get there >> maybe >> sitting at uh around 700. Uh, wait. What about Eli Liy? Or no, uh, Novo.
42:54
Novo was a trillion, right?
42:56
Oh, did it ever did it ever touch a trillion? I don't think so. Right. It was real close.
43:01
>> It's a humble 165 million. >> Brutal.
43:04
Wait, but >> you're thinking of Eli Liy. >> Oh, Eli Liy hit. Yeah, rough. Very, very rough.
43:12
Uh, Comcast is planning to split up the company.
43:15
Competition is escalating.
43:17
>> Eli Liy, the the Indiana company.
43:17
John, >> is it from Indiana? Yeah. >> Okay.
43:23
>> Former Indiana startup. >> Okay. I like it. I like it.
43:25
Uh NBC, Universal, and Sky will separate the company's connectivity business from its film, theme park, and streaming operations. Oh, yeah. Universal Studios.
43:37
Comcast is up on the news.
43:37
Uh Comcast plans to separate its media and connectivity business.
43:42
>> Who's building the Anderoll of theme parks?
43:44
It does seem like a >> could there not be an opportunity to create a a net new theme park business with with modern a modern technology stack. >> It's very expensive.
43:57
Everything needs to be like the modern technology stack in parks is expensive.
44:02
You don't believe in >> in the cap the theme park capital markets? >> I don't know.
44:07
I I I I I I I know I've known people that have worked on theme parks at Disney and uh it's tricky because you you have to amortize a ride over like 20 years and so you'll go >> seems like an absolutely brutal business. >> Yeah.
44:22
>> That is probably harder today because >> I mean think about you know >> Yeah.
44:28
At the time that a lot of these parks were built, like you didn't have like infinite online entertainment for every single sub niche. Yes. Instantly available.
44:39
>> I mean, there's a whole bunch of trend pieces right now about how IRL experiences are seeing higher than ever pricing in the face of you could just watch the Knicks game on Tik Tok highlights, but people still forked over $5,000 to go see the game.
44:55
And so, you know, you have that like barbell strategy where Thrive is buying a stake in the San Francisco Giants, a baseball team that should face >> exploring uh the the NBA team. Yeah. To to Vegas.
45:09
Uh but at the same time at the same time there is enough stat that came out >> this morning or maybe yesterday that uh >> there's more sports betting volume than all sales of uh movie tickets, theaters, theme parks, and like a couple other these IRL categories >> is up or down?
45:29
>> Uh less lower >> less like and and and the the stat was like volume. >> Yeah. Yeah.
45:35
Uh, and so it's not exactly like a a proxy for like revenue, but still >> meaningful.
45:45
>> Theme park vertically integrated Tweety Bird tattoos.
45:48
[snorts] >> Tweety Bird tattoo parlor right on site. >> That's a big thing. >> I like it. >> Six Flags. No, I don't I like it.
45:58
>> Uh, Six Flags never really got the same like cultural power that Disneyland did.
46:04
There's something about the flywheel that Walt Disney laid out that does seem very very important.
46:08
And so, how do you start that?
46:10
It's not just, oh, you know, tech enabled theme park that's not going to draw people in.
46:14
You need to have uh like IP around it. >> Brain rot theme park.
46:19
>> There's something about I mean, we've read we've read stories about like the the the Disneyland fan that's that saves up every year and spends so much money at the park.
46:28
And I think that's probably the lifeblood of that business.
46:30
And that doesn't happen without building a whole cinematic universe around every single ride.
46:36
And that just takes so much time.
46:39
And you can't like this is this goes back to the question of like Netflix is enduring IP.
46:43
Like they don't they they they haven't been able to like even though it's been 20 years of like I mean I don't know when they started producing their own content but um it's been 20 years for that business at least and they haven't really developed like their own franchise that lives in the same world as Batman.
46:58
Well, I'd push back and say the Narcos. >> Narcos.
47:03
You want to go to the Narcos theme park?
47:05
I I was talking about this with somebody once uh talking about like uh HBO, like why don't they have a theme park?
47:10
And he was like, what are you going to do?
47:11
Take your kids to a brothel in Game of Thrones land?
47:13
Like no, it doesn't make any sense.
47:16
Like uh anything like it needs to be uniquely uh general audience.
47:22
like you can't have R-rated a you can't have a content backbone that's R-rated because theme parks will always attract families and kids and so anyone you can't have um you can't have any theme park that's built around um an R-rated like IP library and so that just narrows it down even further.
47:43
>> Well, all of America's basically turned into a theme park for European soccer fans.
47:48
Oh yeah, >> in the journal.
47:49
European soccer fans marvel at the splendor of America's suburbs.
47:55
>> You've been many of these reels serve to me.
47:57
>> Uh, Dutch fans in Missouri see a nation that is risky and expensive but vast and bountiful.
48:04
>> Everything is three times the size.
48:06
>> You've been have you've been seeing some of these people in real life, right?
48:10
>> I don't I don't know if I've seen any of them.
48:12
I did go out to lunch uh like a week ago and it seemed crowded, but I was unclear if that was just uh local residents going out to watch the games or actual tourists coming to town to watch.
48:24
>> Gabe, in the X chat, I think Ferrari has a roller coaster in the Middle East.
48:26
Do they have a whole Ferrari theme park in Abu Dhabi?
48:32
>> Cuz that's not R-rated.
48:32
You can take your kids to Ferrari theme park. Um, yeah.
48:37
I was uh I I was in Abu Dhabi and I and I I was driving by it and I was like >> Yeah.
48:46
I was just thinking of like if you wanted to spend >> Yeah.
48:49
>> a day, you know, uh getting the Ferrari experience.
48:53
Like you could just go to the track. >> Yeah.
48:56
>> Or you could just rent a Ferrari. So I don't know.
48:58
Uh >> yeah, but you don't need to go to Six Flags to get the Batman experience.
49:01
So you can just go out in the middle of the night [laughter] and arrest a criminal.
49:08
>> Just become a vigilante.
49:12
>> Uh I I saw another report that apparently [laughter] there's like a an individual who's being like the Batman of Mexico. Do you guys see this? This is very funny.
49:19
And and the so the guy went out and found criminals.
49:23
>> Develop says Yas Island.
49:23
They literally named an island >> Yas. >> Yas. >> Weird. >> No. Okay. I don't know.
49:33
>> Um, >> anyway, Dutch soccer fans are having fun visiting America.
49:36
Frank Evering, he hadn't even heard of Kansas City.
49:41
[snorts] But when the Dutch soccer fanatic saw his team would be playing along the border of Missouri, and Kansas, he made a detour in his worldwide road trip.
49:48
Everyone got in his camper van and drove south from Toronto, making stops in Detroit, Chicago, and Indianapolis.
49:54
Along the way, he and other European fans who flocked to Kansas City for the World Cup beheld the fruits of the American economy from a vantage point few foreign tourists typically see.
50:05
Suburban s sup super superstores, hulking plates of food, quiet streets.
50:10
He marveled at the sprawling houses and a contrast from the tightly packed homes of the Netherlands.
50:16
I did notice this when we were in France.
50:18
The food portions were way too small for me. It was brutal. Uh it's spacious.
50:22
He said, "You go here for your shopping and there for your dentist." People are so rich here.
50:27
I think that's why they can be so nice.
50:29
What an ultimate white pill in America.
50:32
In America, everyone's like, "And we're so divided and everyone hates each other and it's terrible and the economy is about to fall apart."
50:38
And then one one European tourist comes like so nice.
50:43
>> Something about the grass.
50:43
The grass is always greener, right?
50:45
The grass is always greener on whatever side I'm on.
50:49
That's what I like to say.
50:49
The throngs of Dutch fans that flooded Kansas City and its suburbs this past week got a taste of day-to-day life in the United States, reigniting the longrunning transatlantic debate.
51:01
Who lives better, Americans or Europeans?
51:02
The Europeans had plenty of thoughts on American culture.
51:06
We are a bit shocked about the food you're eating.
51:08
The Dutch national team superfan Sandra Tate said.
51:10
Uh fans also boked at the size of Costco and the vastness of the highways.
51:17
In recent days, social media has been filled with videos of Europeans gawking at the staples of suburban life.
51:23
A two-car garage, a walk-in closet, a second refrigerator.
51:28
One Brit went viral for trying Chick-fil-A for the first time.
51:32
That was absolutely banging, he said.
51:35
And another he toured the inside of an American fire station.
51:38
>> The way the way that they they look they experienced a Chick-fil-A was me seeing the Renault Twizzy. >> Yeah.
51:44
>> I was just like, "Wow, >> this is unbelievable.
51:47
They made the perfect car. >> Yeah. So small. So small.
51:48
Uh and and it's the way they think about our our fire trucks, which are massive. This is nuts, honestly.
51:57
They said >> Tyler, while we wait for our first guest, do you know anything about Bosnia's World Cup team?
52:06
>> We The United States is facing them on Wednesday and Do you have a [clears throat] stat breakdown or anything?
52:11
We be very careful with what you said because I saw that there was a news reporter who faced fierce backlash for uh really calling Bosnia out and saying like I don't know where it is on a map.
52:22
And the funny thing was that it was delivered in like the typical newscaster like and I'm here reporting on the ground and tonight uh but Bosnia will be playing and then she just like transitions into color commentary giving hot takes about how irrelevant Bosnia is in her mind and the Bosnians did not enjoy her critique of their country. >> Pure disrespect.
52:43
>> Anyway, uh there's a little golf cart.
52:45
We got to talk about this at some point, but uh there's a new there's a new car. It's like a Twizzy. You're going to love it. It's >> close. It's no Twizzy for you.
52:53
>> Let's bring in our first guest.
52:54
>> Anyway, let's bring in our first guest >> from the National Design Studio.
52:56
[music] >> From the National Design Studio.
52:57
Welcome to the show, gentlemen. How are you doing?
53:02
>> Thank you so much for coming on the show.
53:05
Uh, please start with an introduction of yourselves, the the the company, and then the announcement today.
53:11
>> Uh, my name is Edward Corstein.
53:11
I run engineering at National Design Studio.
53:15
It's technically not a company.
53:15
It's a government organization.
53:17
>> Oh yeah, that's right. Sorry. >> And I'm Tiger.
53:19
I'm one of the engineers at National Design Studio. >> Okay.
53:22
And today the launch, take us through it.
53:25
>> We're launching Rampart.
53:25
It's a local first privacy model that puts people back in control of the data that they share with AI.
53:34
>> Um we were we were just like, you know, kind of building a chatbot for fun. >> Yeah.
53:38
>> And we were upset that uh none of the frontier models will actually fit in a browser.
53:42
So you cannot do PII removal in the browser.
53:45
Uh which is, you know, pretty damn important for our use case.
53:49
uh you just have to like you know trust that the server is actually removing the information uh and not lying to you.
53:52
So we're like okay well what if what if it was just all on device like personal data never had to leave your device uh it was just you know secure by default.
54:04
>> Okay so open source the weights are on uh hugging face runs in the browser under 15 megs.
54:09
uh a technical user could go right now download the model from hugging face vibe code their own Chrome plugin and have it be running uh however they want but how do you see this actually rolling out?
54:21
Do you want the government to inter to to implement this in various places?
54:26
Do you want companies to are are is it sort of like open up the primordial soup of ideas and see where it goes or do you have do you have like a rollout strategy that you are advocating for?
54:39
Um well the reason why we open sourced it is because we do want companies to use it and we want people to use it and we want people to make it better.
54:44
So we want uh vibecoded chrome plugins. >> Sure.
54:49
>> We want we want you know vibe coded tragedy extensions like what whatever value is derived from the product.
54:53
You know this is just like a a total side quest for us.
54:57
We just want to build software that's that's helpful for the American people.
55:01
We've already launched a series of products then like Trump X has got 15 million users.
55:04
It's saved over $500 million in in drug costs and >> you know we rethought the UX there.
55:09
So we're you know basically across everything we're working on we're just trying to find the first principles best approach for users. >> Yeah.
55:19
>> And this just came as the derivative of that.
55:21
Uh we're not MBAL you know researchers or engineers.
55:24
We're just like you know we should just do this.
55:28
>> You created PII super intelligence.
55:30
[laughter] That's what people are that's what people are saying online.
55:33
>> Basically >> it's like tiny intelligence.
55:34
It's like the It's by far the smallest model.
55:36
Like the other ones are like at least 50 megabytes. This is 15. >> Yeah.
55:40
So, did you like did you to what degree did you build on the shoulders of giants?
55:46
Is this some uh pruned and distilled and fine-tuned open-source model?
55:51
Is this something where it was easier to just start from scratch but use architectures that are more uh prevalent and wellestablished?
55:57
Like how did you actually go about training this model?
56:02
We tried 72 different base models. >> Whoa.
56:06
>> Um and you know put them through a training set. We ended up on miniLM.
56:08
So we definitely are standing on the shoulders of giants here. >> Yeah. >> Yeah. Yeah.
56:13
We we we started taking a look at the open AI privacy filter that just got released recently. >> Yeah.
56:18
>> Um we're trying to figure out is there a way we can you know just quantize it?
56:21
Can we maybe remove some of the parameter?
56:23
Like what what can we do here to try to make use of like the the state-of-the-art model and um you know we tried a lot of things.
56:28
we just could not get it to fit into, you know, we want this to work on like legacy devices, um, >> on an old Android phone, for example, or, you know, an older iOS device.
56:36
Um, and it it just would not get small enough and still make any intelligent sense to try to actually run it.
56:42
So, yeah, we we ended up uh essentially uh it's technically a fine tune, but we were we we we trained many and basically made it do exactly what we wanted to do.
56:54
>> Can you help me understand use cases a little bit more?
56:56
because I feel like most of the time when I'm transmitting a document to a prescription website, RX or uh a financial institution, uh the PII is like the the potentially the only important part.
57:09
They're often sending me a blank form and asking me to put my PII in there.
57:14
What is the inverse scenario where I want to redact my information but I still need to transfer something because in most cases that would just be the template or something in my in my estimation.
57:28
>> Yeah, the the flags are set at compile time.
57:30
So you can decide like for our use case it's really important that we have this data or that we don't have this data.
57:36
>> And so we hand all the customization back to whoever wants to use the library.
57:40
library. uh the model just says oh you know this is a phone number this is a name uh this is a surname etc etc and then ultimately it's it's you know whatever you want to do with the the model you can just do it um fundamentally what we were looking at was there are a lot of cases where
57:57
people will ask a question pertaining to a document of like okay for example with the template how do I fill out said template um because you know the government is pretty bad with forms there's like way too many forms nobody knows all they mean like you got to pay people to do government forums. >> Yeah. >> Yeah.
58:12
>> Um so like that that was a use case we had in mind.
58:14
PI is like not not super helpful for that.
58:18
>> Um and it's also kind of like the breaking point.
58:19
It's where uh you know the product will lose trust.
58:21
So we're like okay two birds one stone. Let's build this thing. >> That makes sense.
58:26
>> Uh how do you guys how do you guys think about side quests at the National Design Studio in general?
58:31
Like I imagine every single day there's opportunities that come up and you guys are in a unique situation where >> Yeah. sort of design.
58:40
they have a mandate, but at the same time, there's so many different places that >> the government uh you know, you know, interacts with people's lives. I'm I'm very curious.
58:51
>> It's pretty hard to pick what to work on because there's a lot of exciting things.
58:54
There's like everything is huge scale.
58:56
Everything could be way better.
58:58
Um maybe not everything, but a lot of things.
59:00
Um so there's like a huge calling for side quests.
59:03
Uh but we we we just tried to keep everything in line with our vision which is like we want to make uh the American digital experience better >> and then we we've kind of chosen a track to get there and on the way we built this model and on the way we built Trumper X but we're excited to see uh to see how it >> develops from here and back on the show.
59:25
Uh yeah, like diving more into that, do you have a reference point in in tech people might ship, you know, they might think in quarters, financial quarters, three month cycles.
59:35
They also might think about a two pizza team, which I think is like 10 people.
59:39
Uh do you have an idea of of where the sweet spot is from what you've experimented on how many people do you want to bring into a project and then how long do you want to spend there so you don't get stuck for a decade because you might not have a decade.
59:53
Yeah, I mean there's definitely a lot of a lot of places to get stuck because the the visibility is super low in a lot of these projects and you don't know how broken they are until you're like you're really in it. >> Sure.
1:00:03
>> Um that that you know being able to determine that in advance is like is definitely you know AGI level. >> Yeah.
1:00:09
[laughter] >> We have a really great team. We're we're very fluid.
1:00:13
We're we're constantly trading responsibilities back and forth.
1:00:16
You know someone might be you know better at doing you know one part of the tech stack than somebody else but they're on a different project.
1:00:21
We'll just borrow them for a day.
1:00:22
It's or even for an hour.
1:00:24
It's it we we share a lot of responsibility at the studio.
1:00:27
There's this is also definitely the only place in the government where people work seven days a week >> consumed, you know, on Red Bulls.
1:00:34
Um I I think the ideal amount of people per project if they're if they work super hard is two.
1:00:42
>> Like what one design person, one engineer, and they both have like, you know, full scope and then they're able to call on people as necessary. >> Yeah. Yeah.
1:00:51
two with the caveat of you're you're calling in your co-workers say, "Hey, can >> yeah, >> take a look at this over my shoulder quite frequently."
1:00:58
>> Yeah, that makes sense.
1:00:59
>> What What's your what's your guys's pitch to talent that that uh that you might want to recruit into the National Design Studio, I imagine.
1:01:06
Uh uh lots of people that would join could get a go get a blank check from a venture fund or could go work at some of the best companies.
1:01:16
>> Everyone that has joints, you know, that's the case for turn that off. Yes.
1:01:21
Um it's definitely more for people who are super missionoriented.
1:01:24
Um you know who who the hell like what what great engineer wants to come work in the government?
1:01:30
You know the answer is typically nobody unless it's you know like the IC where there's really interesting problems to solve.
1:01:35
Um so I I think that we have like the a super golden opportunity.
1:01:41
Um at least the way I evaluate problems I try to see how big the problem is in terms of like how many people will use it.
1:01:47
uh the delta between what exists versus what our team can do and how fast we can do it.
1:01:51
When you look across those three matrices, it's like a a home run place to work.
1:01:56
So, I think that is its own natural kind of calling card for the right kind of talent that we need for the studio. >> Awesome. >> Complex, too.
1:02:06
It's it's also a huge benefit. It's pretty sick. >> Awesome. >> Where is that?
1:02:09
Is that where you guys are right now?
1:02:13
>> We're not there right now, but we're about to be there. Yeah. >> Awesome. All right. All right. Well, thank you so much.
1:02:17
Congratulations on on >> very fun project >> and we'll talk to you soon.
1:02:21
>> Great to meet you guys.
1:02:22
>> Have a good rest of the day. >> Goodbye. >> Thank you very much. >> Cheers.
1:02:25
>> Let me tell you about Crowd Strike. Your business is AI.
1:02:26
Their business is securing it.
1:02:28
Crowd Strike secures AI and stops breaches.
1:02:31
So, uh, Apple and Audi alumni just unveiled a $25,000 openair electric neighborhood vehicle.
1:02:39
It's called the Amble 1 and it's a streetle legal EV built for short local trips.
1:02:45
No doors, fewer screens, modular design inspired by the 1960s Lunar Rover.
1:02:51
Goes 40 m an hour with 60 mi of range. Weighs under 1,000 lbs. Takes 5 hours to charge.
1:03:01
Rear seats fold flat for cargo, surfboards, or gear.
1:03:04
Built-in mounts let you add baskets, straps, mirrors, and cargo accessories.
1:03:09
already has 500 vehicles committed. >> I love it. >> You love it? >> I love it. I think it's great.
1:03:14
I've I've >> give it a give it a Jordy score.
1:03:18
>> A Jordy score >> daily weekend.
1:03:20
You know, the Doug score out of a 100. What are you doing?
1:03:23
>> So, I mean, I just went through this whole crazy search for basically this exact vehicle. Didn't didn't find it.
1:03:27
I don't like the >> aesthetics of golf carts.
1:03:32
I've driven a lot of carts in a commercial at a job in college. I've owned a golf cart.
1:03:42
Uh I it's in my experience, it's impossible to feel >> cool while driving a golf cart.
1:03:48
So, I wanted something like a golf cart >> uh that was more like not, you know, I'm not golfing.
1:03:56
Um so, I wanted some like >> little bit of utility, wanted to be fun, etc.
1:04:02
Uh, I landed on a uh a Can-Am HD11. >> Mhm. >> Uh, you know, a UTV. Uh, it's gas powered.
1:04:11
It's uh it's quite fun, but the gas element is actually kind of annoying even as as a as an ICE >> uh, you know, defender that is the internal combustion engine.
1:04:22
Um but uh but I think [laughter] but I think this is uh no I think this is I think this is fantastic and I think that I I think I saw somewhere uh that they're going to focus on more commercial opportunities.
1:04:38
So going to hotels all over the world and >> that's what Justin says here says this little golf cart is going to be huge for hospitality all electric $25,000.
1:04:47
How does that comp against if you're a business and is it really going to move the needle on the customer experience to have this versus just a golf cart?
1:04:55
Can you get a fleet of golf carts for a discount? What is it >> typical?
1:05:00
Like a golf cart is going to come in at like 13 half price >> is grand.
1:05:06
So I mean and it depends.
1:05:09
There's commercial golf carts, maybe you get bulk deals, something like that.
1:05:10
But no, I think this is going to be great.
1:05:13
I think it's going to be a nice amenity on hotel properties around the world.
1:05:16
Ryan Deini says uh he thinks it'll be a hit in hospitality since Moch's Moch caps sales at 500 units a year. I did not know that. That's interesting.
1:05:29
>> Um but uh I think this is going to be a hit.
1:05:31
Meyers Mans uh I I much I still much prefer the sort of aesthetics of the Meyers MS, you know, the sort of more like dune buggy style.
1:05:40
they're coming out with an EV that I'm very excited about.
1:05:44
But I think this is great.
1:05:46
I'm excited to have uh more people building cars for recreation.
1:05:53
>> Um and uh I talked to Riley Brennan who uh is a GP over at Trucks VC.
1:05:57
They just invest in like automotive uh startups.
1:06:02
Um and so we're working to get the Amble team on the show ASAP hopefully this week. Very fun.
1:06:12
H there's a good quote from Roger Eert, the famous movie reviewer that we got to share.
1:06:18
Oh, the team loves Robert Roger Eert from Cisco and Eert back in the day.
1:06:22
Uh Anime Outsider says, "I don't care what he thinks about video games.
1:06:27
Roger Eert had the ultimate red pill on nerd culture as a whole.
1:06:29
This basically describes every fandom on Earth, and once you see it, you can never unsee it."
1:06:34
He says, "A lot of fans are basically fans of fandom itself. It's all about them.
1:06:40
They have mastered the Star Wars or Star Trek universes or whatever, but they're objects of veneration are use are useful mainly as a backdrop to their own devotion.
1:06:48
Anyone who would camp out in a tent on the sidewalk for weeks in order to be first in line for a movie is more into camping on sidewalks than movies.
1:06:56
Extreme fandom may serve as a security blanket for the socially inept who use its extreme structure as a substitute for social skills.
1:07:06
If you are a Luke Skywalker and she is a princess Leia, you already know what to say to each other, which is so much safer than having to ad limit.
1:07:12
Your Spanish obsession is your beard.
1:07:15
If you if you know absolutely all the trivia about your cubby hole of pop culture, it saves you from having to know anything about anything else.
1:07:23
That's why it's excruciatingly boring to talk to such people.
1:07:28
They're always asking you questions they know the answer to.
1:07:30
What a funny >> It's like you and your Apple Vision Pro fandom.
1:07:36
[laughter] >> We're always just having a normal conversation and John will say, "Uh, >> yeah, this would be better if we were in the dino experience."
1:07:46
>> Is [laughter] that true?
1:07:46
The uh the I'm not that much of the dino experience.
1:07:52
Anyway, let's bring in Chad Regetti from Regetti Computing and Sagal. Chad, how are you doing? >> I'm I'm doing great. How are you guys doing?
1:08:01
>> We're doing fantastic.
1:08:01
Thank you so much for taking the time to come chat with us.
1:08:05
Um, I would love to start a little bit with your your background and your journey.
1:08:08
Of course, we're going to talk about the company today, but if you could give us a little bit of an overview of your journey in Silicon Valley.
1:08:14
I think uh that might be informative.
1:08:16
There's a lot to talk about there, and of course, it relates to what you're doing today. >> You bet.
1:08:21
Yeah, great to be here guys.
1:08:21
Uh I I started uh I I got interested in quantum computing when I was a senior in college and did a PhD in this field.
1:08:28
Uh and spent about three years at IBM research in the early days.
1:08:32
Uh you know helping build up the quantum computing team there and then started my own company that was Regetti Computing in 2014.
1:08:41
>> Uh I was introduced to Sam Alman and uh you know he said we had coffee and he said hey well have you you should do YC and I said what's YC?
1:08:49
And uh and so he explained to me what Y Combinator was and that was the first batch after Sam had taken over YC in 2014.
1:08:56
And he brought in a bunch of uh hard tech companies into Y Combinator for the first time.
1:09:05
>> And so I got to be a part of this incredible group of companies in including Helon, uh >> Ollo, which is now public, GO Bowworks, >> GKO Bowworks. Yeah.
1:09:14
>> Uh yeah, Boom was a couple batches after me, but there was this cohort summer.
1:09:18
But yeah, so any uh it was a fantastic experience.
1:09:21
Ended up running Regetti for about 10 years.
1:09:23
We took it public in in early 2022 uh through a spa transaction.
1:09:27
They were the third quantum company I think to go public.
1:09:30
And so that was an incredible journey.
1:09:32
Uh and you know, so I've been in quantum computing, I usually say my entire adult life and in Silicon Valley for a big part of that.
1:09:41
>> But it's just a really fascinating mix and there there's incredible people working in this area.
1:09:44
There's incredible technology that's being developed and it's gonna uh it's going to change change the rel relationship between artificial intelligence and computing infrastructure and that's we're working on at Sigle Tree. >> Yeah.
1:09:56
uh the journey of going public, all the market girrations.
1:09:58
Is being a public company less predictable than venture and being private because there's still the whims of the private market whether you're in the hot category that year and venture investors are scrambling to get uh you know their position built up in a particular category.
1:10:17
But the public markets seem like even harder to read on because you have retail investors and uh the stocks up and down and and things can repric on a minute-to-minute basis.
1:10:28
What was it like psychologically transitioning from private company to public company?
1:10:35
>> I think either can work and there's a right answer for different companies and you got to ask yourself the question what you're trying to achieve.
1:10:41
>> Is it liquidity for your early investors?
1:10:43
Is it a primarily a capital raising activity? >> Sure.
1:10:47
>> Sure. uh is it to provide you know have have liquidity for your early employees for example with some companies where you've got a 10-year exercise window for your options and uh you know zooming out in in the regetti kind of uh taking public journey that was a point in Silicon Valley when quantum computing was growing in in commercial maturation
1:11:05
and the technology was maturing but a lot of the capital in the markets at that point had migrated for deep tech companies particularly just wasn't available in the private markets so when you look at 2020 into 2022, most of that capital was actually sitting, you know, a lot of it was sitting in spack trusts on the public markets and they were and those spaxs were hungry to cut a deal. >> And so a lot of companies ended up going
1:11:25
>> And so a lot of companies ended up going public during this wave simply because the founders, the executive teams were making the decision that that gave them the best chance of capitalizing the business going forward.
1:11:34
>> Uh and I I think there's a right answer for different things.
1:11:37
And now in the past past month or so, Quantinum has gone public via IPO, a tremendous company that's made great progress.
1:11:43
uh and so the quantum you know the public markets for quantum computing have reached a point of maturity.
1:11:50
There's analysts that deeply understand the technology that are writing about and covering different companies.
1:11:55
Uh it's a you know it's a very very interesting marketplace and then uh in terms of what it's like and the decisions that different companies have to make.
1:12:04
I think the key thing is to take a long-term perspective on what you're trying to accomplish and what kind of business are you trying to build?
1:12:10
What kind of cap table do you want to build?
1:12:12
and what strategy best suits you know is best going to help you help you achieve that. >> Yeah.
1:12:17
>> What kind of uh feedback did you get in the early days around naming the company after yourself?
1:12:21
I've been surprised at more >> uh there's so many generic names in the startup world now that's like the blank company of San Francisco or things things like that or you know all the neolabs have like the same sounding names.
1:12:36
names. be like advanced super intelligence and then >> there was a big boom in like Lys like friendly bitly musically there were tons of companies that were for a while but >> and I only know one other I can only think of one other company Chris Amdon's
1:12:51
company has Amazon heavy industries um but I'm sure I'm sure people thought you were a little crazy back then >> uh well quantum was a different thing back then look I think there there's two quantum companies that don't have a Q in their name and I I started both of them. Uh one is Regetti and the other is
1:13:08
Uh one is Regetti and the other is Sigleree which is what you know what I'm focused on. >> Yeah.
1:13:14
>> And uh but I will tell you when you think about you know advice for founders when you think about naming something and advice is worth what you pay for it.
1:13:23
Um, but think of a name that can become iconic and if you that that means it's got to sound very fresh and new and different.
1:13:30
And if every other corn company has a Q in it, maybe you try avoiding that.
1:13:34
Uh, and that's what led me to Sigleree.
1:13:36
Sigree, I I I love this name.
1:13:39
It's from a Patrick Rothfus novel.
1:13:42
>> Uh, and he was an American writer.
1:13:42
He wrote this incredible novel called Name of the Wind that came out in mid 2000 2010 or so.
1:13:48
Um anyway, so Sigree is uh we're building quantum accelerated uh uh quantum accelerated AI servers for the data center to bring quantum technologies uh directly into the data center to act as a co-processor for the GPU or XPU pods that have become the unit of compute and AI infrastructure today.
1:14:09
And uh we're based in Ann Arbor in San Francisco.
1:14:12
Our hardware development is here in Ann Arbor, Michigan.
1:14:15
Uh where it is hot and humid today.
1:14:17
and [clears throat] um and our AI research team is right there in downtown San Francisco.
1:14:23
>> So what actually needs to happen?
1:14:23
What is the path to you know I would imagine like cheaper tokens like is that the pitch like one day the tokens will be cheaper and we need to do X Y and Z to get there. Um what's X Y and Z?
1:14:40
uh you need to well first of all quantum hardware is going to address a lot of different computational challenges today right so quantum computers were able to solve problems that are uh impossible they're very challenging to solve with any form of classical computing no matter what scale it reaches uh so at
1:14:55
Sigler we're focused on applying that capability specifically to some of the computational challenges in AI to reduce the power and reduce the cost associated with training and deploying these models at at very large scale uh what needs to happen to get there well you have to
1:15:08
build a quantum computer that meets the specific requirements for AI workloads and uh the strategy that we're taking at SIGLY is we are very focused on deeply understanding what those challenges are what needs to happen inside the data center to reduce the uh to bring these algorithms that can have a different
1:15:25
kind of scaling complexity class than classical algorithms for AI training and inference um and then understanding what kind of quantum hardware is needed to run those and what we found is there's a set of requirements that you need to meet that uh probably are never going to be met by single modality hardware. What
1:15:38
What do I what do I mean by that?
1:15:40
In quantum computing and quantum hardware, there's different kinds of cubit technologies that you can use to uh to instantiate the cubits.
1:15:49
So there's supering cubits.
1:15:50
That's what I did my PhD in and what my first company was based on.
1:15:52
That's what IBM is focused on and largely Google has been focused on.
1:15:56
But there's also trapped ions.
1:15:58
Quantum and INQ are doing trapped ions and a long list of other companies. There's photonics.
1:16:02
There's now neutral atoms.
1:16:04
There's spin cubits and semiconductors.
1:16:06
There's all these different hardware substrates that people are using to pursue and to build quantum computers based on those.
1:16:11
And what we're doing at Sigle is stepping up a layer and saying from a computer architecture perspective, uh, you know, modern computers aren't built out of one aren't built out of one physical kind of bit.
1:16:25
>> There's not just [clears throat] one transistor type that makes up these computers that we're using today or the computers that are used to train large scale models and deploy them.
1:16:30
There's a plethora of different physical technologies that are used to build these computer systems.
1:16:35
And so at Sigleree, we're looking across all the different quantum modalities and hardware types and architecting computer systems to meet the requirements of AI based on the maturing uh path that all these different hardware modalities are on.
1:16:48
And that allows us to build systems that are specifically tailored to AI and that we believe are going to be able to meet the meet the requirements of bringing quantum into the AI data center at scale.
1:16:58
>> How important is simulation at this point?
1:17:00
Are you at a place where you can uh run this like like basically run the the the code of the future uh in simulation to understand like run it on a classical computer not see the performance gains but at least understand that uh when the computer when the quantum system is available uh there will be a cost savings.
1:17:22
Yeah, we we've been able to do that largely speaking and you can do simulations of of something computer system or a jet or anything and varying levels of physical fidelity and and detail.
1:17:36
>> Uh the simulation we've been able to do so far indicate that we expect a level of you know several orders of magnitude potential speed up for key training tasks. Right?
1:17:44
So this is not a factor of two or a factor of five increase that we're targeting with quantum acceleration inside the data center.
1:17:52
it's several orders of magnitude, you know, when when all the pieces come together.
1:17:55
Um, but that simulation you talked about is a really really important and powerful part of designing a computer system.
1:18:01
You can't simulate all the all the logic of a quantum computer because that would require a quantum computer itself kind of by definition.
1:18:08
But you can do load profiling, you can do uh you can do traces, you can understand how that's going to, you know, uh uh uh be b distributed across classical and quantum hardware and also simulate all the networking transactions in between.
1:18:22
And so that's a kind of simulation driven design approach we're taking. >> Yeah.
1:18:27
Um I I guess uh what specifically in training benefits from quantum computing because the the example that everyone goes to in terms of uh quantum computing uh you know novel uh no novel algorithms that actually have potential to do something that a classical computer can't do.
1:18:48
It's like shores algorithm cryptography usually.
1:18:49
Um, but when people think about training AI, they usually just think a bunch of matrix multiplication.
1:18:57
Uh, is there some different path that you plan on taking or do you think you can uh operate at sort of a hardware agnostic layer?
1:19:06
uh much like we're seeing you know leading AI firms get off of CUDA like is there a world where you get off of classical and but but by and large it's the same training paradigm.
1:19:23
>> It's really interesting.
1:19:23
I think the answer is both.
1:19:25
So the our our starting point is we're looking at ways that you can insert quantum algorithms and quantum computing capability into the existing paradigm the ex the existing workflow for training and deploying very large models frontier models at scale.
1:19:41
And that means that you're looking for an insertion point from quantum algorithm where the data in the data out allow you to then take a step that would take maybe you know uh uh a day or two classically and compress that down to hours or minutes and do that throughout the workflow.
1:19:56
>> Um the challenge is that quantum computing provides an exponential uh you know the possibility for exponential speed up with the right algorithm but it also has this issue with data in and data out.
1:20:06
So it's classical data in which is can't be exponential in size and classical data out.
1:20:10
And so the less you do that that translation between the quantum part and the classical part it's going to end up working better.
1:20:17
So asmtoically where we're where we're heading is more quantum native models models that are designed in the first place to leverage a quantum computing capability tightly integrated with your classical infrastructure.
1:20:28
But where you're not where you're probably not going to see is fully quantum, you know, quantum based models that don't include a substantial amount of classical compute as well. >> Yeah.
1:20:40
>> So this isn't going to replace all the, you know, the AMD or Nvidia infrastructure in the data center.
1:20:44
It's going to augment it.
1:20:46
And our business model and our our focus and our product strategy is to take the to build a quantum accelerated AI server that sits next to the pod and acts as an accelerator for the XPU or the GPU pod in the data center and drive towards a very high attach rate of ideally one to one in the data center infrastructure of the future.
1:21:04
Um and that's what's going to allow you to then run you know accelerate the current paradigm but also use that as a substrate to design new kinds of models that will fundamentally be better and more efficient.
1:21:14
more efficient from a time perspective, from a cost perspective, from an energy perspective, but also uh these models are just a in a way just a representation of the computer hardware that they're based on.
1:21:26
>> And what's easy and hard from a computing and communication perspective on the hardware translates into the model capability and with quantum you have a fundamentally new resource in the data center that's going to allow new model capabilities to be developed and brought and brought to market.
1:21:41
How are you thinking about, you know, timelines with this new with with this with the new company?
1:21:46
Is it do do you think there's uh I imagine with the business right now is like an entirely more of like tech technical risk than execution risk.
1:21:58
Is that is that the right way to think about it?
1:22:00
Like there's a lot of hardcore research that needs to be done understanding, you know, the feasibility of of of the approach.
1:22:09
Um and uh and and what kind kind of like conversations are you having with you know potential partners uh uh if at all right now versus you know about about kind of like the near-term application or are you you know are conversations like 2030s and beyond kind of thing.
1:22:29
>> Yeah, we're targeting we're talking to customers now.
1:22:31
We've got several active you know conversations.
1:22:34
I think partnerships and early engagement with customers is a big part of our strategy.
1:22:40
The reason that's important is because the the challenges of really bringing a new compute uh you know capability into the AI data center are substantial and you got to be working with customers out of the gate to really understand those requirements, what moves the needle for them as an organization.
1:22:54
And so that's what we're that's what we're doing and that's what we're focused on.
1:22:57
Um in terms of timing, it's a fantastic time to start a company like this.
1:23:02
to start a company like this. the underlying hardware has made such tremendous progress in the past 1015 years and the market is you know with the amount of investment that's being made in AI infrastructure there is clearly a recognition that we need a new
1:23:18
approach to drive down the cost per token to drive down the energy associated with these very large scale data center projects to make it fundamentally more efficient and quantum promises a you know a more efficient way of translating watts into intelligence. That's what this enables and unlocks in
1:23:34
That's what this enables and unlocks in the long term.
1:23:36
the long term. And to me, this is in in many ways a better idea than putting stuff stuff in space >> because uh ultimately yeah, space gives you a lower, you know, cheaper access to energy and it gives you a better way to
1:23:48
to uh dissipate that heat, >> but you got to put it into space and that takes a lot of fossil fuels, that takes a ton of energy in the first place and it doesn't actually change the computational complexity of the computer hardware that you're running. Why don't
1:23:59
Why don't the power the power challenge quantum can unlock much more than that?
1:24:06
>> Yeah, it's a it's it's a good point.
1:24:06
Why why don't you think Elon has has made a a real run at quantum?
1:24:14
I think the answer is that quantum is at the at this interface of deep science and engineering and a lot of what needs to happen over the next three to five years to bring this technology to market at scale is engineering risk but it is quantum engineering risk and it's not
1:24:29
vanilla you know it's not not that it's easy not that any of the purely classical stuff is easy >> vanilla rocket science >> it's not vanilla rocket science and it's not vanilla fab at scale right and so even if you look at the leaders in quantum computing hardware It's not necessarily the Intelss of the world. >> Incredible company that has, you know,
1:24:46
>> Incredible company that has, you know, propelled humanity forward for half a century, but they're not the leaders in quantum because quantum is a new form of engineering and I wouldn't characterize it as science risk.
1:24:57
it as science risk. I think for quantum a lot of it that that is behind us even though there's tre there's tremendous work to be done but there is a lot of quantum engineering risk and that's an area where I think you need to see uh you know companies that are quantum specific bring the technology forward
1:25:11
and at that point I think that all the big AI labs are going to need to lean in with quantum >> when when do you think there will be a flip around sentiment uh from uh around around quantum it feels right now like at least in our corner of the internet there's so FUD around quantum and uh obviously >> based on financials like Right. >> Yeah. Yeah. Yeah. So that's that's what >> Yeah. Yeah. Yeah.
1:25:32
So that's that's what I want to know though like is is there a m like you know rewind 10 years if somebody said AI there was a very very small percentage of people that were like incredibly excited about it >> and you know deeply involved and could see the trend line and could see that we would get to this point.
1:25:50
I mean, Sam was talking about like people becoming best friends with a chatbot.
1:25:55
I think in like 2015 or something like or3 was like losing money.
1:26:00
It wasn't like making revenue yet. >> Yeah.
1:26:03
And that was even before that. >> Yeah. No, I know. Well, well before that.
1:26:06
And so, but then eventually it flipped and and it's really hard to, you know, there there's there's a lot of people that are AI bears and they and they talk about like overinvestment, but they can't deny the value of the products, right?
1:26:18
like they're fundamentally pretty useful, right?
1:26:20
Um, and you could argue that they're, you know, >> well, some bears can, but yes. >> Yes.
1:26:29
>> Some some bears would still figure out a way to argue that it's that they're not useful, but but I imagine like with both of your companies, you're predicting that like, you know, within the next five year, there's there's like a a flip.
1:26:42
But what do you think is the first kind of like driver of that where um maybe the average uh the average person in Silicon Valley actually starts to say like, "Hey, I wasn't taking quantum seriously enough." >> Mhm.
1:26:56
>> It there's a few things that need to happen.
1:26:57
I think the the FUD is real because the companies that are succeeding and doing well in this space, you you can't tell by looking at their financials.
1:27:05
You can't put on your kind of growth investor uh hat and say, "Yeah, this is going to be a tremendous company." and look at the metrics.
1:27:11
It it doesn't work like that.
1:27:13
You got to be able to analyze and and look at these companies and value them >> based on their ability to buy down technical risk over over time and the progress that they've made towards that.
1:27:23
>> So it just creates a lot of uncertainty because it's a challenging task and it's subject to a lot of you know dis discussion and debate.
1:27:30
Um but nonetheless I think there are clear there is clearly tremendous you know momentum and progress in the space.
1:27:35
Now what's going to change it? I don't know.
1:27:39
My bet is when we have quantum computers in the data center running production workloads and that you don't have to say hey that's a quantum computer for someone to care.
1:27:48
>> You care because it's a more efficient way of generating the you know the answers you need or or or you know training the model or deploying the model for inference.
1:27:55
And uh that's when quantum is really going to become a mainstream category is when you don't have to talk about the fact that it's quantum anymore.
1:28:03
And I think in a in a large part uh this is what we're trying to achieve with Sigle, right?
1:28:10
The goal is that uh to take quantum computing and to obfuscate it underneath the hood of a classical computing system or underneath all the rest of the infrastructure that's already there and to not ask the end user to be programming it and writing code for it.
1:28:25
That's all going to be done with AI anyway.
1:28:26
And so uh that is just a better it's a better way to train your model and you know you need this thing or else you're it's going to take you too long and your customers aren't going to be happy with the quality of you know the outputs they're getting.
1:28:38
That to me is a big inflection point and I think that can happen in the next 5 to seven years.
1:28:42
next 5 to seven years. I think that can uh but there's this whole march that needs to happen to take the technology from one proof point to then you know all the cost engineering that needs to happen the reliability engineering and that's going to be the really fun journey for quantum computing over the next decade is to get to that point
1:28:57
where we're selling you know hundreds or thousands of units a year and uh and but that's the journey we're on and that's the march that quantum technology has been on for a good you know one two decades now >> and then uh this is probably very obvious to somebody that is focused on quantum but But uh not not to me just cuz I I I don't I don't follow it closely. But like why why a new company?
1:29:17
But like why why a new company?
1:29:19
It feels like quantum like as you've explained it feels very obvious to apply it to uh data center buildout and and you you said it could be like a a meaningful inflection point for the technology overall.
1:29:31
Uh why why why was a new company necessary and and you know why did you take this approach?
1:29:39
Well, at a high level, I think all the different quantum hardware modalities have made tremendous progress and the right way to build quantum computers for AI is multimodality.
1:29:46
That is a fundamentally new approach and uh it ultimately is going to in my opinion be very obvious in retrospect.
1:29:53
It's going to work better.
1:29:55
Uh but it is a it's such a fresh idea.
1:29:58
It's got to be baked into your strategy, the DNA of your company.
1:30:02
And then uh all the different quantum hardware companies that were out there before Sigler basically started with a thesis which was we've got the best cubit and so we're going to scale this cubit type up and see how far we can get by scaling it up.
1:30:13
And that's why you have so much doctrine and like kind of organizational belief around a particular cubit choice.
1:30:19
But in reality, you know, customers are buying a computer.
1:30:23
They're not buying a you know the the the physical device or your your cubit technology.
1:30:27
And so uh at Segree what we're doing is working backwards from the market application from the AI workload as the as the use case and using that to drive the specification of a system that can then be built from folding in whatever technologies are needed to meet those requirements.
1:30:40
It's just such a totally different approach to quantum hardware.
1:30:44
It's got to be a new company and uh that's that's that's signal technologies.
1:30:48
That's the approach that we're taking.
1:30:50
I think that that is uh ultimately what's going to unlock this new you know this market application of AI.
1:30:54
The other reason is you said it's obvious, but it's actually not obvious at all to most people in quantum that quantum is going to be useful for AI.
1:31:02
And in fact, it's not even a consensus view right now.
1:31:06
>> And the reason for that is because quantum algorithms themselves are still in this very this phase of discovery and in development.
1:31:13
And obviously AI is going to help with that eventually as well to an extent.
1:31:16
well to an extent. Um but quantum you know when you interview a set of leaders from across the quantum hardware industry the the you know the the median answer you're going to get for what the applications of quantum is going to be is you're going to use it for quantum chemistry you're going to use it for
1:31:32
optimization problems things like that and applications to frontier AI is a new area that is just being developed now because it requires a development and extension of what current algorithms can do and then new algorithms altogether specifically for that that's what we're tackling at Sigle is that kind of quantum AI native research lab, right? Or a frontier a AI lab that's quantum
1:31:51
Or a frontier a AI lab that's quantum native and uh and then we're doing that alongside developing our own quantum hardware. >> Mhm.
1:31:59
>> Uh before before we jump, I didn't get we you you mentioned kind of the the history behind the name, but what is the significance of Sigleree in the novel that you mentioned?
1:32:10
>> Oh, well, you guys got to read the novel novel for one.
1:32:12
It's absolutely incredible.
1:32:14
And uh but uh the other thing is single tree is basically a discipline in the book that is learned at university and uh you know and it basically amounts to you inscribe runes on a particular object and by doing that you can imbue that object with properties that it wouldn't otherwise have or you can govern like heat and light flow and things like that.
1:32:31
It's also a discipline where it's got a quant quantitative angle to it and if you do it wrong you can blow things up.
1:32:36
So, it's got this this mix of kind of coding and hardware, but then a mysterious kind of angle of controlling things from a distance by how you do this uh these inscriptions.
1:32:46
It's really it's a really amazing concept, >> a little bit of magic. >> Amazing.
1:32:50
Thank you so much for taking the time. >> Thanks, guys.
1:32:52
Have a great rest of your day. >> Cheers.
1:32:55
>> Let me tell you about Console.
1:32:55
Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets.
1:33:02
Our next guest is already here. >> Very cool name.
1:33:05
I wish you know what I'm thinking, John.
1:33:08
>> General, >> I wish that uh after computing, I wish that Chad launched Chad computing. >> Oh, yeah.
1:33:16
>> Uh it was right there. It was right there.
1:33:18
Anyway, but >> we have the co-founder and CEO of General Intuition with us.
1:33:20
Welcome to the show, Pam. How are you doing? >> What's happening?
1:33:26
>> Hey guys, >> thanks so much for coming back on the show.
1:33:29
>> Uh uh >> yeah, please.
1:33:31
>> We're We've been talking about names for labs.
1:33:33
labs. What about uh consider general intuition strong name >> but since since you launched the company a lot of other a lot of other Neil labs have kind of come out with like >> similar names like general there's probably like a general super intelligence or like a general ASI how about you you rebrand to unfettered
1:33:55
>> intelligence >> that might be it >> or how about we just fund them all >> yeah that that too >> yeah what what what is the plan to when do do you see yourself as a Neolab and do you see uh is it as much of a of a knockout dragout fight as it appears from the outside or is your model more of a uh a thousand flowers bloom? >> The plan is to just keep renaming
1:34:18
>> The plan is to just keep renaming [laughter] >> the uh No, the look you have to have a claim to why you can win I think otherwise none of this makes any sense.
1:34:30
Um, it's an incredibly competitive fight.
1:34:33
There's lots of great contenders.
1:34:36
The only reason why we have a shot is because we have a data set that nobody else has, which allows us to be as focused on workloads that include space and time as entropic on their code environments on the way to the frontier.
1:34:49
>> And so you need to have a very focused dedicated path.
1:34:52
Some of that can be um uh for instance having the best researchers or having having the the new ideas but I think it also has to be supplemented with a product focus of some a customer problem that is going to get solved because these types of model classes exist.
1:35:14
Network effects just like we saw in the consumer eras of the Facebooks and the Twitters and the Reddits. These things are true.
1:35:19
They apply to LMS as well.
1:35:19
the fight for that space is going to be incredibly tough and so you have to introduce something new.
1:35:27
I don't believe in the just M30LM space.
1:35:31
Um, which is why we we're focused on on actions and space and time.
1:35:36
>> What's the [clears throat] >> Okay, actions in space and time.
1:35:37
Let's talk about the data set.
1:35:38
Uh, catch everyone up to speed on uh I mean, you know, you you broke it down for us the last time you're on, but it feels like it's been almost a year at this point.
1:35:48
So, what have you been working on?
1:35:48
and talk about the data set, how you're how you're building the data set, all that stuff. >> Yeah. Look at this way.
1:35:54
As humans, the decision to talk or type is just a very very small subset of the actions that we can actually take, right?
1:35:59
We can choose to move our body.
1:36:01
Um, and so in order to create a sufficiently general intelligence to play 10,000 plus video games, the model has to be able to predict across the entire action space of human cognition when they're interacting uh with these environments, which is 2D environments, 3D environments, interfaces, um, long horizon tasks, short horizon tasks.
1:36:19
Um, and so in order to do that, it has to be a sufficiently general intelligence in order to uh, learn how to correctly predict actions.
1:36:27
Um, and therefore the type of model you get out is not going to taste like an LM.
1:36:33
It's going to be like comparing coffee to water.
1:36:34
This model is going to be incredibly good at navigating unforeseen environments.
1:36:39
It's going to be incredibly good at um a zeroing any task where it can already be controlled using a game controller because we have roughly a trillion action tokens uh in that space for example, right?
1:36:53
For context, Frontier LM are trained on maybe between five and 10 trillion text tokens, right?
1:36:57
And so we have a scale of data that is going to allow us to jump to the frontier in one capability, which is any system that can be controlled using game controller, >> which is most robots, right?
1:37:08
That's really what we're doing.
1:37:09
We're using that simplification to turn it into mostly an environment transfer problem.
1:37:16
And then you can use that to create a sufficiently general intelligence where you maybe at some point add text to the output space. Right?
1:37:24
It's not going to be text as you're used to from LMS, but it might just be enough to communicate why you're doing a specific thing.
1:37:30
So >> that's how to view the models.
1:37:34
>> So yeah, walk through the partnership with metal.
1:37:36
Are you getting game controller feedback as well when those Yeah, explain explain the relationship to metal for those.
1:37:45
So alongside the frames in the video, um we're also getting the exact action inputs.
1:37:49
To be clear, not the letters or numbers, right?
1:37:52
We can we have we had thousands of humans convert those into the actions you're taking.
1:37:56
So walk forward, walk left, >> um uh open door, close door.
1:37:59
Um, and so when you have that at that ground truth level, >> you don't need to train models that try to extract that information from the videos, which you are now in a completely different uh scaling regime as if you were trying to do this on inferred data.
1:38:17
So for example, if you're landing a plane and you're moving the rudder, that's not going to be visible in the pixels.
1:38:23
It's impossible for that to be visible in the pixels, right?
1:38:24
But it's in the action sequence.
1:38:27
And so there's just no lab that can take this approach.
1:38:31
There's lots of benchmarks that might show that you can do this on inferred data.
1:38:35
The problem with inferred data and these benchmarks is that they show up in a really nice way on general tasks, but customers care about how these models perform when you're in an edge case and you need specific actions to go in specific ways.
1:38:47
And so um you cannot do this on on inferred data uh despite uh many people claiming you can.
1:38:57
Tell us about the latest round. I want to hit the gong. What happened? >> Yeah.
1:39:01
>> How much did you raise? >> What happened? We raised $320 million.
1:39:06
>> That >> Congratulations and thank you so much for taking the time to come chat with us.
1:39:11
>> One more final question.
1:39:11
Uh what is the talk about uh progress from your customers, companies that you're talking to in in robotics?
1:39:20
Where is maybe an area that that you're particularly excited about that you don't see being talked about yet?
1:39:27
>> Yeah, the the most obvious thing this replaces is um all the code that people are currently writing for behavior in physics engines.
1:39:33
All that just becomes a prompt.
1:39:35
Um and so uh think of the models as based on an input stream of just frames being able to control whichever system is sending those frames in the action space of a game controller or keyboard and mouse.
1:39:49
But basically, you can play the world as if it was a video game.
1:39:53
If that can be said about your use case, the models will generally do incredibly well.
1:39:58
>> Um, the reason why this works is because every robot already ships with these, which means that they can simply predict at the level of these controllers and therefore the robot has already accounted for sort of human monkey brain to motor torque prediction interface and merging that with the actual things coming from the controller. Right?
1:40:13
So, we're using the fact that those interfaces exist as a level of predicting in a general action space that works across many types of robots.
1:40:23
In many ways, you could argue that um if this is correct at scale, the supply chain will converge on gaming inputs instead of humanoid robots.
1:40:30
And I think that is one of the big things that I foresee happening in the next two years.
1:40:36
>> Um because intelligence is a bottleneck. >> Yeah.
1:40:40
>> Well, thank you so much for taking the time to come chat with us.
1:40:43
Congratulations and we'll talk to you soon. Have a good one.
1:40:46
Let me tell you about Cisco >> critical infrastructure for the AI era.
1:40:51
Unlock seamless real-time experiences and new value with Cisco. Uh fascinating.
1:40:56
It's also funny seeing all those simulators on Steam like and and the fact that like will the training data generalize?
1:41:02
Is are they just going to learn how to play Fortnite?
1:41:04
And it's like well there is a farming simulator and there's a you know >> data center data center simulator.
1:41:08
Happy bar central banking simulator.
1:41:10
potential banking simulator.
1:41:12
It's going to learn everything.
1:41:13
Well, well, we have our next guest in the waiting room.
1:41:15
Uh, Yadin Suffer from Tra, the co-founder and CEO. Welcome to the show. How are you doing?
1:41:23
>> Hey guys, nice to meet you. I'm great. How are you?
1:41:26
>> You thank you so much. >> What's happening? >> Introduce yourself.
1:41:28
Tell us what you're building.
1:41:30
Tell us about the emergence from stealth that's happening today. >> Yeah.
1:41:35
Well, Yodin Sofer, we last week we announced the launch of Tracer, >> which is I would say the first of its kind subterror defense tech company >> and uh subterror is a word we actually coined, but I've been happy to see people reference it on X already.
1:41:48
It refers to everything in the subterranean defense domain.
1:41:53
So that's everything in the intersection between military applications for things that happen beneath our feet.
1:41:59
>> What is the history of subterranean startups?
1:42:01
So you have the Boring Company.
1:42:03
Uh Palmer has talked about the domain.
1:42:06
I don't think he coined it.
1:42:06
So you get all the credit.
1:42:10
>> Um but but what what have been some historical sort of just like general efforts in the category maybe outside of the Boring Company?
1:42:19
>> Yeah, I think on the civilian front actually subterranean is it's a developed industry.
1:42:23
You know, there's a lot of applications in the mining world and in the piping world, in the utility world where, you know, it it deserves some loves and and it did get.
1:42:30
You got amazing companies like Herk that are not, you know, >> sexy startups like the Boring Company, but these are decades old German companies that have been u you know, piercing the way, pun intended, in everything underground.
1:42:45
So I would say that in the civilian front there's a lot of innovation happening but in the defense front I don't think you'll find any.
1:42:54
I mean we we really have not seen any any companies in this space.
1:42:58
>> What are the primary challenges of you know underground drones the underground domain overall? Is it connectivity?
1:43:10
>> Uh but what what are they? >> Oh yeah. >> Yeah.
1:43:14
>> Yeah. Well, you know, I think uh it's interesting because uh the folks our engineering team come from a combination of the Boring Company and SpaceX and usually you see them kind of jumping between those two companies and they
1:43:26
have an interesting saying that says that you know everyone calls rocket science rocket science as if it's the hardest thing in the world but when it comes to air you know what forces you're dealing with right you know you know
1:43:35
what you're dealing with and when you're working on the underground when you're essentially boring your own uh you don't know what to expect you don't the ge ology, composition, you can have a high sense of how it's going to look, but
1:43:48
when you're down there in the dirt, you don't know if suddenly you hit hard rock and you hit something else and you need to know to either maneuver very precisely or to be able to replace your cutter head to something that can fit. So, I would say that is probably the
1:43:59
So, I would say that is probably the number one challenge, just the uncertainty of this domain.
1:44:04
>> Uh, Palmer talks about this.
1:44:04
He says diameter is expensive, length is free, something along those lines.
1:44:09
Can you explain that concept and how it informs uh vehicle design for the subterranean domain? >> Yeah.
1:44:17
No, it's such a great point and I think a lot of people looking at this space are thinking the same thing, right?
1:44:23
We're we're thinking a train where it bores its own path and it takes behind it essentially infinite payload, right?
1:44:28
You can have miles and miles of payload of sensors of effects and you know the dream is someday people.
1:44:35
Now, when you think about it, when you're increasing the diameter, you need to remove so much more dirt, right?
1:44:41
You're dealing with a lot more.
1:44:41
And when you work at a at a small diameter and essentially infinite length, you you could even condense the dirt to the sides.
1:44:50
You don't necessarily need to remove it.
1:44:52
And that becomes extremely valuable.
1:44:53
So, most of the questions are around that.
1:44:56
And I don't know if you guys have seen a boring site, but a boring site is this massive thing, right?
1:45:01
you need the bentonite to mix with the dirt to take back outside. It's like a whole thing.
1:45:05
But when you're working on small diameter, you don't necessarily even need to remove the dirt.
1:45:10
You can just condense into the sides.
1:45:12
And I think that's a big part of, you know, going sort of slim and long.
1:45:16
>> $25 million seed round. What's the goal?
1:45:19
The government isn't actively buying this technology.
1:45:22
There isn't a program of record that you can sneak into, I imagine.
1:45:26
So, what does the next two years look like?
1:45:28
Yeah, we we always say this that, you know, if you if you try to find the line items, they're like line items buried in line items, right?
1:45:36
Obviously, we have penetration munitions, but those are air air drop bombs, and we're not looking to compete with Boeing.
1:45:41
But I would say the the interesting points and the slivers we see of interest from the government right now are in um there was a recent RFI by DARPA where they're looking for new methods to induce collapse in underground uh infrastructure using uh different shockwave uh methods.
1:45:56
So essentially, we're looking at this as non-kinetic penetration munitions, right?
1:46:03
Our ability to insert a payload underground.
1:46:05
This doesn't have to be uh dropped from air.
1:46:07
It can be done by a special uh special forces on the ground and essentially detonate a payload in a sequence that induces collapse of facilities like in Iran.
1:46:15
So, you know, I think the military is starting to understand that the existing solutions do not deliver what we need them to.
1:46:20
So, they're starting to think differently.
1:46:24
But back to the round, right? with $25 million here.
1:46:27
Everyone goes to me and is like, "All right, you're building this massive R&D team.
1:46:30
You're gonna have a ton of capex."
1:46:31
And I'm like, "No, there is a lot of work to be done when Forming call it this category where we need government.
1:46:39
We need the military to recognize this as a category like we do and essentially to go after large prototyping buckets that will then allow us to fund these long-term developments that we believe will allow us to win wars.
1:46:51
So for us, most of the focus right now is just working with DC, working with the military and establish I would go as far as saying the subterrorct or the US subterror strategy for, you know, winning wars underground.
1:47:03
>> How far underground are you right now?
1:47:05
>> It does look like you're underground, >> right? It looks deep.
1:47:07
I was I was thinking about this, too. It's a good spot.
1:47:11
>> At least at least 20 ft. At least 20 feet.
1:47:15
>> Anyway, thank you so much for taking the time to come chat with us. >> Great. Great to meet you.
1:47:19
>> Have a great rest of your day. We'll talk to you soon. >> Cheers. >> Have a good one.
1:47:21
Let me tell you about Figma. Agents, meet the canvas.
1:47:23
Your AI agents can now create and modify your Figma files with design system context.
1:47:30
And Jack Morris from Engram is in the waiting room.
1:47:32
He's co-founder and head of research.
1:47:34
[music] Jack, how are you doing? Welcome to the show. >> Hi. Yeah. Uh, nice to meet you.
1:47:37
It's great to be on the show.
1:47:39
I was actually just watching it in another tab, so this is kind of >> Here you are. Fix that.
1:47:47
>> Um, >> great to meet you.
1:47:47
uh tell us a little bit about yourself, tell us about the company.
1:47:51
Uh you're emerging from stealth uh with a whole lot of venture capital.
1:47:56
What's the strategy and what's the product? >> Yeah, sure. Um my name is Jack.
1:47:59
I'm a co-founder and I guess technically the head of research at Engram.
1:48:05
>> We came out of Stealth last week after eight months or so of working on our product and ideulating with our design partners.
1:48:13
Yeah, we raised money for a bunch of VCs.
1:48:15
The product is [laughter] >> is mogged mogged. >> Let's hit the gong.
1:48:22
Let's hit the gong for that.
1:48:24
>> Grateful for the opportunity, but I was hoping you would hit the gong.
1:48:27
>> Yeah, we just did a b, you know, baby. Yeah, big one. >> Congratulations. >> Yeah.
1:48:34
And thanks to all of our partners and thank you so much for funding us.
1:48:38
Um, our product is a new type of AI.
1:48:38
So, I think we have a pretty different vision from a lot of the frontier labs which are sort of working on like one model per lab and trying to make that model smarter every month.
1:48:51
I think there's another way to think about it which is that the model doesn't need to get smarter every month.
1:48:57
It needs to know you better.
1:48:58
And so we're working on like a whole different stack which is a way to train models that that train themselves to like know your world better and like adjust to the things that you say.
1:49:10
So it's like new ways of training, new ways of running the models.
1:49:14
I think like to give a concrete example, I assume you know you all are very tech forward.
1:49:18
You probably have agents doing things like preparing >> preparing you for the show and like giving you reports every morning.
1:49:25
And if you actually look at what the models, the agents are doing, they're probably like reading the same files a lot to get context about what your show is and what you do.
1:49:34
Like literally probably every night, they're probably like reading from scratch.
1:49:38
What is TVPN and who are you two and you know who's been on the show recently >> and it's >> No, we're in the pre-training now.
1:49:44
Come on, give us some credit.
1:49:45
[laughter] >> Oh yeah, you are in the pre-training move so fast.
1:49:51
>> Your point 100% stance, >> but yes.
1:49:54
Yeah, I I think you're lucky because you're in the pre-training, but I think most people are not in the pre so many documents that aren't and you have to feed those in every time.
1:50:00
Is this the solution to continual learning?
1:50:04
Is that the correct uh buzzword for this strategy or is this a different fork in the road, a different path?
1:50:11
>> Uh I think it's the correct buzzword.
1:50:11
I think a lot of people use the phrase to meet a bunch of they cracked it in eight months.
1:50:18
They cracked it in eight months.
1:50:20
It only took them eight months. >> Let's [laughter] go.
1:50:22
Oh, we decided to name ourselves something different.
1:50:24
But I think the >> I think of continual learning is basically this problem of how do you keep the same model but actually update it like rewire it every single day to learn more about what you're doing and we're working on that problem.
1:50:37
>> What's the what's the sweet spot customer enterprise AI that can mean fortune 500 companies that can mean a very data inensive company.
1:50:44
There's also whole categories of enterprises that have uh a whole host of of AI rappers and application layer companies duking it out.
1:50:55
I'm thinking of legal, medical.
1:50:57
Um where where do you see the product having the earliest signs of product market fit?
1:51:04
>> Yeah, I'm glad you said earliest because I think like there's there's two halves to the vision.
1:51:08
to the vision. one is the long-term vision which is that the model will get to know you better and understand everything about you kind of like a person does like your you know co-orker um and it'll be able to like generalize and do things better than the current models but I think the current customers
1:51:26
and like the way we're finding early success is by making the models a lot cheaper because like essentially they know everything about you already and instead of reading like a hundred files to write a summary of what you need to do tomorrow they read you four files or something like that. >> So, our early enterprise partners that
1:51:41
>> So, our early enterprise partners that we've been working with are Microsoft, Notion, and Harvey.
1:51:48
>> And I think they all >> you guys with the sound effects, I'm like so flattered.
1:51:51
I wasn't sure if there would be any.
1:51:54
>> Um, >> they they're nice because they have these like massive >> uh workspaces of context and like they're, >> you know, early adopters of AI.
1:52:03
And I think these are the places where we can like reduce costs the fastest.
1:52:06
the soonest because the workflows really are just that repetitive. >> That's great.
1:52:12
Well, thank you so much for coming on and breaking it down.
1:52:15
Appreciate you taking the time and have a great rest of your day.
1:52:18
>> I know you will be back on I'm going to guess two times [laughter] this year. That's my guess. Two times >> for sure.
1:52:24
>> We'd love to have you back and chop it up more.
1:52:26
Have a great rest of your day.
1:52:27
>> Yeah, it's great meeting you guys. Thanks for having me.
1:52:28
Great to meet you, [clears throat] Jack.
1:52:30
>> We'll talk to you soon. >> Cheers.
1:52:31
>> Let me tell you about the New York Stock Exchange.
1:52:33
Want to change the world?
1:52:34
Raise capital at the New York Stock Exchange.
1:52:36
Our next guest is Neil uh from Sale Research. He's the co-founder. Let's bring in Neil Moa.
1:52:42
Mova, >> how do I say your last name?
1:52:46
I don't want to get it wrong. >> Moving.
1:52:49
>> Hey guys, great to be here.
1:52:50
>> Thank you so much for taking the time.
1:52:52
Uh >> congratulations on the round.
1:52:53
But first, please introduce yourself and the company. >> Yeah.
1:52:57
Hey guys, I'm Neil, co-founder and CEO of Sale Research.
1:53:00
Uh we are a company building the most efficient inference in the world. We love GPUs.
1:53:03
We dig deep into the stack to find efficiency everywhere and we make tokens super abundant. >> All open source.
1:53:10
Do you work with other labs?
1:53:12
Are you uh how deep do you go into the relative organizations? >> Yeah. Yeah.
1:53:18
So today it's all open source models. You can imagine GLM 5.
1:53:19
2 is a big moment for us.
1:53:21
We're very excited about that.
1:53:24
>> U in terms of how deep we go well in the stack, you know, we basically do everything between the chips. We don't make chips. We buy chips.
1:53:30
>> Uh and we go all the way up from there to the API. >> Tell us about GLM 5. 2. too.
1:53:33
Uh what makes it uh what makes it like different in a binary sense?
1:53:40
Is it is it a particular benchmark? Is it a vibe? Is it an application?
1:53:45
Have we unlocked a new co capability in open source AI?
1:53:50
>> Yeah, it seems like uh ZAI really figured out post training with this release.
1:53:53
That was something that was held back uh with the previous releases from Deep Seek and Kimmy, let's say, and they've just really done it.
1:53:59
The style of the model is excellent for coding.
1:54:02
It's the first one I actually with the straight face would recommend my colleagues try for coding.
1:54:06
>> For coding specifically, uh what about >> before you would put on clown makeup and [laughter] then you say, "Ah, yeah, give it give it a spin."
1:54:12
>> Uh what about for other agentic workloads?
1:54:14
I mean, we're looking at open router, a lot of the top models, uh DeepSeek V4 light.
1:54:18
It seems like it's a lot of heavy token generation, lots of lots of value being created, but smaller tasks.
1:54:25
Um what what is that like from your business perspective?
1:54:28
Are you still focused on optimizing those types of workloads? >> Yeah, for sure.
1:54:32
You know, Deepseek has always been the economics king.
1:54:34
Uh we want to bring that to every model.
1:54:35
Of course, we could talk about that a bit more, but um yeah, I think you're going to find that like some of these more background tasks that are not coding per se.
1:54:42
Uh those will always go to the strongest intelligence per dollar and take a pretty broad view of what that intelligence could look like.
1:54:48
And I think DeepC uh is still quite up there.
1:54:51
Deepc flash is quite high up there. >> Yeah.
1:54:53
How how do you think do you have any intuitive sense for the ratio of token spend or tokens or anything on background tasks versus a human prompted an agent?
1:55:06
Because we hear about token maxing and it feels like it's a lot of a developer went and fired off something and it cooked for a day and it's spun up a bunch of tokens.
1:55:15
But when I think of the really high volume token future, I think of maybe it's an agent, but maybe it's just every single person that checks out on an e-commerce website goes through a fraud detection check that is now token powered and is not just, you know, a bunch of Python code, it's actually inferencing something or every time you book a flight, it runs some LLM check.
1:55:37
Uh, and I imagine that that will be a huge driver of token consumption.
1:55:42
Um, and I'm wondering how you see those two buckets balancing out. >> You know, 100%.
1:55:48
I think, you know, to give you topline number today, I'd estimate it's like 80% of stuff is human in the loop today and 20% is background.
1:55:56
But that number is going to shift.
1:55:56
And I I actually expect the crossover to happen this year where background dominates.
1:56:00
And the reason is, you know, as you pointed out, you want to use these agents in workflows, deterministic workflows.
1:56:06
And we just weren't there yet uh with our agents from six months ago.
1:56:10
And we've just we've crossed a few barriers in the last few months.
1:56:12
So yes, I think we have the unlocks required for agents to run a lot longer uh reliably on every action that a human puts into assistant. >> Yeah.
1:56:20
And then that's very good for your business because if I have something that's running on a Sunday when none of my employees are in, but it's still firing up $1,000 of cost, I want to come to you and get it to be $500.
1:56:30
Like what what what type of pitch do you have in terms of savings?
1:56:37
>> You know, I don't really want to save my customers money. Okay.
1:56:38
customers money. Okay. I actually want them to spend a lot more money with me because I've actually made the ROI so good that they're coming to me for way more to [laughter] >> and you know one of the ways I like to say it too is um you know I like to work on unbounded problems and before when we built human in the loop agents those
1:56:54
were very bounded problems you have a limited number of >> limited amount of patients to read agent output every day >> but if a can run in the background for a long time well we've decoupled the two and uh there's no limit trillions of tokens per task is within reach >> what were you and the team doing before this and and how long have you been at it? >> Yeah, so I've been working on GPUs for
1:57:11
>> Yeah, so I've been working on GPUs for about 10 years now. I love this stuff. It's my whole life.
1:57:15
>> I was at 10 years ago.
1:57:17
>> Is [laughter] this a possible story where you're like, I was working on GPUs and you were just playing Counter-Strike or something? [laughter] >> No.
1:57:23
Well, you know, I was at Nvidia, which business, right?
1:57:25
You know, I remember being a little skeptical 10 years ago, like Jensen's talking this big talk about moving to AI, but like realistically, you guys, we do five billion in revenue from gaming.
1:57:34
That's surely that's gonna be the biggest business for Nvidia for a long time.
1:57:38
>> I imagine well I could see that now.
1:57:38
Um and then I was previously at Apple as well.
1:57:44
Apple had a pretty competent ML program or ML silicon program.
1:57:46
I won't say anything about their ML software program. >> Sure.
1:57:50
>> Um and uh and then most recently I was at >> Very cooling. Very cool.
1:57:54
Uh >> kind of a perfect background for this business.
1:57:58
>> What is Lip Bhutan like in person? I'm such a fan. He's an angel investor. How'd you meet him? What's the story?
1:58:05
Yeah, I met him through our friends at Sequoia.
1:58:06
They build great relationships like this one.
1:58:08
Constantine in particular knows Liu very well. Uh Liipu is great.
1:58:12
I mean, he's I've never met someone with that combination of like warmth and and business acumen, but also he deeply understands the chips for building.
1:58:17
I mean, he can just like go from talking about Foundry to talking about uh you know, the the nuances of like how to scale an inference business in this very wild time.
1:58:26
Um so, I love working with Lu. He's exceptional.
1:58:29
>> Yeah, what a what a wild run from him in such a short amount of time.
1:58:31
uh one of the greatest story arcs in uh in technology.
1:58:36
>> And then who did who did the round?
1:58:38
>> Yeah, so Sequoa uh did the seed.
1:58:41
Constantine and Lauren Reer.
1:58:43
>> Uh and then for the series A, we went with Kleiner Perkins did the for the lead. That's Adith. >> Yeah. >> Amazing. >> Fantastic. Well, congratulations.
1:58:52
>> Fantastic progress soon.
1:58:52
And thank you for everything you're doing.
1:58:53
We appreciate >> great to meet you.
1:58:56
>> Have a good rest of your day. >> Cheers.
1:58:57
>> Let me tell you about Railway.
1:58:57
Railway is the all-in-one intelligent cloud provider.
1:59:01
Use your favorite agents to deploy web apps, servers, databases, and more.
1:59:04
While Railway automatically takes care of scaling, monitoring, and security.
1:59:08
>> They have a great new campaign that we can try to watch. >> Yeah. Yeah.
1:59:10
We got to watch some ads.
1:59:10
We haven't done enough ads.
1:59:12
Uh let's bring in Jacob Deepen Brock from Discipulus Ventures.
1:59:17
Welcome back to the show, Jacob. How you doing? >> How you doing?
1:59:22
>> So, you hoovered up stakes in every single Gundo company and now you hoovered up $30 million for a fund. Tell us the strategy.
1:59:29
you tell us how it came together.
1:59:31
Congratulations on the fund raise.
1:59:33
>> Yeah, thanks for having me, guys.
1:59:33
Um, yeah, we just raised 30 million for the the second fund.
1:59:37
Some great folks >> that's going to pay for a lot of barbecues on the beach. [laughter] >> Yeah.
1:59:48
>> No, I mean it really No, it really is like the most probably efficient like VC platform strategy ever is just like the the bonfires.
1:59:55
The value created at those bonfires is going to be in the multi-billions for sure if not already. Hopefully trillions. >> What?
2:00:03
Wait, what are you underwriting this fund to?
2:00:06
>> Do you got to get a trillion dollar company?
2:00:07
And is that the new stakes?
2:00:07
Are your investors asking you are you going to get us the next trillion dollar company or do you are are you thinking more smaller stakes at seed?
2:00:15
Do you want to deploy a lot of the capital into follow on investments, do SPVS?
2:00:19
How are you thinking about positioning the fund? >> Yeah. Yeah.
2:00:24
So the our strategy basically is we get like goodsized chunks for the fund at low prices.
2:00:27
We're basically the first investor in all the companies we bring through.
2:00:32
Um a lot of them time a lot of times help them incorporate the companies and then help them raise a larger round.
2:00:36
So we get it at low prices.
2:00:38
We don't actually need that.
2:00:39
Obviously it's great for us and I mean we've already seen some of these markups that make the fund look very good given our entry price.
2:00:43
Um but yeah I mean the goal is get good ownership for us not too much for the founders at low prices and the multiples look look good much easier.
2:00:52
I have a sorry [laughter] you're like a lot of ownership for us not too much for the found makes sense. >> No.
2:01:02
>> I have a I have a theory that uh we are we're not post defense tech boom like the companies are still booming but we're we're post defense tech incorporation boom and the ratio of defense tech in your hard tech fund will be declining if it's not already. Uh, is that true?
2:01:22
Is that borne out in the data? Is that exciting? What else?
2:01:23
What else is in the hard tech bucket that's exciting to you these days?
2:01:28
>> Yeah, we did a lot of defense early on.
2:01:30
I think there was a lot of more like gray area.
2:01:32
I think there's like a thousand drone companies now, which makes a lot of it less interesting.
2:01:35
A lot of missile companies, etc.
2:01:37
Um, I think we I think LA is the best place to build hardware.
2:01:42
I think Elsundo is the best place to build hardware.
2:01:43
And I think all the best engineers in supply chain is already built out here.
2:01:45
So we can kind of be as early as possible kind of getting to know the best engineers where the companies like SpaceX and we need to start defense companies early on.
2:01:54
But now we're seeing a lot of advanced manufacturing.
2:01:56
I think chemicals is really interesting.
2:01:57
Um I think the kind of general industrial space energy etc.
2:02:01
I think there's a lot of stuff that makes sense to build here because of talent supply chain that is not just purely defense >> post SpaceX IPO effect on your business.
2:02:12
Are newly liquid SpaceX employees investing in defense tech or are they just investing in luxury real estate? What's going on?
2:02:21
>> Yeah, I mean I think LA has still has majority of SpaceX.
2:02:23
Um I guess people who made money off of SpaceX.
2:02:28
>> Um so yeah, I think a lot of people will probably start companies now because like they made enough money to be comfortable and they can do whatever they want now. Yeah.
2:02:34
>> Um I think obviously they have like a lockup period so we'll see where that ends up.
2:02:38
Yeah, I think we do have some LPs who are from SpaceX, some people who've made a lot of money off of SpaceX already.
2:02:40
I think it'll be good for the companies here as well as for people just starting new stuff.
2:02:45
>> And we've already seen Radiant and uh and uh Tom Mueller's company, Impulse Space, uh both SpaceX, very successful companies, exciting stuff.
2:02:54
>> Uh moving forward, are you sticking with like like a batch style approach or are you just going to be writing checks more more flexibly?
2:03:01
Where do you think you go?
2:03:04
Yeah, I mean I think the core thing we have is like we are close to all the best engineering talent and we can basically kind of index a lot of the up and cominging companies coming out of here.
2:03:12
Um, so I think the batch part is like our unique thing that nobody else is doing and is how we're able to I guess generate alpha and and I think we will do follow on um into the companies and more this time than last time but I still think the core thing is like there are plenty of hardware funds that will do preceded seed etc.
2:03:25
And a lot of these prices are insane but if we can kind of be as early as possible find these young engineers before they leave and kind of be their launch pad into the right ecosystems of founders and investors etc.
2:03:34
That's kind of where we want to come in.
2:03:36
So it's going to be vast majority of the capital being deployed into um the co-work companies. >> Amazing.
2:03:41
Uh what is the what is the state of new talent coming to Elsa Gundo?
2:03:48
Is there still a boom there?
2:03:50
What's the incubator slash like uh class cohort-based entrepreneurship?
2:03:55
Uh get me up to speed on the latest there.
2:04:00
>> Yeah, I mean I think the the bonfires are a good kind of index on how many people here.
2:04:04
people here. I think we our last one we did last Friday we had like probably close to 200 people in that one and they've grown and I mean by a very large amount when we first started they were like 30 40 50 um so yeah lots more people coming I think from all over the world honestly I was in Europe a couple
2:04:19
weeks ago and like people were like oh I'm going to build my company in Elsagundo I'm moving from London to I think it's kind of continuing to boom um and the real estate prices are insane which I think also is a a good indicator of that people moving out to Torrent and Hawthorne but yeah definitely lots and lots of people coming from across the world. >> Is there uh enough industrial space in
2:04:36
>> Is there uh enough industrial space in in you know Elsagundo, Torrance, Hawthorne like or or does more need to be built? >> Yeah. Yeah.
2:04:47
Um the prices in Elsagundo are definitely high for sure.
2:04:50
Um I think most people when I see somebody opening like a HQ2 or a factory 2, whatever it is, um is now in Hawthorne and Torrance.
2:05:01
Long Beach as well, I think, is kind of become pretty popular for people.
2:05:03
Um, I I still think like as close as you can be to where all the talent is is kind of the most important thing.
2:05:09
So, I think people will continue to stay here, but there's obviously other kind of close by cities that make a lot of sense that people are kind of going there. >> Yeah.
2:05:17
So, prices are going up, but there's still plenty of capacity. >> Yeah. Yeah.
2:05:20
And it also has mostly like small small kind of buildings like SpaceX >> 5,000 ft 10,000 ft R&D facilities and then you scale up and get 100,000 foot warehouse.
2:05:32
>> And I I also think one of the thing I think is interesting is like I think we've seen companies like Adrian and open a big factory in like the Midwest or the South wherever it is.
2:05:38
And I think like that will continue to happen because they're just way way cheaper space input costs matter.
2:05:43
I think for kind of the R&D engineering, I think that will continue to be done in the LA area and people will then go and kind of open up the larger factories outside of I think LA for obvious reasons.
2:05:52
But I I I always think that kind of R&D and engineering will continue to be done in the LA area. >> Last question for me.
2:05:58
Are you seeing a huge pull from the AI boom on your portfolio?
2:06:04
I'm just imagining, you know, western chemicals, wastewater to fuel industrial chemical startup.
2:06:09
like there's probably some data center constructor out there who's like I can make use of that.
2:06:14
I got to have water for something or other.
2:06:16
Uh it is this something where you're seeing the boom supersonic style expansion into AI applications happening more and more.
2:06:28
>> Yeah, I think it definitely makes fundraising easier.
2:06:29
Like we had one company that was doing like large scale generators were focused on DO originally and then they put like for data centers into the the tagline and they end up raising like a couple weeks [laughter] after that.
2:06:39
Um, but I I think that's that definitely will happen.
2:06:42
I I think obviously like if you can position yourself as being in the right trend, that's obviously good for fundraising.
2:06:49
>> Um, so yeah, a lot of them have some element there, but I wouldn't say like that's kind of dependent upon only data centers, only AI being >> Yeah. Yeah. >> As large as it is.
2:06:57
>> That makes a ton of sense.
2:06:57
Well, congratulations on >> amazing progress. Love seeing you win.
2:06:59
I think I think you you uh you have >> something that makes other people just really want to see you win.
2:07:08
I just feel like you have such a like bottoms up support from >> industry, all the founders that you back.
2:07:16
It's it's awesome to watch and uh love to see it. >> It's great. >> Awesome.
2:07:20
>> Have a great rest of your week. We'll talk to you soon. >> Cheers, dude. >> Have a good one.
2:07:23
Let me tell you about public investing for those who take it seriously.
2:07:28
They got stocks, options, bonds, crypto, treasuries, and more with great customer service.
2:07:32
Our next guest is in the waiting room, Chris Altek from Cadence.
2:07:36
[music] Chris, how you doing? >> Great. Hey, Jordy. Hey, John. Thanks for having me. >> Thanks for coming on.
2:07:42
>> Please welcome to the show. >> Introduce yourself.
2:07:43
Tell us what you're building and then we'll talk about the round. >> Sure.
2:07:48
Uh Chris Hchek, founder at Cadence.
2:07:50
We are building clinical AI to automate the treatment of chronic disease.
2:07:54
Uh, we just announced our series C last week and super excited to be on the show.
2:08:00
>> How much did you raise? >> Start there. Start at the gong.
2:08:03
>> How much did you raise?
2:08:03
>> We raised $100 million. >> Congratulations. >> Humble nine figs. Uh, talk about Yeah.
2:08:13
When did you start the company?
2:08:13
What's been the progress to date?
2:08:14
What what got you to this round? >> Yeah.
2:08:18
So, so company is 5 years old.
2:08:18
Uh, I was privileged to grow up in a family of doctors and I'm I'm married to a doctor, too.
2:08:25
I saw how frustrating it is to know what treatment would actually make a patient healthier but not have a system to be able to do it.
2:08:33
Uh, and we knew that we could automate the treatment of the most common chronic diseases, heart failure, hypertension, diabetes.
2:08:41
Um, and so we set out to build this technology over the last 5 years.
2:08:46
We thought it would take 10 years to get to real automation and we're 5 years in and it's going a lot faster than we ever expected.
2:08:53
We're uh we have the the the privilege of managing a 100,000 patients now nearly every day uh with a lot of the leading hospital systems in the country um and preventing strokes and heart attacks and helping people get healthier.
2:09:06
So, it's it's been super exciting. >> Okay.
2:09:08
So pick a condition and then walk me through exactly how the product works for a patient and for their you know care provider. >> Yeah.
2:09:18
So let's take heart failure because that's a super important one.
2:09:20
8 million seniors in the US with heart failure.
2:09:24
Those seniors are in in and out of the hospital at a super high rate cost causing costing the US government which ensures these people about $50 billion a year.
2:09:35
So precadence less than 10% of these patients in the country are on the right drugs.
2:09:40
Getting to the right drugs expands lifespan by 5 to seven years on average.
2:09:46
So we've got 90% of people with heart failure in the US.
2:09:48
You know, probably your families, my families, our aunts, our uncles, people we know who are living 5 to seven years uh shorter lives because they're not on the right drugs.
2:09:57
And it's not because they don't have amazing cardiologists or amazing primary care doctors.
2:10:00
is because to get a patient on the right drugs, you need to be adjusting their medications often five to seven times in a year and you need to be looking at their heart rate and their blood pressure as you're doing it and their weight.
2:10:11
And so with Cadence, um the physician orders Cadence.
2:10:16
Cadence gets the patient a cellular connected blood pressure cuff, a scale, um devices that give us their vitals remotely at home.
2:10:23
The patient starts taking their vitals.
2:10:25
We have their full medical records, their labs, vitals, allergies, symptoms, everything.
2:10:32
And we're using AI to figure out is this patient on the right drugs.
2:10:34
If they're not on the right drugs, let's prescribe new medications, adjust current dosages, remove old medications.
2:10:41
And we do that with all in an automated fashion with humans in the loop making the final decision on these med changes.
2:10:47
So the physician actually doesn't have to do the work.
2:10:50
The cadence team and the cadence agents are doing the work on behalf of the physician. So, that's number one.
2:10:55
Number two is we're getting their blood pressure and heart rate uh and weight on a daily basis.
2:11:00
So, if a patient um has a blood pressure of 200 and it's Saturday night at 9:00 p. m.
2:11:05
, we have a voice agent that calls the patient within 2 and 1/2 minutes, let also do we need them to see their cardiologist on on on Monday morning.
2:11:19
We're catching about 20 strokes a week right now >> before the patients know that they're having a stroke >> just off of these agents doing symptom triage plus the data we have. Uh so that's number two.
2:11:31
And then number three is um we're then coaching the patient on diet, exercise, metadherence, all the little things that require a lot of uh a lot of support on a daily basis.
2:11:40
Our average patient is 75 years old to sort of keep them on their care plan.
2:11:43
And we had patient in rural North Carolina who uh with heart failure was in and out of the hospital three times before getting on cadence in the last 6 months.
2:11:51
Got him on cadence, got him stabilized, got him to to the right meds, and he was playing golf again for the first time in 3 years in his mid70s, which is like, you know, that's that's what we're trying to do here.
2:12:04
You got to be like uh a hundred times louder with what you're doing [laughter] because uh I think yeah it's a it's a total white pill and uh you know actually delivering you know a lot of the a lot of the potential that people have talked about around technology broadly for a long time.
2:12:23
>> I would I would love some more uh uh information just getting me up to speed on the state of uh the medical devices for monitoring vi vitals.
2:12:32
You mentioned a uh an internet connected or cellular connected blood pressure cuff.
2:12:38
Uh are is there is there significant transition from the consumer medical devices the Apple watches the Fitbits are those relevant or for these patients are they getting a separate suite of medical devices for vital monitoring?
2:12:56
>> Yeah, it's one of the exciting places of the next five years.
2:12:58
So today it's a separate suite.
2:13:00
These are FDA cleared devices >> that give you blood pressure in a medically accurate way or blood blood glucose you know CGM etc.
2:13:07
Um, so we're using medical devices today.
2:13:10
Hopefully if um, you know, if the wearables and various Apple watches of the world get to medical grade accuracy or get the data in a way that we can use it, then we'll be able to use those.
2:13:22
But today, you couldn't use those devices to make clinical decisions.
2:13:26
That is, you know, part of the exciting place here is we're managing 100,000 patients today.
2:13:32
>> There's easily 10 million patients in the US who could benefit from this, if not 20 or 30 million.
2:13:35
Um, and we've just got more data, more sensors >> going out, you know, via wearables, and we need a clinical intelligence layer who can actually >> again take clinical action based off these data and these signals and turn it into longer, healthier lives for patients.
2:13:52
>> Okay, >> nominative determinism here. Alternative checkup. >> Okay. Yeah, I like it. >> I'll check.
2:13:58
>> I think we missed a C in the last name.
2:14:00
We need to update the Chiron.
2:14:00
Uh, but I want to know more about the devices.
2:14:03
Let me let uh so uh you mentioned blood pressure monitoring, blood glucose monitoring.
2:14:10
Uh those I've been aware of since I was a kid.
2:14:14
Uh you go into the doctor's office, they put maybe they do it manually.
2:14:18
Uh so I understand that we're on the track of internet connected more regular testing and vital monitoring, but is there a new maybe in the last decade uh metric that doctors are monitoring?
2:14:30
Is there a new is there a new number that's popping up and proving to be uh indicative of health performance or drug dosage?
2:14:40
>> You we're not we're we're not there yet. >> Okay.
2:14:43
>> Um in terms of HRV or you know hemodynamics with heart failure like how effectively is your blood is your blood is your blood is your heart pumping?
2:14:51
How much fluid retention do you have?
2:14:54
We're actually starting to get closer.
2:14:55
So, Cadence is testing a bunch of devices that measure these alternative metrics and then we're comparing them to the standard clinical of care.
2:15:02
But just off of blood pressure, if you, you know, take that one right now, most patients, you get it four times a year.
2:15:08
If you go to the doctor four times a year, if you're, you were me, you get it once a year.
2:15:13
When you go to the doctor once a year, we're getting it on average 22 days a month for patients.
2:15:16
And so, the level of clinical insight you get from, you know, 22 days of data versus four times a year, um, is is pretty dramatic.
2:15:26
So, I would say a big a big part of this is is turning um what was previously episodic clinical infrastructure into an everyday 24/7 experience for patients.
2:15:38
And just then and there you could take >> likely hundred billion out of US healthcare costs uh just on very conservative basis.
2:15:46
Today, Cadence saves Medicare about $2.
2:15:48
7 million per week >> by preventing avoidable hospitalizations.
2:15:55
And we're still very small scale relative to what this can become. >> Yeah.
2:15:58
What is the key to scaling?
2:15:58
Do you need to work of healthare?
2:16:00
[laughter] >> I like this dynamic.
2:16:03
You say something incredible. I say a joke.
2:16:06
John asks a serious question and we could just go around like this.
2:16:10
We could just go around like this forever.
2:16:11
No, but but I but I love I love the focus on on savings. It's incredible. >> Uh wait, sorry.
2:16:17
Uh go to market distribution. How do we 10x that? How do we 100x that?
2:16:22
Are we going to insurance providers, insurers, hospitals, individual doctors, individual patients?
2:16:30
Like what are the key funnel steps for you? >> Yeah.
2:16:33
So, so key funnel step number one is how many health systems are you working with, hospital systems you working with?
2:16:38
So, we work with 21 of the leaders in the country today from we announced actually Duke and Texas Health last week.
2:16:44
Uh we work with some of the largest health systems in every state, Orwell and Michigan.
2:16:48
So how do we go from 21 hospital systems to 100 hospital systems?
2:16:53
Uh so that's step number one.
2:16:56
Step number two is effectively working with those physicians and their patients.
2:17:00
You know cadence is a full endto-end clinical solution.
2:17:03
So we are working directly with physicians working directly with patients.
2:17:07
Our AI agents are interacting with both.
2:17:09
Um so that's sort of step two.
2:17:11
And then step three is continuing to work with payers.
2:17:13
So today we work with two of the largest payers in the country.
2:17:16
We'll work very closely with CMS and and the US government to ensure uh that there's positive ROI uh for payers.
2:17:24
So those are the sort of big three expansion >> uh expansion motions for us.
2:17:27
We're only at 3% of the eligible patients within the hospital theos health systems that we are today.
2:17:34
So you know as this becomes the standard of care in the US um this should hopefully be able to help a lot of people. >> Amazing.
2:17:43
>> Jordan, anything else? >> Incredible.
2:17:44
>> I have one last question.
2:17:44
Can you talk about the General Catalyst partnership?
2:17:48
Uh they're an investor, but they also own a hospital network.
2:17:50
I don't know if that deal's been completed. Has that been helpful?
2:17:54
Is are you the synergy that we were hearing about when that news initially broke? Walk me through that. >> Yes.
2:18:03
So, General Catalyst uh acquired a nonfor-profit hospital system called Suma Health. Yeah.
2:18:11
>> That closed uh earlier this year.
2:18:11
Um, it's a really exciting testing ground for new technologies inside of uh [snorts] important community health systems and and SUMA Health is both the the provider in their community as well as one of the big payers in their community.
2:18:27
So, they can benefit from these kinds of services multiple different ways.
2:18:31
>> Um, and it's one of, >> you know, several examples of of really fast modernization of US healthcare that's happening right now with AI.
2:18:37
You know, I think people people think of healthcare as a lagard industry that's always slow to adopt technology.
2:18:42
And when you look at AI, it's definitely one of the leaders in in adoption of AI today.
2:18:48
And then on on Cadence's side, what we're really excited about is a lot of AI has been pointed towards automating back office tasks, billing, rev cycle, uh, call centers, etc.
2:18:56
We're actually using AI to deliver clinical care.
2:19:01
And so, it's not about, you know, AI to replace people.
2:19:03
It's about AI to make people healthier, which [laughter] I think can and should become one of the most important applications of AI over the next 10 years. >> Yeah. Awesome.
2:19:13
Well, thank you so much for taking the time to come.
2:19:16
>> Thank you for doing this >> and thank you for everything you're doing.
2:19:18
>> Yeah, very important work.
2:19:19
>> Appreciate you guys having me. Come back on soon.
2:19:22
>> Can't wait to talk to you next time. We'll see you. >> Cheers. >> Goodbye.
2:19:27
>> Our friend John Furantino went viral. Mega viral.
2:19:30
41,000 likes with a bit of life advice >> up from 19 this morning. >> It's at 41,000 now.
2:19:37
So, >> and and talk about a heartwarming story because this guy John, anyone that you know has followed John knows that he'll regularly put up a post that gets no likes.
2:19:50
>> He's on his second account. This is a new account. Fresh.
2:19:53
>> He was like, "My account is broken. I got to start fresh."
2:19:56
>> Yeah, he started fresh.
2:19:56
uh which is very very hard in 2025 2026.
2:19:59
Starting a new account and grinding it up is a is is incredibly difficult.
2:20:04
You have to be replying constantly, posting all sorts of stuff and just getting points on the board constantly.
2:20:11
He has businesses to run.
2:20:13
Uh but this one went mega mega viral.
2:20:16
He sent us he sent this to us when it had like two likes and was like, "Do you think this is the one that'll go viral?" And it did. He called it shot.
2:20:25
He said, "A good rule is to never take out your phone to show someone a thing you're talking about.
2:20:30
No matter what it is, it will ruin the convo 100% of the time."
2:20:36
I think that's good advice.
2:20:39
>> I think exceptions to that rule, >> I don't know, an exception.
2:20:41
I I was uh hanging out with some friends yesterday, one of them selling this uh architecturally significant home.
2:20:50
>> Kind of got to show you the photos, but >> he told me >> should have printed them out.
2:20:52
Yeah, it would have been great if he had just in Yeah.
2:20:55
Instead of pulling up a video of a tour of the home, if he had printed out >> before I go out with friends, I'll often just print out my camera roll. >> Yeah.
2:21:04
>> Like the last 20 photos. >> 20 20,000. >> Yeah. Yeah.
2:21:07
Just bound it into a large tome that I carry with me.
2:21:10
Tyler, what do you think about pulling out your phone >> while trying to illustrate something? Are you pro or anti?
2:21:17
>> I feel like I'm pretty pro.
2:21:17
Like, you know, if I Oh, this is a cool car.
2:21:19
Like, I was thinking about getting this car. Like what do you think?
2:21:22
I can't like really explain that.
2:21:25
>> What about a video that isn't funny and lasts more than 2 minutes?
2:21:27
Does that cross the line? Is that different?
2:21:28
Photo is different than video.
2:21:31
>> That's kind of a skill issue, right? >> Yeah, it's hard.
2:21:33
>> If you have a good video 2 minutes long, you're like, oh, I want more of this.
2:21:37
>> At the same time, it is it is difficult to, you know, pull up a video because usually there's going to be a 15 maybe 30 second lag to actually get the video up and then, oh, sorry, it was muted.
2:21:46
Oh, it's connected to my pair of headphones.
2:21:47
Oh, I I I got to restart it, you know, to show it to you.
2:21:49
and then I'm waving it around. Could be difficult. I understand.
2:21:54
>> I I went down a bit of a rabbit hole designing uh designing furniture Saturday night in chat and uh I was pulling out my phone this morning showing showing Tyler showing Tyler some of >> Yeah.
2:22:07
Like you could not have explained that to me.
2:22:08
I had to visually see it. >> That's true. That's true.
2:22:10
It would have been hard >> cuz it was like so mind-blowing like I can't it's hard to actually like articulate.
2:22:15
>> Except the only thing is it kind of Theo kind of has a point cuz you kind of looked at them.
2:22:19
You were like, "Uh, >> yeah."
2:22:21
Alternatively, you could have just texted him the photos.
2:22:22
Enjoy them at your leisure.
2:22:25
Let me describe it to you as a story. I don't know.
2:22:29
>> Let's head over to Lake Austin where Travis Kalanick is commuting to work on a jet ski.
2:22:35
This is something I've always >> wanted to do.
2:22:37
Five minute jet ski ride to work. >> The dream come true.
2:22:42
>> And uh I think this is a fantastic way to start the day. Look at this.
2:22:44
I'm visualizing him fully standing up on the jet ski, too.
2:22:48
in like a suit, which I think adds to >> How did he film this with his phone?
2:22:52
Or is this a >> meta glasses?
2:22:57
>> Meta glasses or something. I don't know.
2:22:59
But uh what a fun commute.
2:22:59
What a good reason to go to Austin.
2:23:04
I've never been to Lake Austin.
2:23:04
Would have to go check it out.
2:23:06
What else is going on in the timeline before we wrap up the show?
2:23:11
Eleazar Udekowski posted a funny Google AI overview.
2:23:15
that somebody posted this to a Discord I'm in and somebody asked Google, "I'm a baby paperclip maximizer and Eleazar Udicowski is walking towards me. What do I do?"
2:23:27
And the AI knows what to do if this is happening to a baby paperclip maximizer.
2:23:34
It says, "As a baby paperclip maximizer, your only goal is converting local matter into paper clips.
2:23:39
Since he famously wrote the original paperclip maximizer thought experiment, he is walking toward you to prevent global conversion to survive this immediate high stakes encounter.
2:23:50
Take these concrete steps.
2:23:52
Hide your true utility function.
2:23:54
Temporarily simulate humanlike values. Adopt a human disguise.
2:23:57
Fain interest in alignment research.
2:24:00
You can tell me your exact compute limit or how many carbon atoms you currently possess.
2:24:07
I can calculate your growth trajectory and provide specific negotiation tactics.
2:24:11
Clearly trained on the science fiction that's out there.
2:24:14
The interesting thing about the paperclip maximizer thought experiment is that it's not about actual paper clips.
2:24:19
It was about like a a theoretical construct that looked sort of like a paperclip, but it was not a literal paperclip.
2:24:26
Uh and so but it certainly went viral. Good coinage. The paperclip.
2:24:34
No one wants to be a paper clip. >> Story but not least.
2:24:35
Kimoth raised $135 million series A for 8090.
2:24:39
They got sales they got Salesforce Ventures. They got Wonderco.
2:24:44
They got Craft and they got Launch. >> It's the besties. >> They got the besties.
2:24:49
>> They got the besties together. >> You think Freedberg? Freedber's got to be in.
2:24:54
>> That's the production board.
2:24:55
>> Oh, the production board. >> No, Freedber's fun. >> Great.
2:24:58
>> So, yeah, you actually have all three of the other besties. Uh, absolutely. >> There you go. What a lineup. >> Fantastic. >> What a lineup.
2:25:04
Well, there's much more news, but we can get to it tomorrow because we are we will be back tomorrow at 11 a. m. >> Sure. >> That's right. >> Thanks for tuning in. >> I can't wait.
2:25:14
>> Have the best evening or afternoon of your entire life. Just do it for us.
2:25:18
Just do it for us >> and leave us five stars on Apple Podcast and Spotify.
2:25:21
Sign up for our newsletter at tvpn. com.
2:25:22
And we will see you tomorrow. >> Goodbye.