Open-Source AI Battle, Google Throttles Meta, Micron Margins Moon, Rocket Lab buys Iridium

I see a large IP on the horizon.

You're surrounded by journalists. Hold your position.

Overnight success.

The laser 3000.

Double leg.

That's misinformation.

Clearing order inbound.

Let's just roll.

We are surrounded by journalists. Hold your position.

Bump it up.

Trust the extra.

Five.

We are extra.

Founder.

Very good.

I see multiple journalists on the horizon. Stand by.

UAV online.

Blaze.

Double blaze.

Triple glaze.

Double kill.

Cop is.

Wait.

Team Deathmatch.

We are experts.

Triple blades.

Let's just roll.

Right.

Marky clearing order inbound.

Come on, get up.

We are surrounded by journalists. Hold your position.

Strike two.

Activate golden retriever mode.

Mark clearing order inbound.

Five.

I see multiple journalists on the horizon. Stand by.

Friendly IPL in triple glaze. Another one.

Yeah, you're watching TBPN. Today's Monday, June 29th, 2026. We are live from the CVPN Ultradome, the temple of technology, the fortress of finance, the capital of capital.

Let me tell you about Ramp.com. Time is money. Save both. Easy to use corporate cards, bill payments, accounting, and a whole lot more.

All in one place. I got to adjust my IM.

Well, on the front of the Wall Street Journal today, this is how you know this is the whole AI 2027 Washington waking up. The AI stories are making it to the front page, the world news section, not just the business and finance section more and more. So the very front page of the Wall Street Journal, of course the picture is about the heatwave but the lead, the story with the largest text is about artificial intelligence. China resets the AI race with the United States as security models mark gains. We're going to get into it. This is a fascinating debate because I thought that we'd have a conclusion to the open source AI debate by now. Either the frontier would have collapsed and there would be, you know, perfect commoditization or they would have fallen.

A conclusion, 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 the story completely, 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. We are going to be talking about a first iteration on-device PII redaction model that is far smaller than existing models.

It's actually tiny. It's 15 megs, which is, and you can do it in the browser.

And we have Chad Rigetti. He's coming on to talk about a whole lot of quantum mumbo jumbo.

We'll see what's going on there.

And Pim's coming back on from General Intuition. And we got a bunch more founders coming on. Jakob Diepenbrock announcing a $30 million oversubscribed fund with tons of TBPN guests already in the portfolio. The rest of the portfolio soon to be on the show, I'm sure. Anyway, open source AI. So the big story is centering around GLM 5.2 from Z.AI. It was officially released June 13th. So it's taken a couple weeks for it to really break through to the front page of the Wall Street Journal, but there's been some strong performance on benchmarks, some positive reviews from developers. I have a whole review from Tyler we can go through in a little bit, but we're now entering another round of debates around open source AI. What can the model actually do? Is this a threat to national security? What are the geopolitical ramifications here? And so I'm sure this will be an ongoing conversation throughout this week, probably next week. We have some guests lined up to help contextualize it. But laying down the facts from the journal, security researchers said that a new AI model released this month by China's Z.AI, also known as Z.A.I., can match the latest US models when it comes to finding security bugs. A development poised to reset the global tech race and pressure the White House in its overhaul of US AI policy. So unlike models from Anthropic or OpenAI, Zhipu's GLM 5.2 is open weight. You can just download it, run it anywhere. You don't need to go to an API. You don't need to go to a private company and pay them. You can run it on your own server, provided you have the electricity and GPUs to do so. It is expensive to run as we'll go into, but it is open weight. 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. Open weight models are ideal for users who want unfettered access to systems they control. But they're also ideal for hackers who want to run them in the shadows.

Unfettered intelligence.

Unfettered. Oh, that's completely out of, we were completely out of names for new Neolabs.

That's a good Neolab name. Yeah.

Unfettered intelligence. Intelligence is good. GLM 5.2 has ranked as one of the top 10 most used AI models according to data from OpenRouter, a company that provides access to more than 400 AI models. And what a fantastic business. Alex Atala over there absolutely cooking at OpenRouter. It's such an exciting way to plug into the AI race without actually needing to play the benchmark game so much. Be the front door. Anyway, in some benchmarking tests, according to cybersecurity company Semgrep, GLM 5.2 bested Anthropic's Claude Opus 4.8 model, which was released in May. When given further instructions, Opus 4.8 and GLM 5.2 can match Mythos in bug finding ability according to researchers. So prior to this launch, and there's a chart that we should pull up here about overall AI capability. 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. And I saw a lot of American AI fans sort of cheer for this. Hey, we have the capital markets, we have the data centers, we have the researchers. And so we are able to push the frontier at a different rate. 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 AI industry. Now this chart sort of goes back and forth and there's some debate over it. It's in the newsletter. You can go sign up at tbpn.com. While we're pulling that up, let me tell you about Codex. Codex is a powerful workspace for getting work done with AI agents. Whether you're writing code, analyzing data, creating content, or automating business workflows, Codex helps you move projects forward from start to finish. So this chart, which we can pull up, shows progress from GPT-4o to o1, o3 mini, o3, Opus 4, GPT-5, 5.2, Opus 4.6, GPT-5.4, GPT-5.5. Showing a, you know, linear trend in this ELO, which is a blend.

It says GLM 5.2 sounds too much like a gray market peptide.

It actually does. It does sound a lot like that. And then you can see the red line are the Chinese models which are also improving over time but at a slightly lower rate. And so the question was, are they going to plateau while America's progress continues to advance? And this latest model GLM 5.2, it 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 CAISI, I think that's how you pronounce it, the Center for AI Standards and Innovation. They have this way to calculate like the ELO of a model. It's basically a kind of approximation of a bunch of different benchmarks. Some of those are proprietary, like they're not open, so it's actually hard to run these.

Also because I was basically trying to bench like all the recent models since this was published. It was, I want to say, May 1st.

Yeah, it'd be great to throw 5.6, Soul, Mythos, and Fable. It would be great to just continue this chart because it's an interesting trend.

So a lot of those benchmarks aren't actually public. So it's very hard to estimate. But I tried, you can look at some of the benchmarks that are public that you can reference. You can kind of match them up to previous models.

Um.

5.2 looks like it is a big step up from the Chinese trend line, right?

But even then, I think it's hard.

I think that the group of benchmarks that were chosen for this ELO definitely accentuate the gap between US and Chinese labs.

I think there's a bunch of other groups like Epoch AI has done a chart. They've basically seen a relatively stable gap between closed source and open source models.

Since like 2023, like a long time.

Yeah, and perhaps at this point the discussion should be more centered around cost per task more than cost per token.

Yes, definitely. Because even like, you know, new models a lot of times when they come out, like okay maybe the token price is actually the exact same but the token efficiency is much better. Then when you do a lot of these tasks, it's not the price per token, it's price per, you know, something completed and then you actually see it go.

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 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.

Yes, most people are feeling like 5.2 is very token hungry, right? So it uses a lot of tokens. So maybe it definitely is much cheaper than the frontier models.

On a per token basis.

But on the per task basis it might be more expensive.

Yeah, I mean that's still generally not. But 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.

Well, let's revisit John Luddig's post from May 2024. This was pre-DeepSeek, talking about his prediction about why the future of foundation models is closed source. He got a lot of pushback from this because a lot of people like open source models, but he laid out a thesis around closed data, closed source data flywheels, exponential capex intensivity of training. And he said open source will have a home wherever smaller, less capable and configurable models are needed. Enterprise workloads, for example. But the bulk of the value creation and capture in AI will happen using frontier capabilities. The impulse to release open source models makes sense as a free marketing strategy and as a path to commoditize complements, but open source model providers will lose the capital expenditure war as open source ROI continues to decline. And that was the thesis around the time that the open source AI discussion was primarily driven by Mark Zuckerberg's work at Meta on the Llama family of models. The idea was that Meta would benefit from attracting talent. It was good marketing. It told the story that Meta has an AI story and has AI talent in-house, even if they weren't monetizing it and sharing a really fast takeoff in ARR around those models. It showed that hey, they're able to develop these models and that might help them cut their costs in the long term. Very interesting that that 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. But then China sort of woke up and the DeepSeeks launched at the start of 2025. And the game theory became way more complicated. So George Hotz sort of summed this up nicely. 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. He says this explains why the Chinese are giving much more moderate resources to train models away for free. They love to see deflationary economics in the US. It is much less of a service-based economy. 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. In his formulation, 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 likens what's happening in DC to sort of rearranging deck chairs on the Titanic. It's a very fun piece. But 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 Amodei when he was testifying in front of Congress in 2023. 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 from 3 years ago but some of his predictions were very prescient as of where the frontier is today. So he said, 'I'm very concerned about where things are going. 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. But he's talking about 2025, 2026. Remember he was saying this in 2023. We're there now. '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. And again if the path continues I think we could get to a very dangerous place.' He was worried about cybersecurity and biorisks being open sourced and then not having a counterweight to that. Now the good news is that we've talked to the CEOs of cybersecurity firms like CrowdStrike and Palo Alto Networks and they've been working with Mythos and GPT-5.5 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. 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. So security stances must adjust. It's not that closed source is falling behind, so it's never going to be an issue. There will be this gap and how the American cybersecurity industry and eventually the biocurity industry implements changes and fixes before open source catches up or commoditizes and makes that particular capability widely available is going to continue to be important. So let's go over to Tyler's quick review of GLM 5.2. Why don't you take me through your bullet points that we shared in the newsletter at tbpn.com and you can tell us like what is the shape of this model? How are the reviews?

Yeah. So I think so far one of the main things is like people are saying it's distilled, right? This has been a big thing with a lot of these open source models, especially the Chinese ones. Oh, the only reason that they're good is because they're distilled. It's very hard to actually figure out how true this is.

But people are, you know, it certainly seems like there's some, you know,

I think aspects of Anthropic models. Didn't Anthropic openly accuse Alibaba of distilling a number of these labs?

Yeah. And there's also been a big professionalization of the gray market where a whole bunch of different individual groups will 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. 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. Is that not something people can do? I don't understand.

Yeah. 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 public GitHub repos and those were all made with open source models. 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 because they're just basically taking the closed source labs.

There's also this weird thing with distilling where as more and more of the public internet and GitHub broadly and open source repos become LLM outputs, 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 going to wind up learning 'it's not this, it's that'. And the same thing applies for different code conventions in open source repos that have effectively been completely rewritten by closed source models.

Yeah. And so I think it's safe to say that like we've generally seen that distilled models generally will generalize worse, right? So you'll see really good benchmark scores.

Maybe they're benchmarks, maybe they're not. But even if they're not directly benchmarks, you still find that they generally

Yeah. They're kind of accidentally benchmaxing.

Yeah. Yeah.

So you should always, so I think initially you should just be a little bit suspicious of these super high benchmark scores.

Yeah.

Um.

But they lack that big model generalization.

Yeah. And this is like anecdotally reinforced. A bunch of people have been saying, you know, for coding, these models are really great. GLM is a very good model, you know, for creative writing or something like this where you'd imagine it's a bit harder to kind of benchmax this.

They'll perform a bit worse.

Yeah. I wonder, have people been testing it with the like Tiananmen Square bench? Like does it reject that stuff? Because it felt like that was something that was widely misunderstood by American audiences that in fact that might not be the biggest deal for the CCP.

Yeah. So 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 it's a bit harder than that but I think you can kind of get around that kind of stuff.

Okay. Yeah. So we talked about the token hunger and the API price. 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. And if you look at OpenRouter, 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. So

Yeah, so I think like what we've seen is

Yeah, that's the job.

Like a marginal IQ point of the models is like extremely expensive.

Um.

Frontier models are getting very expensive. People have to cut back. You know, they're token maxing. This is like massive bill on their balance sheet, whatever. I think it seems like there's now basically like two classes of models that people really use. There's like the frontier ones.

And they're using coding agents. They need the best thing. If you're doing cyber like you just need the best model because you know the risk of someone hacking you is so great. You just need the best thing, you pay whatever it is.

Yeah.

And then there's the second class which is like these very small, very fast, very cheap models that you can use for these kind of point solution things. Maybe you have some orchestration where using a really big model to have these like little agents using these very cheap models.

Yeah.

I think in the middle it's hard to actually figure out what is the real use case.

Maybe it's like hobbyists using these coding agents and they don't want to pay the super expensive tokens of the closed source labs.

But generally and you see this on OpenRouter where like what are the top models by token usage, it's these very small models. It's like, you know, DeepSeek Flash.

You're just spamming them for like, you know, every receipt that goes into Ramp gets processed by an LLM at this point. Does it need to be a frontier model telling me that I spent $10 on a coffee? No. It can just do standard OCR.

That'd be my preference.

Yeah, you want super intelligence overseeing your expenses most likely. But no, you use the right tool for the job and that's clearly what's happening on

But also I think like it is a very good model, right? Like we should not fully dismiss

I think the idea that oh the gap is widening, we really don't have to worry about these open models. I think they are like very good.

Yeah. Yeah.

And maybe if you're super worried about distillation, maybe something changes if the models are kept to these big partners, right? Like what we've seen recently with government coming in. But I think we can't really fully dismiss these labs.

Yeah. It throws a little bit of a wrench in the monetization potential, like how long can you monetize a new frontier model, that's more tricky. And then the other one is just like if you're going to keep a model behind KYC or behind an approval for specific companies like the government has been sort of edging towards and moving towards. 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 GPT-7 or whatever, but by the time the government got back to me, my company got access to GLM-6 and it's close enough. And so 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, which has been yeah back and forth as always. Anyway, let me tell you about Shopify. 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.

Google, Meta's Gemini use as AI demand strains capacity in the Financial Times. Surging appetite for advanced models is turning computing power into the tech industry's scarcest commodity. And they have a picture here of a Google Gemini bicycle which looks fantastic.

What does that have to do with Meta though?

I think that was just the best Gemini picture.

It is hard. Otherwise, it's just a picture of a phone screen. I mean, you saw in the Z.AI, it's just a picture of the app, which is like so boring.

Imagine writing this.

Or it's the stock image of the brain with the neurons. That's always good.

What do you think? I mean this kind of ad placement.

Like on a, what do you actually call this? It blocks the water. No, no, no. That's just the part that blocks water from, you know, if you were to ride in the rain.

Yeah. Some type of fender thing.

Like a Mansory kit for a city bike.

Exactly. Exactly. But imagine riding that Gemini bicycle in the rain. Fantastic.

Google has.

That's what it's for. So the water doesn't come up and splat you. Interesting. Okay. I never knew what that was for.

You never knew that. No. I learned this is educational.

This, the experience of hosting the show is educational for both of us.

It is.

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. Google told Meta around March that it could not provide all the Gemini capacity the company wanted to purchase, according to three people familiar with the matter, in a move that has disrupted and delayed some of Meta's internal AI projects. So

I don't understand how much this is possible.

Yeah. Yeah. So one.

Google spent $200 billion on capex.

Okay. So of course like this.

Around this time token maxing was becoming a thing. 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.

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 can't fulfill that.

But I'm wondering how much more we should read it, like is it worth reading? I mean it sounds extremely bullish for Google like if this tracks with what they talk about on earnings calls.

Yeah. Yeah.

Um.

Yeah, Google Cloud acceleration crazy.

You do have to wonder like could distillation be part of this story? Could that be a factor here? I have no idea.

I don't know. Zero Hedge said Meta puts limits on Claude and Codex fearing distillation, The Information.

But so this story is different. This is Meta telling its own employees, don't use Claude and Codex in certain parts of our business because we don't want to accidentally do distillation is what Meta is saying. So that's different. I was wondering like is Google thinking like whoa that's a lot of, you know, cool it, you know.

Yeah. Owing to the restrictions which remain in place as well as broader push to streamline AI costs, Meta has encouraged staff to be more efficient with AI tokens. Several other Google clients have been affected by the restrictions although to a lesser extent. Meta has been particularly impacted because of its exceptionally high demand for Google's models.

Interesting.

Very interesting. I would love to see a pie chart like breaking down what all the different ways they're using Gemini.

Yeah.

In their business because

Google has not broken out Gemini revenue.

No. No.

At all to date. 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. So that's a search product probably insane token demand there, right? You've seen the chart of like they're in the quintillion or quadrillion tokens category. And then you have YouTube now has Gemini plugged in and you can chat with any video and transcribe it. That's got to be incredibly token heavy. And then you have Gemini app users and free users and paid users. So there's got to be a lot of just Gemini internal usage, but it's remarkable. Yeah, I would love to see that Meta pie chart because I thought that they were spending a ton with Anthropic. 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. And if you go into the Meta app, you now have access to Muse Spark. And if you go into Instagram and you search for something, it populates it with a Llama 4 result. And so I would imagine that even though that product has not broken through like crazy, I would imagine that it's still generating a lot of tokens just because of the scale of Instagram. Like Instagram has a billion, two billion users, something like that. It's huge. And so even if it's people sort of accidentally winding up in an LLM powered workflow, it probably is generating a lot of tokens just because of the scale of that system. On the topic of Meta. Yeah.

Meta shared this morning.

What'd they do?

A new milestone.

It is a mind reader.

Mind reader.

Non-invasive brain-to-text decoder research. Brain-to-Text v2, building on v1 which was published today in Nature. Brain-to-Text v2 is the highest performing end-to-end pipeline capable of real-time sentence decoding from raw brain signals. Advances beyond character level performance to decoding words and semantics enabling accuracy for overall communication. So if you thought, you know, Instagram was listening to you, if you thought I was listening to your conversations, now you can have a new conspiracy at home which is that they might be just listening to your thoughts. Do you know the device? They say this is a non-invasive device. I just shared an image of this device and I want you to tell me, do you consider this non-invasive or invasive?

Look at this image of the magnetoencephalography device.

No, you got to go. You need to scroll up a little bit because you can't even see the whole thing here. Scroll up. It's not invasive because it looks like the device could actually potentially carry on for like a whole half of football.

It really does seem like it's just put yourself in this room-sized device. No, of course this will shrink.

I'm giving him credit here. Non-invasive.

Okay.

As long as he.

You're putting this thing on. You're daily driving this thing.

I don't know if I'm ready to daily it.

I don't know if I'm ready to daily it. Yeah.

This will be a cool demo. Like this will actually when you can just walk in, sit down in a chair and see your thoughts on a screen.

No, we were debating it earlier. My buddy Rob Taves has been on the show twice. Dropped five predictions in Forbes recently. We can go through them at some point. He's going to come on the show. But four of the five were very, very reasonable. You know, Anthropic's going to be bigger and, you know, TSMC is going to face more competition. And then he predicts that in 2030 telepathy will be commonplace, which is a very aggressive prediction in my estimation. You know, it's certainly not like a straight trend line since like, you know, TSMC has competitors right now. The prediction is just that there will be more competition. But truthfully, telepathy is not really existent outside of like a few demos like this. It's not really something where it's like, oh yeah, like 5% of people have the Meta Ray-Bans that take pictures. So like face cameras are going to be bigger in five years. And it's actually only three and a half years until 2030, which is sort of crazy to say, but we are getting quickly to the future. To the future. Never sell your company. Should you ever sell your company? David Sacks says no. He says the best founders in the world would never sell their company. You could never acquire Elon, Bezos, Zuck, Jobs, Ellison, Jensen, Dell, Page, and Brin. Scott Wu has turned down billions and keeps saying no. This is a great clip. Went super viral. I don't know, did I lose internet or something? I don't know. Anyway.

Tyler's app is in shambles.

I don't know about that. But there's some debate over this because Elon.

That's not my app.

Elon, yeah, this is just a, Elon did sell two companies. He sold Zip2 and he also sold PayPal. And then Jobs sold NeXT back to Apple. Does that count? I don't know. He did sell Pixar to Disney. That sort of counts. And I mean Elon never sells his company. He just sold XAI to himself, but I guess that doesn't count. But yes, it is a funny thing. Didn't Mark.

Yeah. So what's the, do you know the backstory here from Sasha?

I don't know. Tyler looked it up. Apparently there's a Business Insider report from the time that this happened in 2007. How Terry Semel fumbled Yahoo's Facebook deal. How much is Facebook worth? 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. As Yahoo continues its soul-searching, here's an unpleasant rendition of Semel's catastrophic decision, courtesy of Wired. When Yahoo came calling with a bid of 1 billion in cash, the pressure became too much. Zuck relented in July of 2006. He was just like 18 months into building the company, something like that. Verbally agreeing to sell Facebook to Yahoo. He said yes. He said he was going to sell Facebook to Yahoo allegedly. Strategically it seemed like a good match. Yahoo had hundreds of millions of users but its foray into social networking was struggling. Facebook had cool tools and was looking for a mass audience. The timing, however, could not have been worse. In the days after Zuckerberg agreed to sell, Yahoo announced it was projecting slower sales and earnings growth and that the launch of its new advertising platform would be delayed. Its stock price tumbled 22% overnight. Terry Semel, Yahoo's CEO at the time, reacted by cutting his offer from 1 billion to 800 million. He just took 20% off, but Zuckerberg, who had been warned about Semel's reputation for last minute renegotiations, walked away. And that's probably reasonable. I mean, if they're cutting the price there, you have to imagine that as it gets papered, you get cut down again. Then the earnout, you get cut down again, and all of a sudden, you're walking away with barely anything. But two months later, Semel reissued the original $1 billion bid. 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. He rejected the offer and became famous as the cocky youngster who turned down $1 billion from Wired.

Legendary. Legendary. It's so interesting to imagine the road not traveled there because the dynamic, the way Facebook is built as a social network, like could it have been successful under Yahoo's stewardship or would it have been less exciting, attract less talent, ultimately been disrupted and would they have had the capital and the guts to go buy WhatsApp and then also buy Instagram? You know, to actually maintain the dominant position in social networking. What do you think?

I think Yahoo should make another offer. We were hanging out with Jim, CEO, last week, dear friend of ours, and I would like to see Yahoo make another bid.

Hey, Meta's trading down. Just keeps going. If it continues at this point, it goes down to 99.99% might be able to pick up continues at this trend.

Anyway, let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI. Own the data platform that powers it. Moving on. What else is in the news?

Chipmakers are profiting off AI at the expense of just about everyone.

This is on the cover of the business and finance section today. 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.

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. Micron along with Korea's Samsung Electronics and SK Hynix 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. Soaring data center demand simply jacked up prices. Micron's soaring profits are for its customers soaring costs. We are witnessing an enormous transfer of cash. They said profit shifts 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. 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 single-digit percentage. It said last week prices for NAND flash memory, also used in data centers, jumped more than 80%. 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. Apple last week raised prices for MacBooks more than 15%. Closer to home, closer to home for me, the memory I bought on Amazon.com a year ago to build a super quiet computer — I hate fan noise — good, good color commentary here — has tripled in price and now costs more than the CPU. For an industry in which prices usually drop every year, it's a huge turnaround in consumer electronics. Passing on higher prices helps limit demand for chips, just as higher oil prices reduce consumption. But the AI companies aren't passing on higher prices because they are able to throw money at supply problems. 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. Everything is still priced to bring in new customers, yet not yet to make money. So higher input costs create a nasty problem: either losses will be bigger, or higher prices will be needed, putting off potential customers. And you can see the price of Micron's stock has been through the roof as the company joins the $1 trillion club and becomes the first trillion-dollar company headquartered in Boise, Idaho. And Idaho got a trillion-dollar company before New York, I believe, and also before Florida and Austin, maybe something like that. It's rare. It's rare.

Mostly on the West Coast. Anyway, there's a whole bunch of bull cases for Micron still. The stock could double from here, says Barron's. I love Adam Levine and Barron's sharing the bull case. Tyler, how many trillion-dollar companies are there in Europe? Out of curiosity.

I'm going to go with zero.

That's true. That is true. You are correct.

The other ASML could get there, maybe, sitting at around 700. Wait. What about Eli Lilly? Or no, Novo. Novo was a trillion, right? Oh, did it ever touch a trillion? I don't think so. Right. It was real close.

It's a humble 165 million.

Brutal. Wait, but you're thinking of Eli Lilly.

Oh, Eli Lilly hit. Yeah, rough. Very, very rough. Comcast is planning to split up the company. Competition is escalating.

Eli Lilly, the Indiana company. John, is it from Indiana? Yeah. Okay. Former Indiana startup. Okay. I like it. I like it. NBC, Universal, and Sky will separate the company's connectivity business from its film, theme park, and streaming operations. Oh, yeah. Universal Studios. Comcast is up on the news. Comcast plans to separate its media and connectivity business.

Who's building the Anduril of theme parks? It does seem like a... could there not be an opportunity to create a net-new theme park business with a modern technology stack?

It's very expensive. Everything needs to be like... the modern technology stack in parks is expensive. You don't believe in the theme park capital markets?

I don't know. I know people that have worked on theme parks at Disney, and it's tricky because you have to amortize a ride over like 20 years, and so you'll go... seems like an absolutely brutal business. Yeah. That is probably harder today because... I mean, think about, you know... Yeah. 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.

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 TikTok highlights, but people still forked over $5,000 to go see the game. 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 the NBA team. Yeah. To Vegas. But at the same time, there is enough stat that came out this morning or maybe yesterday that there's more sports betting volume than all sales of movie tickets, theaters, theme parks, and like a couple other of these IRL categories.

Is up or down? Less, lower. Less, like, and the stat was like volume. Yeah. Yeah. And so it's not exactly like a proxy for like revenue, but still meaningful.

Theme park vertically integrated Tweety Bird tattoos. Tweety Bird tattoo parlor right on site. That's a big thing. I like it. Six Flags. No, I don't... I like it. Six Flags never really got the same like cultural power that Disneyland did. There's something about the flywheel that Walt Disney laid out that does seem very, very important. And so, how do you start that? It's not just, oh, you know, tech-enabled theme park that's not going to draw people in. You need to have like IP around it. Brain rot theme park. There's something about... I mean, we've read stories about like the Disneyland fan that saves up every year and spends so much money at the park. And I think that's probably the lifeblood of that business. And that doesn't happen without building a whole cinematic universe around every single ride. And that just takes so much time. And you can't like... this goes back to the question of like, is Netflix enduring IP? Like they haven't been able to... even though it's been 20 years of like... I mean, I don't know when they started producing their own content, but 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. Well, I'd push back and say Narcos.

Narcos. You want to go to the Narcos theme park? I was talking about this with somebody once, talking about like HBO, like why don't they have a theme park? And he was like, what are you going to do? Take your kids to a brothel in Game of Thrones land? Like no, it doesn't make any sense. Like anything like it needs to be uniquely general audience. Like you can't have R-rated... 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 any theme park that's built around an R-rated like IP library. And so that just narrows it down even further.

Well, all of America's basically turned into a theme park for European soccer fans. Oh yeah, in the journal. European soccer fans marvel at the splendor of America's suburbs. You've been... many of these reels serve to me. Dutch fans in Missouri see a nation that is risky and expensive but vast and bountiful. Everything is three times the size. You've been... have you been seeing some of these people in real life, right? I don't know if I've seen any of them. I did go out to lunch like a week ago and it seemed crowded, but I was unclear if that was just local residents going out to watch the games or actual tourists coming to town to watch.

Gabe, in the X chat, I think Ferrari has a roller coaster in the Middle East. Do they have a whole Ferrari theme park in Abu Dhabi? Because that's not R-rated. You can take your kids to Ferrari theme park. Yeah. I was in Abu Dhabi and I was driving by it and I was like... Yeah. I was just thinking of like if you wanted to spend a day, you know, getting the Ferrari experience. Like you could just go to the track. Yeah. Or you could just rent a Ferrari. So I don't know. Yeah, but you don't need to go to Six Flags to get the Batman experience. So you can just go out in the middle of the night and arrest a criminal. Just become a vigilante.

I saw another report that apparently there's like an individual who's being like the Batman of Mexico. Do you guys see this? This is very funny. And so the guy went out and found criminals. Develop says Yas Island. They literally named an island Yas. Yas. Weird. No. Okay. I don't know.

Anyway, Dutch soccer fans are having fun visiting America. Frank Evering, he hadn't even heard of Kansas City. 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. Everyone got in his camper van and drove south from Toronto, making stops in Detroit, Chicago, and Indianapolis. 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. Suburban superstores, hulking plates of food, quiet streets. He marveled at the sprawling houses, a contrast from the tightly packed homes of the Netherlands. I did notice this when we were in France. The food portions were way too small for me. It was brutal. It's spacious. He said, 'You go here for your shopping and there for your dentist. People are so rich here. I think that's why they can be so nice.' What an ultimate white pill in America. 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.' And then one European tourist comes like, so nice.

Something about the grass. The grass is always greener, right? The grass is always greener on whatever side I'm on. That's what I like to say. 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 long-running transatlantic debate: who lives better, Americans or Europeans? The Europeans had plenty of thoughts on American culture. 'We are a bit shocked about the food you're eating,' the Dutch national team superfan Sandra Tate said. Fans also balked at the size of Costco and the vastness of the highways. In recent days, social media has been filled with videos of Europeans gawking at the staples of suburban life: a two-car garage, a walk-in closet, a second refrigerator. One Brit went viral for trying Chick-fil-A for the first time. 'That was absolutely banging,' he said. And another toured the inside of an American fire station.

The way that they experienced Chick-fil-A was me seeing the Renault Twizy. Yeah. I was just like, 'Wow, this is unbelievable. They made the perfect car.' Yeah. So small. So small. And it's the way they think about our fire trucks, which are massive. This is nuts, honestly. They said... Tyler, while we wait for our first guest, do you know anything about Bosnia's World Cup team? The United States is facing them on Wednesday. Do you have a stat breakdown or anything? Be very careful with what you said because I saw that there was a news reporter who faced fierce backlash for really calling Bosnia out and saying like, 'I don't know where it is on a map.' 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, but Bosnia will be playing,' and then she just 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.

Anyway, there's a little golf cart. We got to talk about this at some point, but there's a new car. It's like a Twizy. You're going to love it. It's close. It's no Twizy for you. Let's bring in our first guest. Anyway, let's bring in our first guest from the National Design Studio. Welcome to the show, gentlemen. How are you doing? Thank you so much for coming on the show. Please start with an introduction of yourselves, the company, and then the announcement today.

Okay, so open source, the weights are on Hugging Face, runs in the browser under 15 megs. A technical user could go right now, download the model from Hugging Face, vibe code their own Chrome plugin, and have it be running however they want. But how do you see this actually rolling out? Do you want the government to implement this in various places? Do you want companies to... is it sort of like open up the primordial soup of ideas and see where it goes, or do you have a rollout strategy that you are advocating for?

You created PII super intelligence. That's what people are saying online. Basically, it's like tiny intelligence. It's by far the smallest model. Like the other ones are like at least 50 megabytes. This is 15. Yeah. So, did you... to what degree did you build on the shoulders of giants? Is this some pruned and distilled and fine-tuned open-source model? Is this something where it was easier to just start from scratch but use architectures that are more prevalent and well-established? Like how did you actually go about training this model?

Can you help me understand use cases a little bit more? Because I feel like most of the time when I'm transmitting a document to a prescription website, Rx, or a financial institution, the PII is like the potentially the only important part. They're often sending me a blank form and asking me to put my PII in there. 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 estimation.

That makes sense. How do you guys think about side quests at the National Design Studio in general? 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. They have a mandate, but at the same time, there's so many different places that the government, you know, interacts with people's lives. I'm very curious.

And back on the show. Yeah, like diving more into that, do you have a reference point in tech? People might ship, you know, they might think in quarters, financial quarters, three-month cycles. They also might think about a two-pizza team, which I think is like 10 people. Do you have an idea 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?

What's your pitch to talent that you might want to recruit into the National Design Studio? I imagine lots of people that would join could get a blank check from a venture fund or could go work at some of the best companies.

Let me tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. So, Apple and Audi alumni just unveiled a $25,000 open-air electric neighborhood vehicle. It's called the Amble 1 and it's a street-legal EV built for short local trips. No doors, fewer screens, modular design inspired by the 1960s Lunar Rover. Goes 40 mph with 60 miles of range. Weighs under 1,000 lbs. Takes 5 hours to charge. Rear seats fold flat for cargo, surfboards, or gear. Built-in mounts let you add baskets, straps, mirrors, and cargo accessories. Already has 500 vehicles committed. I love it. You love it? I love it. I think it's great. Give it a Jordy score. A Jordy score. Daily weekend. You know, the Doug score out of 100. What are you doing?

So, I mean, I just went through this whole crazy search for basically this exact vehicle. Didn't find it. I don't like the aesthetics of golf carts. I've driven a lot of carts in a commercial job in college. I've owned a golf cart. In my experience, it's impossible to feel cool while driving a golf cart. So I wanted something like a golf cart that was more like, not, you know, I'm not golfing. So I wanted some like little bit of utility, wanted to be fun, etc. I landed on a Can-Am HD11. Mhm. You know, a UTV. It's gas powered. It's quite fun, but the gas element is actually kind of annoying even as an ICE, you know, defender that is the internal combustion engine. But I think this is fantastic and I think I saw somewhere that they're going to focus on more commercial opportunities. So going to hotels all over the world.

That's what Justin says here. Says this little golf cart is going to be huge for hospitality, all electric, $25,000. 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? Can you get a fleet of golf carts for a discount? What is it typical? Like a golf cart is going to come in at like 13 and a half grand.

So I mean, and it depends. There's commercial golf carts, maybe you get bulk deals, something like that. But no, I think this is going to be great. I think it's going to be a nice amenity on hotel properties around the world. Ryan Deini says he thinks it'll be a hit in hospitality since Moch caps sales at 500 units a year. I did not know that. That's interesting. But I think this is going to be a hit. Meyers Mans, I much prefer the sort of aesthetics of the Meyers Mans, you know, the sort of more like dune buggy style. They're coming out with an EV that I'm very excited about. But I think this is great. I'm excited to have more people building cars for recreation. And I talked to Riley Brennan who is a GP over at Trucks VC. They just invest in like automotive startups. And so we're working to get the Amble team on the show ASAP, hopefully this week. Very fun.

There's a good quote from Roger Ebert, the famous movie reviewer that we got to share. Oh, the team loves Roger Ebert from Siskel and Ebert back in the day. Anime Outsider says, 'I don't care what he thinks about video games. Roger Ebert had the ultimate red pill on nerd culture as a whole. This basically describes every fandom on Earth, and once you see it, you can never unsee it.' He says, 'A lot of fans are basically fans of fandom itself. It's all about them. They have mastered the Star Wars or Star Trek universes or whatever, but their objects of veneration are useful mainly as a backdrop to their own devotion. 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. Extreme fandom may serve as a security blanket for the socially inept who use its extreme structure as a substitute for social skills. 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-lib. Your Spanish obsession is your beard. 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. That's why it's excruciatingly boring to talk to such people. They're always asking you questions they know the answer to.' What a funny... It's like you and your Apple Vision Pro fandom. 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.' Is that true? I'm not that much of the dino experience. Anyway, let's bring in Chad Rigetti from Rigetti Computing and Sigil. Chad, how are you doing?

We're doing fantastic. Thank you so much for taking the time to come chat with us. I would love to start a little bit with your background and your journey. 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. I think that might be informative. There's a lot to talk about there, and of course, it relates to what you're doing today.

Yeah. The journey of going public, all the market gyrations. 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 their position built up in a particular category. But the public markets seem like even harder to read on because you have retail investors and the stocks up and down and things can reprice on a minute-to-minute basis. What was it like psychologically transitioning from private company to public company?

Yeah. What kind of feedback did you get in the early days around naming the company after yourself? I've been surprised at more... there's so many generic names in the startup world now that's like the blank company of San Francisco or things like that. Or, you know, all the neolabs have like the same sounding 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 I only know one other, I can only think of one other company, Chris Altchek's company has Amazon Heavy Industries. But I'm sure people thought you were a little crazy back then.

So what actually needs to happen? 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. What's X, Y, and Z?

There's not just 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. There's a plethora of different physical technologies that are used to build these computer systems. 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 path that all these different hardware modalities are on. 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 requirements of bringing quantum into the AI data center at scale.

How important is simulation at this point? Are you at a place where you can run this, like basically run the code of the future in simulation to understand, like run it on a classical computer, not see the performance gains, but at least understand that when the quantum system is available, there will be a cost savings?

Yeah, we've been able to do that largely speaking, and you can do simulations of something, a computer system or a jet or anything, at varying levels of physical fidelity and detail.

The simulation we've been able to do so far indicates that we expect a level of, you know, several orders of magnitude potential speedup for key training tasks. Right? 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. It's several orders of magnitude, you know, when all the pieces come together. But that simulation you talked about is a really, really important and powerful part of designing a computer system. You can't simulate all the logic of a quantum computer because that would require a quantum computer itself, kind of by definition. But you can do load profiling, you can do traces, you can understand how that's going to be distributed across classical and quantum hardware and also simulate all the networking transactions in between. And so that's a kind of simulation-driven design approach we're taking.

Yeah. I guess what specifically in training benefits from quantum computing? Because the example that everyone goes to in terms of quantum computing, you know, novel algorithms that actually have potential to do something that a classical computer can't do, it's like Shor's algorithm, cryptography usually. But when people think about training AI, they usually just think a bunch of matrix multiplication. Is there some different path that you plan on taking, or do you think you can operate at sort of a hardware-agnostic layer, 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, but by and large it's the same training paradigm?

It's really interesting. I think the answer is both. So our starting point is we're looking at ways that you can insert quantum algorithms and quantum computing capability into the existing paradigm, the existing workflow for training and deploying very large models, frontier models at scale. And that means that you're looking for an insertion point from a quantum algorithm where the data in, the data out allow you to then take a step that would take maybe, you know, a day or two classically and compress that down to hours or minutes and do that throughout the workflow.

The challenge is that quantum computing provides the possibility for exponential speedup with the right algorithm, but it also has this issue with data in and data out. So it's classical data in, which can't be exponential in size, and classical data out. And so the less you do that translation between the quantum part and the classical part, it's going to end up working better. So asymptotically, 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. But where you're probably not going to see is fully quantum-based models that don't include a substantial amount of classical compute as well.

So this isn't going to replace all the AMD or Nvidia infrastructure in the data center. It's going to augment it. And our business model and our focus and our product strategy is 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. And that's what's going to allow you to then 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 — more efficient from a time perspective, from a cost perspective, from an energy perspective — but also these models are in a way just a representation of the computer hardware that they're based on.

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 to market.

How are you thinking about timelines with this new company? Do you think there's — I imagine with the business right now it's like an entirely more of like technical risk than execution risk. Is that the right way to think about it? Like there's a lot of hardcore research that needs to be done understanding, you know, the feasibility of the approach. And what kind of conversations are you having with, you know, potential partners, if at all right now, versus, you know, about kind of like the near-term application, or are conversations like 2030s and beyond kind of thing?

Yeah, we're talking to customers now. We've got several active conversations. I think partnerships and early engagement with customers is a big part of our strategy. The reason that's important is because the challenges of really bringing a new compute capability into the AI data center are substantial and you've got to be working with customers out of the gate to really understand those requirements, what moves the needle for them as an organization. And so that's what we're doing and that's what we're focused on. In terms of timing, it's a fantastic time to start a company like this. The underlying hardware has made such tremendous progress in the past 10, 15 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 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 more efficient way of translating watts into intelligence. That's what this enables and unlocks in the long term. And to me, this is in many ways a better idea than putting stuff in space.

Because ultimately, yeah, space gives you a lower, you know, cheaper access to energy and it gives you a better way to dissipate that heat, but you've 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. The power challenge, quantum can unlock much more than that.

Yeah, it's a good point. Why don't you think Elon has made a real run at quantum? I think the answer is that quantum is 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 vanilla — you know, it's not that it's easy, not that any of the purely classical stuff is easy.

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 Intels of the world. 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. I think for quantum, a lot of that is behind us, even though 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 companies that are quantum-specific bring the technology forward. And at that point, I think that all the big AI labs are going to need to lean in with quantum.

When do you think there will be a flip around sentiment around quantum? It feels right now like, at least in our corner of the internet, there's so much FUD around quantum and obviously based on financials, like, right.

Yeah, yeah, yeah. So that's what I want to know though. Like, is there a moment — 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. I mean, Sam was talking about like people becoming best friends with a chatbot. I think in like 2015 or something, or 2013, it was like losing money. It wasn't like making revenue yet.

Yeah, and that was even before that. Yeah, no, I know. Well, well before that. And so, but then eventually it flipped and it's really hard to, you know, there's a lot of people that are AI bears and they talk about like overinvestment, but they can't deny the value of the products, right? Like they're fundamentally pretty useful, right? And you could argue that they're, you know, well, some bears can, but yes. Yes. Some bears would still figure out a way to argue that they're not useful, but I imagine like with both of your companies, you're predicting that like, you know, within the next five years there's a flip. But what do you think is the first kind of like driver of that where maybe the average person in Silicon Valley actually starts to say like, 'Hey, I wasn't taking quantum seriously enough.'

There's a few things that need to happen. I think the FUD is real because the companies that are succeeding and doing well in this space, you can't tell by looking at their financials. You can't put on your kind of growth investor hat and say, 'Yeah, this is going to be a tremendous company,' and look at the metrics. It doesn't work like that. You've got to be able to analyze and look at these companies and value them based on their ability to buy down technical risk over time and the progress that they've made towards that. So it just creates a lot of uncertainty because it's a challenging task and it's subject to a lot of discussion and debate. But nonetheless, I think there is clearly tremendous momentum and progress in the space. Now, what's going to change it? I don't know. 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.

You care because it's a more efficient way of generating the answers you need, or training the model, or deploying the model for inference. And 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. And I think in a large part, this is what we're trying to achieve with Sigleree, right? The goal is 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. That's all going to be done with AI anyway. And so that is just a better way to train your model and, you know, you need this thing or else it's going to take you too long and your customers aren't going to be happy with the quality of the outputs they're getting. That to me is a big inflection point and I think that can happen in the next five to seven years. But there's this whole march that needs to happen to take the technology from one proof point to then 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 where we're selling, you know, hundreds or thousands of units a year. And 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 this is probably very obvious to somebody that is focused on quantum, but not to me just because I don't follow it closely. But like, why a new company? It feels like quantum, like as you've explained, it feels very obvious to apply it to data center buildout and you said it could be like a meaningful inflection point for the technology overall. Why was a new company necessary and why did you take this approach?

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. That is a fundamentally new approach and it ultimately is going to, in my opinion, be very obvious in retrospect. It's going to work better. But it is such a fresh idea. It's got to be baked into your strategy, the DNA of your company. And then all the different quantum hardware companies that were out there before Sigleree basically started with a thesis which was, 'We've got the best qubit and so we're going to scale this qubit type up and see how far we can get by scaling it up.' And that's why you have so much doctrine and like kind of organizational belief around a particular qubit choice. But in reality, you know, customers are buying a computer. They're not buying the physical device or your qubit technology. And so at Sigleree, what we're doing is working backwards from the market application, from the AI workload 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. It's just such a totally different approach to quantum hardware. It's got to be a new company and that's Sigleree Technologies. That's the approach that we're taking. I think that is ultimately what's going to unlock this new market application of AI. 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. And in fact, it's not even a consensus view right now.

And the reason for that is because quantum algorithms themselves are still in this very phase of discovery and development. And obviously AI is going to help with that eventually as well to an extent. But quantum, you know, when you interview a set of leaders from across the quantum hardware industry, 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 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 Sigleree is that kind of quantum AI-native research lab, right? Or a frontier AI lab that's quantum-native. And then we're doing that alongside developing our own quantum hardware.

Before we jump, you mentioned kind of the history behind the name, but what is the significance of Sigleree in the novel that you mentioned?

Oh, well, you guys got to read the novel for one. It's absolutely incredible. But the other thing is Sigleree is basically a discipline in the book that is learned at university 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. It's also a discipline where it's got a quantitative angle to it and if you do it wrong you can blow things up. So it's got 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 these inscriptions. It's a really amazing concept.

Amazing. Thank you so much for taking the time.

Thanks, guys. Have a great rest of your day.

Let me tell you about Console. Console builds AI agents that automate 70% of IT, HR, and finance support, giving employees instant resolution for access requests and password resets. Our next guest is already here.

Very cool name. I wish, you know what I'm thinking, John. General — I wish that after computing, I wish that Chad launched Chad Computing. Oh, yeah. It was right there. It was right there. Anyway, but we have the co-founder and CEO of General Intuition with us. Welcome to the show, Pim. How are you doing?

What's happening?

Hey guys, thanks so much for coming back on the show.

We've been talking about names for labs. What about, uh, consider General Intuition, strong name. But since you launched the company, a lot of other, a lot of other new labs have kind of come out with like similar names, like General — there's probably like a General Superintelligence or like a General ASI. How about you rebrand to Unfettered Intelligence? That might be it. Or how about we just fund them all? Yeah, that too.

Yeah, what is the plan? When do you see yourself as a new lab and do you see — is it as much of a knockout dragout fight as it appears from the outside or is your model more of a thousand flowers bloom?

The plan is to just keep renaming. No, look, you have to have a claim to why you can win, I think, otherwise none of this makes any sense. It's an incredibly competitive fight. There's lots of great contenders. The only reason why we have a shot is because we have a dataset 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. And so you need to have a very focused, dedicated path. Some of that can be, for instance, having the best researchers or having the new ideas, but I think it also has to be supplemented with a product focus of some customer problem that is going to get solved because these types of model classes exist. Network effects, just like we saw in the consumer eras of the Facebooks and the Twitters and the Reddits. These things are true. They apply to LLMs as well. The fight for that space is going to be incredibly tough and so you have to introduce something new. I don't believe in the just M30LM space, which is why we're focused on actions and space and time.

Okay, actions in space and time. Let's talk about the dataset. Catch everyone up to speed on — I mean, you know, you broke it down for us the last time you were on, but it feels like it's been almost a year at this point. So what have you been working on? And talk about the dataset, how you're building the dataset, all that stuff.

Yeah. Look at it this way. As humans, the decision to talk or type is just a very, very small subset of the actions that we can actually take, right? We can choose to move our body. 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 with these environments, which is 2D environments, 3D environments, interfaces, long horizon tasks, short horizon tasks. And so in order to do that, it has to be a sufficiently general intelligence in order to learn how to correctly predict actions. And therefore the type of model you get out is not going to taste like an LM. It's going to be like comparing coffee to water. This model is going to be incredibly good at navigating unforeseen environments. It's going to be incredibly good at zeroing any task where it can already be controlled using a game controller because we have roughly a trillion action tokens in that space, for example, right? For context, frontier LMs are trained on maybe between five and 10 trillion text tokens, right? 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 a game controller, which is most robots, right? That's really what we're doing. We're using that simplification to turn it into mostly an environment transfer problem. 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? It's not going to be text as you're used to from LLMs, but it might just be enough to communicate why you're doing a specific thing. So that's how to view the models.

So yeah, walk through the partnership with Meta. Are you getting game controller feedback as well when those — Yeah, explain the relationship to Meta for those.

So alongside the frames in the video, we're also getting the exact action inputs. To be clear, not the letters or numbers, right? We had thousands of humans convert those into the actions you're taking. So walk forward, walk left, open door, close door. 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 scaling regime as if you were trying to do this on inferred data. 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. It's impossible for that to be visible in the pixels, right? But it's in the action sequence. And so there's just no lab that can take this approach. There's lots of benchmarks that might show that you can do this on inferred data. 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. And so you cannot do this on inferred data despite many people claiming you can.

Tell us about the latest round. I want to hit the gong. What happened? How much did you raise?

What happened? We raised $320 million.

Congratulations and thank you so much for taking the time to come chat with us.

One more final question. Talk about progress from your customers, companies that you're talking to in robotics. Where is maybe an area that you're particularly excited about that you don't see being talked about yet?

Yeah, the most obvious thing this replaces is all the code that people are currently writing for behavior in physics engines. All that just becomes a prompt. And so 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. But basically, you can play the world as if it was a video game. If that can be said about your use case, the models will generally do incredibly well. 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? 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. In many ways, you could argue that if this is correct at scale, the supply chain will converge on gaming inputs instead of humanoid robots. And I think that is one of the big things that I foresee happening in the next two years, because intelligence is a bottleneck.

Well, thank you so much for taking the time to come chat with us. Congratulations and we'll talk to you soon. Have a good one. Let me tell you about Cisco — critical infrastructure for the AI era. Unlock seamless real-time experiences and new value with Cisco.

Fascinating. It's also funny seeing all those simulators on Steam, like, and the fact that, like, will the training data generalize? Are they just going to learn how to play Fortnite? And it's like, well, there is a farming simulator and there's a, you know, data center simulator, happy bar central banking simulator, potential banking simulator. It's going to learn everything. Well, we have our next guest in the waiting room. Yadin Sofer from Tracer, the co-founder and CEO. Welcome to the show. How are you doing?

Hey guys, nice to meet you. I'm great. How are you?

Thank you so much.

What's happening?

Introduce yourself. Tell us what you're building. Tell us about the emergence from stealth that's happening today.

Yeah. Well, Yadin Sofer, last week we announced the launch of Tracer, which is I would say the first of its kind subterranean defense tech company. And subterranean is a word we actually coined, but I've been happy to see people reference it on X already. It refers to everything in the subterranean defense domain. So that's everything in the intersection between military applications for things that happen beneath our feet.

What is the history of subterranean startups? So you have the Boring Company. Palmer has talked about the domain. I don't think he coined it, so you get all the credit. But what have been some historical sort of just like general efforts in the category maybe outside of the Boring Company?

Yeah, I think on the civilian front actually subterranean is a developed industry. 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 deserves some love and it did get. 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, you know, piercing the way, pun intended, in everything underground. 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. I mean we really have not seen any companies in this space.

What are the primary challenges of, you know, underground drones, the underground domain overall? Is it connectivity? What are they?

Oh yeah. Yeah. Well, you know, I think it's interesting because 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 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 what you're dealing with and when you're working on the underground, when you're essentially boring your own, you don't know what to expect. You don't know the geology, composition. You can have a high sense of how it's going to look, but 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 number one challenge, just the uncertainty of this domain.

Palmer talks about this. He says diameter is expensive, length is free, something along those lines. Can you explain that concept and how it informs vehicle design for the subterranean domain?

Yeah. No, it's such a great point and I think a lot of people looking at this space are thinking the same thing, right? We're thinking a train where it bores its own path and it takes behind it essentially infinite payload, right? You can have miles and miles of payload of sensors of effects and, you know, the dream is someday people. Now, when you think about it, when you're increasing the diameter, you need to remove so much more dirt, right? You're dealing with a lot more. And when you work at a small diameter and essentially infinite length, you could even condense the dirt to the sides. You don't necessarily need to remove it. And that becomes extremely valuable. So most of the questions are around that. And I don't know if you guys have seen a boring site, but a boring site is this massive thing, right? You need the bentonite to mix with the dirt to take back outside. It's like a whole thing. But when you're working on small diameter, you don't necessarily even need to remove the dirt. You can just condense into the sides. And I think that's a big part of, you know, going sort of slim and long.

$25 million seed round. What's the goal? The government isn't actively buying this technology. There isn't a program of record that you can sneak into, I imagine. So what does the next two years look like?

Yeah, we always say this that, you know, if you try to find the line items, they're like line items buried in line items, right? Obviously, we have penetration munitions, but those are air-dropped bombs, and we're not looking to compete with Boeing. But I would say the interesting points and the slivers we see of interest from the government right now are in — there was a recent RFI by DARPA where they're looking for new methods to induce collapse in underground infrastructure using different shockwave methods. So essentially, we're looking at this as non-kinetic penetration munitions, right? Our ability to insert a payload underground. This doesn't have to be dropped from air. It can be done by special forces on the ground and essentially detonate a payload in a sequence that induces collapse of facilities like in Iran. So, you know, I think the military is starting to understand that the existing solutions do not deliver what we need them to. So they're starting to think differently. But back to the round, right? With $25 million here, everyone goes to me and is like, 'All right, you're building this massive R&D team. You're going to have a ton of capex.' And I'm like, 'No, there is a lot of work to be done when forming, call it this category where we need government, 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.' 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 subterranean or the US subterranean strategy for, you know, winning wars underground.

How far underground are you right now? It does look like you're underground, right? It looks deep. I was thinking about this, too. It's a good spot. At least 20 feet. At least 20 feet.

Anyway, thank you so much for taking the time to come chat with us.

Great. Great to meet you.

Have a great rest of your day. We'll talk to you soon.

Cheers. Have a good one. Let me tell you about Figma. Agents, meet the canvas. Your AI agents can now create and modify your Figma files with design system context. And Jack Morris from Engram is in the waiting room. He's co-founder and head of research. Jack, how are you doing? Welcome to the show.

Hi. Yeah. Nice to meet you. It's great to be on the show. I was actually just watching it in another tab, so this is kind of — Here you are. Fix that.

Great to meet you. Tell us a little bit about yourself, tell us about the company. You're emerging from stealth with a whole lot of venture capital. What's the strategy and what's the product?

Yeah, sure. My name is Jack. I'm a co-founder and I guess technically the head of research at Engram. We came out of stealth last week after eight months or so of working on our product and ideating with our design partners. Yeah, we raised money from a bunch of VCs. The product is — is mogged, mogged.

Let's hit the gong. Let's hit the gong for that.

Grateful for the opportunity, but I was hoping you would hit the gong.

Yeah, we just did a b, you know, baby. Yeah, big one.

Congratulations.

Yeah. And thanks to all of our partners and thank you so much for funding us. Our product is a new type of AI. 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. I think there's another way to think about it which is that the model doesn't need to get smarter every month. It needs to know you better. And so we're working on like a whole different stack which is a way to train models that train themselves to like know your world better and like adjust to the things that you say. So it's like new ways of training, new ways of running the models. I think like to give a concrete example, I assume you all are very tech-forward. You probably have agents doing things like preparing you for the show and like giving you reports every morning. And if you actually look at what 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. Like literally probably every night, they're probably like reading from scratch, what is TBPN and who are you two and, you know, who's been on the show recently.

No, we're in the pre-training now. Come on, give us some credit.

Oh yeah, you are in the pre-training. Move so fast.

Your point 100% stands, but yes. Yeah, I think you're lucky because you're in the pre-training, but I think most people are not in the pre-training. So many documents that aren't and you have to feed those in every time. Is this the solution to continual learning? Is that the correct buzzword for this strategy or is this a different fork in the road, a different path?

I think it's the correct buzzword. I think a lot of people use the phrase to meet a bunch of — they cracked it in eight months. They cracked it in eight months. It only took them eight months.

Let's go. Oh, we decided to name ourselves something different. But I think of continual learning as 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.

What's the sweet spot customer? Enterprise AI can mean Fortune 500 companies, that can mean a very data-intensive company. There's also whole categories of enterprises that have a whole host of AI wrappers and application layer companies duking it out. I'm thinking of legal, medical. Where do you see the product having the earliest signs of product-market fit?

Yeah, I'm glad you said earliest because I think like there's two halves 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 coworker, and it'll be able to like generalize and do things better than the current models. But I think the current customers 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 we've been working with are Microsoft, Notion, and Harvey.

And I think they all... you guys with the sound effects, I'm like so flattered. I wasn't sure if there would be any.

Um, they're nice because they have these like massive workspaces of context and like they're, you know, early adopters of AI. And I think these are the places where we can like reduce costs the fastest, the soonest, because the workflows really are just that repetitive.

That's great. Well, thank you so much for coming on and breaking it down. Appreciate you taking the time and have a great rest of your day.

I know you will be back on. I'm going to guess two times this year. That's my guess. Two times for sure.

We'd love to have you back and chop it up more. Have a great rest of your day.

Yeah, it's great meeting you guys. Thanks for having me. Great to meet you, Jack.

We'll talk to you soon. Cheers.

Let me tell you about the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange. Our next guest is Neil from Sale Research. He's the co-founder. Let's bring in Neil Movva. Movva, how do I say your last name? I don't want to get it wrong.

Thank you so much for taking the time. Uh, congratulations on the round. But first, please introduce yourself and the company.

All open source. Do you work with other labs? Are you, uh, how deep do you go into the relative organizations?

Tell us about GLM 5.2. Uh, what makes it, uh, what makes it like different in a binary sense? Is it, is it a particular benchmark? Is it a vibe? Is it an application? Have we unlocked a new capability in open source AI?

For coding specifically. Uh, what about... before you would put on clown makeup and then you say, 'Ah, yeah, give it, give it a spin.' Uh, what about for other agentic workloads? I mean, we're looking at OpenRouter, a lot of the top models, uh, DeepSeek V4 Light. It seems like it's a lot of heavy token generation, lots of, lots of value being created, but smaller tasks. Um, what, what is that like from your business perspective? Are you still focused on optimizing those types of workloads?

Yeah. 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 agent? 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 spun up a bunch of tokens. 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. Uh, and I imagine that that will be a huge driver of token consumption. Um, and I'm wondering how you see those two buckets balancing out.

Yeah. 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. Like, what, what, what type of pitch do you have in terms of savings?

What were you and the team doing before this and, and how long have you been at it?

Is this a possible story where you're like, I was working on GPUs and you were just playing Counter-Strike or something?