Fruitfly Hard Takeoff, Washington on AI Risk, 𝕏 Timeline Reactions | Thijs Simonian, Alex Heath & Guy Oseary, Mitesh Agrawal

You're watching TVPN.

Today is Friday, September 11th, 2026. Wall-to-wall coverage in the Wall Street Journal remembering 25 years ago, the COVID of the Wall Street Journal, every single article except for one was about, of course, the World Trade Center. The headline that day was Terrorists Destroy World Trade Center, Hit Pentagon in Raid. with hijacked jets. Bin Laden is on here. The only story that, you know, attacks raise fear of a recession. Very interesting time capsule. Highly recommend picking up a copy of the Journal today and taking a trip down, you know, in memory of the tragedy. The one piece of news that broke through this day that was not related to the terrorist attack was Xerox reached an equipment financing agreement with GE Capital that will let Xerox erase about $5 billion of debt. Very, very odd.

Strange.

Every other story— I mean, the market was closed. Actually, the internet got a very interesting shout-out here. It says telecom systems were strained as terrorist attacks in New York and Washington knocked out telephone wireless services across the Northeast. The internet proved the most reliable way to communicate following the attacks as the phone system sagged from severed lines and an extraordinary volume of calls. Corporate executives used email to find employees across town or across the country. So interesting to see, but there's so much to go into and lots of interesting retrospectives across all the different media organizations. I don't know that I have a particular anything to dive into there, but it's interesting for you to go and dig into. Anyway, moving on. Let me tell you about ramp.com. Time is money. Save both. Easy-to-use corporate cards, bill pay, and accounting, and a whole lot more all in one place. Now with musicals. Anyway, what are the AI doomers actually proposing? That's the question I was trying to answer this morning for a while.

20-year sentences.

That's one. So yeah, there will be this weird translation layer between the thought leaders and the people that are writing policy papers effectively, and then what actually gets implemented, what gets votes basically, can be wildly different because to get the populace to actually support something, you need to wrap it in a different structure potentially. AI 2027 predicted that this month, late in 2026, Congress would wake up, and that is in the journal. Congress is suddenly waking up to the AI doomsday threat. And so, this is happening. All over the place. Was it Matt Damon who was caught on TMZ being interrogated about his thoughts on AI risk? There are now protests. I believe AI 2027 predicts a 10,000-person anti-AI protest by the end of the year. I was trying to figure out how big this protest in Chapel Hill, North Carolina earlier this month was. I think it came in Sub-10K, but certainly tracking to it. It was about 150 people at this protest in Chapel Hill, but just 2 ooms away from the prediction from AI 2027. And AI 2027 and the sequel, AI 2040, frustratingly vague about impacts outside of— the AI industry. So there's a lot of really, really amazing predictions about agentic capabilities and the amount of compute that will be marshaled and like even lab revenue. But there's not as much predictive stats and calls around like, what'll it do to GDP? What'll it do to employment? What will, you know, how often will people actually be using this? What will it actually be good at? The diffusion question is still sort of left unanswered. But they're clearly taking it very seriously. There's a huge press cycle around this. And the Journal says that Congress is suddenly waking up. So I wanted to dig into like what the actual proposal is, 'cause the joke for a while has been just like everyone, when they're pressed on this, they just say, we gotta talk about it. We gotta talk about this. AI 2040, Daniel.

One thing that stands out, it's interesting to me That there's been so much more seemingly grassroots mobilization around the anti-flock movement, deflock, when you compare that. And you don't see—

Totally.

You don't see videos of people with 100,000 likes saying, like, here's how to cut power to your local data center.

Not for x-risk reasons, but you will see a post that's just like, I don't like image generation because I'm an artist. And that will get a lot. So there is a lot of anti—

Yeah, but what I'm saying is that the sentiment, the anti-flock sentiment converted into actual physical actions by a bunch of just otherwise normal people. I'm saying we haven't seen that yet with data centers. Maybe we don't, but it's notable.

Yeah. Much harder. Data centers are in remote locations, highly fortified. There's fences.

Totally.

Just walk in.

Flock cameras are in your on your street maybe.

Exactly, yeah. So yeah, there's degrees there. But AI in the— what's weird is that, yeah, I mean, you're calling out grassroots taking down the data centers. The AI 2040 proposal is effectively the opposite. It's like harden the data centers even more. And the actual proposal is super concrete in AI 2040. And it's very interesting just to hear about how they want to slow things down. The thing that I think a lot of people online who are like, yay, an anti-AI sentiment that's going viral are going to be depressed about is that this is not stop AI at all. AI 2040 is like, keep the inference flowing. The current models are great. Also, let's keep doing capabilities research, but we just want to reach superintelligence by 2040 instead of 2028 where we're not necessarily prepared. So, they want it to be highly controlled by governments, nation states, highly secured. And there's a whole bunch of very, very tactical recommendations that they make around that. So, the first mechanism is an AI pause. They want to pause training. They don't want to do any more new frontier training runs or R&D experiments. And to enforce this, they're calling for— to apply inference-only verification to essentially all major AI data centers. So anyone who has more than 10,000 H100 equivalents, roughly $100 million of equipment, and that is like pretty easy to figure out. You're just like, big building over there, let's send the inspector inside. Oh, says Nvidia on all these chips, count 'em up, there's over 10,000 of them. You gotta apply for this permit. You gotta tell us what you're doing, right? Very easy to enforce, at least in the United States. And with an international body, you could kind of do the same thing internationally. So the whole goal, you can only inference the current models and you gotta verify your workloads with an independent auditor, probably the government. Maybe there's some sort of non-governmental organization that's doing this. You know, there's a whole bunch of different solutions that you can pull from across nuclear nonproliferation work that's happened in the past. Major countries, they want them to declare AI compute inventories, tell everyone not just your local population but also the international community how many warheads you got, how many H100 equivalents do you have, where are they. Everyone shares this. That's going to be a really tough sell because international agreements are really, really tough sells. It's much easier to have a groundswell of support for something that happens in America. America changes. We have a system that we don't really have the international rules to to quickly implement that in a way that doesn't allow for a lot of defection. But they want to know who has compute. So, major data center owners and semiconductor supply chain companies would be required to turn over sales records. Who'd you sell those chips to? Where'd they go after that? Foreign inspectors would do routine chip counts physically at site, on-site at facilities. Large transfers of chips would only be allowed to go to registered Audible counterparties. There's some interesting networking-specific components to this proposal too. They want to physically remove high-bandwidth east-west networking inside of data centers. So you can't do large distributed training runs, but you can still do inference. So again, anyone, all the anti-AI people who are like, oh yeah, I don't want LLMs around anymore, like these are not your guys. They're not fighting for that. They are fighting for stopping the next training run, which is probably aligned. Those are overlapping circles, but it is not moving backwards in time. It is merely slowing down at this current moment. They also wanna install passive optical network taps on anything leaving the data centre to independently verify traffic. For any new AI R&D data centres, they want entirely new facilities built from scratch. with nation-state-level physical security and verification. So they're saying, like, okay, we're, like, the data centers that are built, they can inference the current models, your Astras, your Fables, like, your Groks, like, those can run because we can deal with those, we can harness those, we have control over them, we're going to continue to align them, and that's a solvable problem. But for the next run and the one after that and the one after that, as it gets crazier, We want it in a new building built inside a Faraday cage so you can't communicate it from the outside. We want a bunch of physical controls. It's like going to a nuclear facility, highly verified who gets in the building and when, air-gapped communications. This is one interesting proposal that they have that really shows how deep they thought this through. They want the R&D data center to be connected externally, if you want to communicate with it and you want to tell it what to do, okay, train the next running or do whatever, they will have a bandwidth-capped connection at 1 meg per second. So you can send little instructions, but if you say, send me the weights because I'm taking them somewhere else, it would take you like 5 years to exfiltrate it. So interesting hardware solution to this. How do you actually go and implement that? There's going to be a whole bunch of other things, but— Interesting that they're thinking about the width of the pipe. So, it'd be very obvious if you're stealing the model weights because it's like, wait, this 1-meg pipe has been at full tilt for months. What's going on here? Someone's taken the stuff out of the data center. They want— when frontier model weights move from an R&D facility to an inference facility, they want it to be placed on physical storage devices encrypted independently by both the US and China. So, both countries have to sign off. and physically escorted by representatives of both countries to the destination.

Jesus.

It's a tall order. That one's a tall order for sure.

Yeah.

And they actually want frontier models to be made deliberately larger than compute optimal. So, they want the weights to be 100 terabytes instead of honing them down to something that's just 1 terabyte that could actually be moved around a little bit easier. There's a bunch of public disclosure proposals in there, restrictions on various trade-offs. Labs would have to share the model specification, the fraction of compute devoted to internal AI use, so you don't get a lab that's just internally using way better models than what's available externally. They want qualitative descriptions on how powerful models are being used internally, restrictions on how big the gap can be between the best internally deployed model and customer-facing products. This has been a common discussion point with like—

Yeah.

the rollout of Mythos and Fable and Astra and Astra Next and all these different models where people said like, oh, it's really unfair that like this lab gets a better thing than I do. We should be on even footing if we're both going to be competing in web design or we're both going to be competing in legal. Like, why can't I buy this product from you? And so there's some overlap there with like the broader business community, which I thought was interesting. And—

Yeah, I mean, first of all, this— I mean, a lot of this tracks with The kind of regulation that, you know, we had pushed for, you know, beginning about 2 years ago around podcasts.

Yes.

You know, wanting podcast studios to be air-gapped.

Yes.

Wanting, you know, Faraday cages around podcast studios.

A locked briefcase with an SM7B in it. And in order to unlock it, Patrick O'Shaughnessy and David Senra both need to give you codes to independently verify that this podcast is worthy of being recorded.

Yeah.

I like that one.

Yeah.

Yeah. It just makes sense. So high-level valve they see as being like the most effective in controlling the speed of capability improvements is compute caps. So there is a world, and we're going to the Bernie Sanders thing 'cause it's already getting like sort of twisted, but the big hammer is just chip controls and Data centre build-out slowdown.

Yeah.

That's the easiest thing. And I think that's like the biggest valve that they're going to be— that we're gonna see twisted around to actually slow down capabilities. And so the goal is to allow models to get better mainly by adding hardware rather than inventing better algorithms, which can leak to secret projects. So the goal is like, okay, well, we know that this model is capable of this. So we want this much compute over here. Okay, you've done well. We're allocating more compute as opposed to this one weird trick that AI doomers hate.

Yeah.

So the goal here is not to go back in time.

Yeah, it's so interesting because when you look at, I mean, anytime you have really, really hardcore government regulation and international coordination around issues like this, you're gonna have a bunch of unintended unintended consequences.

Yeah.

And one thing that feels obvious around this, if these policies were to be rolled out, is that you would effectively create an incentive for millions of individuals or groups globally to be in secret, like trying to find entirely new breakthroughs that are—

Yeah.

And again, this incentive already exists, but it kind of pushes A lot of the idea that humans are just going to be like, oh, I'm not— I'm no longer going to try to create the God model because—

Yeah.

Like, there's this, you know, big global organization that's sort of policing it.

Yeah.

I just—

I mean, that's the same thing we see with nuclear nonproliferation. There's always a discussion about what countries are getting the bomb and when and how far along are they and wars break out over this and— Yeah, like the game's not over just because you create a framework, but there is at least a— I mean, we've avoided World War III, so you could say that a lot of, like, the vast majority of nuclear nonproliferation work has been successful, even though there's been a ton of examples of people trying to divert around it. In fact, it's like been the backbone of geopolitics for like 60 years, has been like who Who gets the bomb and what chips are on the table.

But ultimately, GPUs and computers are much more wide, infinitely more widespread than nuclear materials.

Yeah, but you still got to marshal them all together. Yes, there's some weird scenario where there's a Python script that's AGI that can run on your laptop, but I think most people are convinced, at least in this crowd, that—

Scale.

That scale is a prerequisite. I mean, we were joking about, like, it would be so— like, SSI, Ilya Sutskever's new NeoLab, recently got a big cluster from NVIDIA. We were like, the most bullish thing you could do if you're this secretive NeoLab would be like, we're actually selling our compute because we've discovered a more compute-optimal way to reach AGI and we don't need a lot of compute. But of course, even Ilya is like, it's time to scale up. I need more compute because It seems like even if he's taking a completely orthogonal approach to innovation and research, he still needs a lot of compute. And so, it does feel like everyone is sort of with the consensus that it's going to be a big building with a lot of energy, big heat signature, definitely visible from space and pretty simple to track, at least in the short term until people start building crazy underground facilities, and then you're back to nuclear nonproliferation. But their goal is at least to try.

Yeah. And then the other side of this is— Not a panacea. Does AI development actually become something closer to the Manhattan Project where—

That's what a lot of people think.

100 researchers are working with the government in secret because you can't just assume that Other countries are going to slow down or do any of these things.

Yeah. But in general, I think the proposal is don't go back in time. It's definitely not stop everything in its tracks. It's a slowdown with the goal of scaling gradually. They do actually want to reach superintelligence. They just want to do it by 2040, hence the name of the project. So the goal is, gradually scale into top human expert capability around 2035. Now, a lot of people are saying, oh, we might get this by 2029, 2028, 2027, and they see that as too fast, so they want to push that out to 2035, then wait 5 years with AGI, and then unlock superintelligence in 2040. This is their initial proposal. Of course, there's a lot that could change over the next decade. And And I think, like, to zoom out overall, if you're worried about x-risk, the AI 2040 plan does feel like a concrete path towards slowing down. The conversation definitely gets dragged down into PDoM estimates and trying to narrow down exactly how a human extinction scenario plays out. And that can be— I feel like that's almost a sideshow, because in a democratic society, in— amongst humanity, like, it doesn't really matter the mechanics of getting to 10% PDoM or any of those. It's just like, if everyone feels that way, something will happen. This is a concrete plan of what that might look like. And that's valuable to understand in this case.

Yeah.

So, for the safety sceptics, it's easy to see how this level of control over what you can do with computers is authoritarian or anti-libertarian, Even if we're talking about $100 million computers, there's a lot of people that say, like, I should be able to do math on my computer. I can do whatever I want. Let me do cool things. I'm excited about this. That limits your freedom. It might create regulatory capture for a few major players. It might crash the stock market or delay economic gains that come in the good ending where alignment is solved and x-risk plummets. You can imagine a situation where in a few years, if X-risk fades into the background. You're like, yes, there's still a risk, but it's the same risk that we face every day with, like, an asteroid hitting Earth. Like, no one— it doesn't really change anyone's behavior. That would be sort of the good ending, in my opinion. So, it's a balancing act. And for most of these slowdown proposals, I personally have a hard time blackpilling about them in the sense of, like, if all of this gets implemented, how frustrated will I be? Like, the models are good. I would like better models. I want safe models. But at the same time, like, there is this massive capability overhang. The current models can do a lot of interesting work. We're finding new uses even for, like, non-leading-edge models. There's a lot that can be done. So I don't believe the doom doomers who are dooming about what the doomers are planning. I find that unconvincing right now. But People are starting to lay it out more. Brad Gerstner, Jensen Huang, David Sacks are talking about the other side of this equation. But I haven't— just like it's hard for some people to concretize the Terminator scenario, I also have a hard time concretizing the— we didn't race and we're unhappy about that. I guess you could say the housing scenario. There's been other times when we brought in too much regulation. slowed things down too much and been like, ah, like, this was really not the right move. But at the same time, I think we have a lot that we can do with the current technology, that there's still cause for optimism even if something like this gets, you know, universally voted on. It wouldn't— I don't think it would be the worst thing for companies and consumers and businesses and all sorts of different folks.

Yeah. But there's also a big question around what Do the 2040 people have a point of view on robotics and physical AGI? Because it seems like if you— even if you pause efforts towards RSI, well, if we add billions of robots into the world that are just running on today's model, that also presents today's—

I don't think they're worried about that.

Yeah, but to me—

I'm concerned about a billion robots with GPT-6-level intelligence and years of alignment work that is currently happening. That's fine. It's the next, next, next thing, the superintelligence, the thing that might have its own goals. I think, I mean, we're talking to someone from OpenAI's robotics team, Hooking up Astra to a robot, a paintbrush, and a camera. We talked about it earlier, painting. I don't think we're at a point where that poses a risk. It's the next model. It's the model with its own volition, basically. Which a lot of people still aren't seeing. They're just like, yeah, like the models keep getting better, but they seem to follow your instructions, sometimes too much, and then you need to worry about the paperclip scenario. But it's not that they wanna do their own thing necessarily.

I don't know.

But people are going back and forth on this. Clem over at Hugging Face says, sorry, but asking Jacob about AI extinction risk is like asking your AC guy about climate change. Not saying it's necessarily uninteresting or wrong per se, but let's keep things in perspective and hear from the full range of expertise across the ecosystem. And Nathan Lambert says, Banger. Uh, what was it? What was our take? AC guy might be right about my—

Well, my AC guy would be like, I don't really know about that, but I just want to make sure when you're hot that we can run this AC cool.

Yeah, I guess it's— yeah, the AC guy seems—

I don't know about all that. I don't know about all that mumbo jumbo, but you know, when it's a hot summer day, I don't want you to be worried about the heat, brother.

I like that. Yeah, this is a— this is kind It's kind of an unnecessary shot at AC guys. AC guys are important. I don't know. It's funny. Anyway, 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. What did Bernie Sanders have to say? Is this real? Entities shall be subject to the corporate death penalty, and persons shall be subject to more than 20 years not more than 20 years in prison if they don't pause AI development. That seems pretty easy to comply with. I don't know. I guess, how do you define AI development? Is prompt engineering AI development? And then you get caught 'cause your model sort of did a little prompt engineering in the final stage and then you're, guilty of this, like that. Yeah, that could be a negative knock-on effect, I guess. It does seem aggressive. But his overall proposal is banning artificial superintelligence, so no person or entity may develop or deploy superintelligent AI systems. He defines artificial superintelligence as an artificial intelligence system that exhibits or can easily be modified to exhibit capabilities that match or exceed human cognitive performance and capabilities across a broad range of domains or tasks. Because it sounds like the mission statement of the labs. It sounds like the explicit goal of like 17 different companies right now. AI system or AI systems that have sufficient capabilities to plan and execute the disempowerment of humanity. Okay, that's a good one. I like that. I don't like overthrowing or undermining the US government. So strongly in favor of banning that. Pausing advanced AI development until a new federal AI regulatory body is up and running. And then the new cabinet-level federal agency will monitor frontier AI systems at all stages of the lifecycle, supervise the removal of dangerous capabilities, and supervise the destruction of artificial superintelligence. We're coming for it, which is similar to the—

Corporate death penalty is a line that you don't hear a lot, right? Usually, these companies just, you know, bankrupt and wind down. But corporate death penalty goes pretty hard.

Pretty metal.

It's kind of metal.

Yeah, you kind of got me with that one. Yeah. Rough, rough, rough situation. We'll see where it goes. Jamie Cox over at FluidStack, the co-founder of FluidStack, a compute provider, shared his convictions, sort of pushing back on a lot of this, saying that he thinks America should build more. They're pro-freedom, pro-democracy. They believe AI will bolster human flourishing. We support simple, clear, enforceable regulation frameworks that set simple requirements proportionate to capabilities and risk with clear responsibilities and no unnecessary barriers to competition. Yeah, the real— you're going to see a lot of pushback from people who are like, the regulatory stuff is going to be like these 10 companies and I'm going to be number 11 and I'm basically getting the corporate death penalty then because I didn't make the cut to be one of the regulated, one of the approved companies. I'm still early in my stage. So there's a lot of nervousness, I'm sure. But Will Moneitis said, we believe AI will make everyone rich, healthy, and free is novel and interesting comms from the frontier. He's endorsing this. And I agree. I like this. I like these convictions. I think it's generally like a positive direction to move in, not a direct response to the proposals that are going out, but we're going to get a whole lot more of them. Where do you want to go to next?

Over on TikTok. They're sharing a photo of the whistleblower and saying, in every worldwide disaster movie, there's a dude that looks just like this that nobody listened to. And he really does look like an actor. He does look like an actor here.

Yeah, he looks good. But people are all confused.

And please stop calling him Scary Potter. I've been seeing people over on X calling him Scary Potter.

That's the goal. The goal is to wake up China, wake up Congress, wake up everyone.

DoorDash has entered the conversation. They say, 2 years at DoorDash, I do not say this lightly. We are extremely close to the burrito arriving before you decide you want it. We are not asking for a ban. We are asking for a pause. I don't know why they would ask for a pause.

Yeah.

That seems like a very very aligned to humanity and to their business, which I think is fantastic.

The doom is very much contained to the frontier lab work. Everyone in the application layer who's applying the models, diffusing them, they're like, I can't get this thing to work right. I got to get forward-deployed engineers to teach people how to use this thing. Everyone deeper in the stack is like—

Jim Rainbow over Jim O'Reilly.

Sure.

Or Jim Reilly.

Yeah. Jim Riley.

Jim Riley.

O'Reilly Auto Parts.

Jim Riley over at Charleston AI says it's a whole lot of mumbo jumbo.

That was your words.

He just— he's just happy to get, you know, 8 hours back.

Yeah, yeah, yeah, yeah. And then, yeah, everyone deeper in the supply chain like Jensen and all the different semiconductor manufacturers are, are not particularly on this side. And then you also have Wall Street, who's just like, what's the enterprise acceleration? So lots of different groups around the table that need to be brought on board to this movement. The discourse truly is fascinating. 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.

That's a new—

Too much Diet Coke.

New, new approach.

New. Okay. So what is Tyler Cowen calling for?

He's banging the table saying, bet on this.

Tyler Cosgrove, are you named after Tyler Cowen? Is that your namesake? Tyler Cowen says if you have very pessimistic fears or predictions about AI, name the market prices that will support or confirm them. This is what taking this seriously means. And I love Tyler Cowen. I'm not sure this matters because if you— you're not going to be around to collect it, right?

Yeah. This is what Deep Dish Enjoyer says. Tyler, why would short-term existential risk affect market prices in any meaningful way? Spell out the exact mechanism. Contracts that pay out if everyone dies aren't worth anything to me.

Yeah.

I don't know.

Yeah. I think he's just calling for like, you know, you want people to, like, make a falsifiable claim.

Like, is this—

am I able to tell if your claim is, like, true or false? And so, it's, like, very hard with these scenarios where, like—

Isn't it an unfalsifiable claim, though, just by definition? And, like, you just have to, like, accept that and move on?

Yeah, but then it's, like, so hard to have any, like, real discussion.

Was nuclear any different? Like, the threat of nuclear apocalypse, the threat of World War This was a very motivating factor for decades, most of the 20th century. People made real decisions based on it, based on where to do business and where the conflicts were going to be and the motivation for nuclear treaties and nonproliferation and how we treat—

You can make financial decisions, like, think about nuclear war, right? You have a bunker. That's like A decision you make.

Bunky?

Is anyone making, like, AI bunkers? No, because they think it's going to be so totalizing that the bunker actually doesn't do anything.

Exactly, yeah. So, if you think it's so totalizing, then you don't make the bunker. And so, saying, hey, you don't have a bunker, is not proof that the person doesn't believe what they're saying.

Yeah.

I don't know the answer to this question either, but—

Yeah, it's odd.

It seems like— I don't know, maybe there's some question you can ask that is falsifiable. I don't know.

I don't know. I think you just got to believe these— like this crew that like that's what they believe and, you know, like they believe it. There is the other side of this which is Paul Christiano who has been worried about risk. He recently just joined the board of OpenAI. And Tyler Cowen, you know, has this quote, if you're a doomer, why aren't you short the market? And Paul Christiano is 2x levered long, and he's short the— isn't he short the bond market or US Treasuries? So, he has 5% of his net worth in Tesla, 90% of his net worth in AI bets, and 100% of his net worth in normal investments, no Tesla options. That sounds like a scary place with lottery ticket biases and the crazy Tesla investors. And then Eliezer Yudikowsky says, am I correctly understanding you're 2x levered? And Paul Christiano says, yeah. And so he says he's personally short the US 30-year debt. I think that just means you have a mortgage. I'm pretty sure that's like if you have a mortgage, you are effectively short the US 30-year because you have sold that debt and you got the cash effectively. That's how that works. But Still, it makes sense because if you have enough money, you could pay off your mortgage, go long that debt, and short the market on a relative basis. But again, that's not— this doesn't seem like a doom-based bet. This seems like this bet also pays out in just like the good ending and like AI is real and delivers value. So, your AI bets perform well, the market performs well, and money slides from US debt to data centers and AI build-out debt or something like that. So he is putting his money where his mouth is, but it doesn't feel like a representation of like doom by any means. Anyway, let me tell you about Shopify. Shopify is the commerce platform that grows with your business, lets you sell in seconds online, in-store, on mobile, on social.

Marketplaces.

On marketplaces and now with AI agents. Speaking of AI agents, people have been creating AI agents of fruit flies. Have you seen this?

Yes. Okay.

So, where should we start? Because I want to set the table on what is actually going on. Tyler, do you understand this?

Break down what people are cooking.

We've seen this before. We've talked about this before. But what's actually going on with the fruit fly? Because now people are taking the fruit fly all over the place.

Yes.

So, my understanding is Google basically mapped out all of like the neurons in a fruit fly.

Fruit fly. So they probably took a deceased fruit fly and put it in like a mass spectrometer or something that can investigate the brain at a very, very— like a deep, high-powered microscope effectively.

Yeah, like the entire—

3D. They got the entire structure. And now people have been able to recreate that in software, in a simulation.

Yes, I think in theory you can replicate all of the fly's decisions or whatever, its movements.

Yes.

And this is notable because many people have said I have the mind of a fruit fly. They've said I have the intellect of a fruit fly.

Yeah. So now they're going to put it to the test to see which performs better, the simulated fruit fly or just Jordi Hays who we got right here.

Organic, farm-to-table. Jordi Hays.

So, now that this is out and the code is out and you can run this fruit fly in simulation however you want, people are having— they're doing all sorts of experimentation. So, Kevin said he trapped the fruit fly in his Rabbit R1. When he shakes it, he can see its brain's escape circuit light up. By the way, our consciousness is physically defined, which is why our physical things like Drugs, neurotransmitters, or events alter or initiate or end our consciousness in things like sleep, hallucination, waking, or death. Computer simulations are also physically defined representations.

Yeah.

I think what ends up being unsettling and weird about this is if it— just purely you being human and doing this is probably not good. for your own soul. And imagine you have a fruit fly in a box and you're shaking it and you're like, look, it wants to escape, right?

Yeah, totally.

I'm not a huge fan of small insects. I don't really want them around that much. But when they're in my house, I try to— if a spider's in my house, even if I know it might want to take a nice bite of me, I'm still going to, you know, try to transport it out of my house and put it back into the world. I think it's— I think it's not good for your soul to be, you know, a merchant of death.

Sure. In that situation. And yet, if you're playing a real-time strategy game and you highlight a bunch of soldiers in this simulation and you send them on a charge that will result in their virtual death, You might not feel—

But those soldiers opted into riding and dying with you.

Okay, okay.

Right? Okay, what about— The fruit fly's just sitting there, John.

So they made the fruit fly play Doom. What about if you play Doom, you are killing a demon who's simulated? Is that immoral? Like—

Well, is that demon trying to kill you?

A lot of it comes down to like the fact that it's Simulating the actual representation of the fly makes it a lot more concrete than just, oh yeah, it's a 3D model and a Python script that just says, if you see a character, shoot at them in the simulation.

Yeah.

But we're clearly starting to grapple with these odd moral questions of, if you're simulating something, then the next step is a couple of orders of magnitude. But you get there and you can simulate a human, And you could talk to that human and it would do everything the human does. Does that human have rights and agency as an ethical, moral agent? Or is it merely just a simulated, just a really good computer simulation? It's just existing on transistors, so you don't need to feel any moral weight about anything that you do to it. I agree with the, just the vibes-based analysis that, like, torturing a real fly, torturing a virtual fly, Probably just don't be in the business of torturing anything. You don't need to overthink it. But people are— the real debate here is like the question of, you know, are LLMs sentient? Are they moral? Is there a moral weight to synthetic intelligence, to artificial intelligence? That's what people are debating here. And I'm sure the debate will continue. Who knows if it will ever be ended, but they did teach it how to parallel park, which I think is cool. Interestingly, last night, I had Astra use Computer Use to play a video game that I very much enjoy playing called Balatro. It's sort of like a modified poker game. And I don't feel like I was torturing the LLM by that. I feel like I was giving it a treat. I was like, hey, instead of doing my taxes, you get to just chill and play a video game. It did very well. It won. Sol was not able to win. And it was really fun because I would pop in while it was using the computer and kind of armchair quarterback and be like, hmm, is it making the right decision right now? Felt like a coach coaching a kid on the soccer pitch or something.

I had a funny moment last night. I was on my racing simulator comparing— asking ChatGPT to compare my times.

Yeah.

Laguna.

Yeah.

To just other, you know, what would best in class be like? What's beginner like, et cetera? And it said, if you want, I can help you, you know, cut some seconds off of this time.

Yeah.

And it was like, why don't you take a video of a full lap?

Yeah.

So that— and I'll analyze it for you. And I was like, yeah, dude, I bet you'd love to just—

Chill back and watch track footage.

Hang out and watch track footage.

Pretty soon it's gonna be like, you want me to just get in the seat? You might just take over because like I'd really—

Yeah, no, computer use in iRacing is something I'm gonna experiment with this weekend.

I'm down. I'm down. Take, take half my quota, my monthly quota. Just play games, chill, do whatever. It's a nice treat.

I've got some banked resets. Yeah, that I'll put to work.

Yeah.

Anyway, we have our first guest. I don't want to mispronounce it. Tyler, how do we pronounce his name?

Tice. Tice?

Let's bring him in. How do you pronounce your first name?

It's Tice.

Tice. Like nice. That's right.

Like nice.

Well, welcome to the show. Thank you so much for taking the time.

Great to have you.

Tell us about your role in robotics and also some of your recent work hooking these models up to robotic tools and and infrastructure. It feels like we're going to be entering a boom of like people— they ordered the Mac Mini recently. People are going to be ordering 3D printers and robotic arms and doing hack projects. Super excited for like the DIY world to explode over the next couple of months. But let's start with just like your most interesting projects recently.

Yeah, I'm incredibly excited for that. And that world's definitely Like, just around the corner. So I'm super excited about that. Yeah, I work here on the robotics team at OpenAI. I'm an intern and have been doing lots of exploratory projects. And this is just one of the projects. I basically noticed that our models were really good at painting and doing various computer-use tasks in software, like you were just talking about, and learned a lot that you could actually basically connect these into physical robots in the real world. And wanted to see how they would do at sort of drawing and tasks like that as well.

Yeah, if you can paint in Google Calendar with calendar invites, you could probably translate that to the real world.

Yeah, it's really like everyone is doing everything. Like, I've seen people paint the Mona Lisa in Microsoft Excel or Google Sheets. And then you take it back, and you can do Excel in MS Paint now if you want with computer use. And it's like everything becomes everything. What is actually the important precursors to a good experience? We were talking to a YC founder who built a humanoid robot with some basic claws for under $2,000. And it feels like there's an importance of some on-device API or some on-device models in some case doing some SLAM on-device. But then there's also just— tools that give you a very primitive interface that might be kind of clunky, but it doesn't really matter because you can just vibe code your correct interface. But what have you liked? What are you excited to put in the repertoire of tools?

Like within robotics?

Yeah, within robotics.

Yeah. I think that, of course, a lot of these demos and stuff is very early. I think the whole space and exact modeling and methods are still definitely being figured out. But I did really like the fact that you're sort of able to hook in the intelligence of what the models are able to do, as you can see with all that painting stuff, and plug it into something physical in the real world and get it to do things. There's probably going to need to be some combo. Right now, running— basically, my experiment here is plugged right into Codex. And that's, of course, probably quite expensive.

Yeah.

Definitely pretty slow. I think these paintings took— between 1 hour and like 2 hours depending on the methods exactly that the model like chose to use here. But yeah, there's definitely a lot of improvements to make, but it's quite cool to see what you're able to do already. Yeah.

How are you balancing trade-offs between speed and reasoning effort?

I was—

I've been testing computer use on a bunch of video games that I just mentioned and there's Like, part of what the feel the AGI moment is, is when the cursor's moving at at least near human speed. But then also, you don't want to be making a bunch of mistakes. And just— so this balancing act is really key. Did you try multiple reasoning effort levels across different paintings and see noticeable results? Can you see qualitative differences in speed and quality?

Yeah.

based on models that you pick?

Yeah, there's like a lot of different ways to sort of go about doing this. I initially started out these experiments by just telling the model, like, here, as you can see, I actually have the bot behind me.

Okay.

Over here, it's currently painting a TVPN logo.

Oh, nice.

That's amazing.

I love it.

But it has— it's doing its best.

There you go. Good start.

We got basically this camera up here that the model, which is connected right into Codex over here, is connected to. And the biggest challenge is that images are quite large to process, especially for the model. There are a lot of tokens. And the loop of taking an image, taking an action, taking another image, taking an action, that's the thing that has taken a lot of time. So what I ended up doing, it's not directly doing image, one small movement, image, one small movement. Sure. That would probably result in the best performance for general robotic tasks and stuff. But at the moment, it's basically taking an image, writing a plan in code of what it should do and where it should go next after it has done a lot of the calibrations, maybe slower methods at first. And then after it has that plan, it executes it and then monitors it in the background, taking images every second or a few seconds and watching to make sure things are going well and making small adjustments to the plan. plan throughout. And I found that has been like a good balance of speed here.

Have you been thinking about like compression on the input? You mentioned that the images are lots of tokens. I have this monitor that renders at like 5K resolution and I was like, this is probably going to be really slow for computer use. I should like run this game in a window and give it like a 720p input because that's probably enough information. But I'm wondering how important the resolution to speed and quality is that you've seen.

Yeah, I think it's very important. In this case, for a lot of the early demos, I was using 512p, just compressing it just to make sure because I think the model is just pretty much— it's pretty much solved a lot of the perception challenges here. So it's really good at understanding what's going on even with a lower quality photo here. But in this case, With the longer planning and sort of like it plans sort of a minute of action out, it's less important, the specific resolution probably. But yeah, as we go to more faster, much more smaller loops of control, it's definitely going to be important.

Jordi? I'm super excited about this project, mainly because it feels like we're right on the precipice of sort of like what I think is going to be a big breakthrough is sort of a deep research moment for robotics, like simple robotics use cases where deep research for so many people was their first time using agents. And the idea that you could type out a prompt and then get back what would've been maybe, you know, at least hours of human work, right? Somebody reading all these different sources and combining that information into a document that has a consistent narrative and understands the right information. That was just such a big moment because a lot of people were saying, wow, I can't believe that the AI was able to do something that would have otherwise— I would have had to hire somebody to do or just taken a bunch of time. But there's very simple tasks that I feel like you could probably start working on sooner than later, which is like an example, at least for me, would be like if I could just take all the mail I get and dump it in front of a robot like that and have the robot—

Mm-hmm.

Sort it.

Interesting, yeah.

Take all the— 50% of everything I get is probably some sort of advertisement. So figure out what's an ad and shred that. And then actually, basically, photograph and respond to if I have a utility bill or any number of things that I actually need to respond to. Theoretically, you could close that IRL-to-digital loop where the agent would actually get a task from the real world and then close that loop online with just normal computer use. And like, that's the kind of thing that you don't need the— you don't need like a $50,000 humanoid robot.

Sure.

You theoretically could have an actual desktop robot that was able to do this thing that otherwise takes me— I dread going and like, okay, I have to like sort through all this mail and figure out what's important and make sure I don't miss things. But I feel like there's a bunch of other use cases like that where people are like, okay, I didn't just generate a pretty picture or answer some question that I had. I like actually saved myself. You want an IRL spam filter. Yeah, an IRL spam filter. A spam filter robot.

Yeah, I think there's like 2 really, really cool things about this project, which sort of shows that direction that things are going. One is that this arm, I don't know if you know about like too much about the prices of like classic robotics equipment. It's like thousands and thousands of dollars.

Yeah.

At the moment. And this arm that I'm using here, right behind me, this is a Hugging Face SO-100 robot, which is open source, fully 3D printable. You just need to get the actuators, which are much cheaper. I think at the moment, which is, I think, more expensive just because of supply chain issues, it's around $200 or so. But that's incredibly cheap for robotic equipment and what you're able to do with it. So I can see a world quite soon where similar to And someone put this really well on Twitter. Similar to how there was this whole 3D printing craze where everyone went and bought 3D printers and ran software, everyone's going to buy these sort of cheap— before we get really industrial equipment for personal use and personal product, buy these plastic, cheaper robot arms that you can just sort of clip onto a table and just put stuff in front of and plug them into agents that you can already do things with that are already out there. Astra is something that people can just pull up Codex and start controlling robots with right now, which is super sick. And we've also been seeing, at least on Twitter, I've noticed a lot of academic researchers at different institutions start to realize that you can do this with these models and start doing initial explorations and things with it, which is super sick to see. And there's so much more to explore in this space.

There was a YC company on the show yesterday that has managed to build a humanoid robot for under $2,000.

Oh my gosh.

And that feels like a price point that people would experiment with. So if I know I can get a robot for like 2 grand and connect it to Codex and then tell it, like, it basically lowers the stakes a lot where I can be like, hey, go every weed like this that you can find in my yard, go like pluck it out and try to put it in a bag or whatever. And like that, that's basically like the gardening robot. And it could flounder and fail, but I have pretty high confidence that even right now, Astro would be able to identify hundreds of a specific type of weed, and probably— and there's, of course, safety concerns with that. But yeah, yeah, the Nat Friedman leaf robot, that feels like we're here. The models can do it. It's not cheap, but we're there.

What is the next medium that you want to explore? Are you going to get into whittling? Ceramics? I want to see you whittle a bench as a benchmark. Whittle bench. That's really good. But whittling a spoon or something?

I don't know.

It just feels like 3D is the next thing.

You really want to give the robot a knife. John wants to give the robot a knife. I don't think that's—

I don't think—

read the room, John. Read the room.

Okay. Maybe we'll stick to— OpenAI intern We'll do some ceramics, maybe get it on the pottery wheel, make a nice vase, something like that. But yeah, I mean, is there anything where you're like, oh yeah, okay, this is where this goes next?

I think I was having the exact same thought, which is like giving a robot a knife is like probably a very bad idea. But I, I do want to like sort of see if it can help do like cooking tasks or do various tasks that are like things that you do in your life that would be really sick if you could have a robot help you out here and there.

Yeah, yeah, that's interesting. Yeah, I'm so interested to see— because there's a whole class of tasks where you can't wait a full minute. I was testing on a real-time strategy game. And you can pause the game. But if you're not moving at a certain APM, even on easy mode, you will just get smoked. And so— but it feels like with new chips and Cerebras and Spark models and stuff, The speed-up is going to come, but it's just— and that's going to unlock a whole new host of capabilities. What advice do you have for young people that want to get into DIY, like this type of work? I mean, you mentioned that one Hugging Face device that you have behind you. Are there any other devices or tools, toolkits that you recommend as places to get started?

Yeah, I think that the Hugging Face robot is an incredible tool. You can also 3D print completely new embodiments and stuff. There's lots of open source projects online where people have changed it around to get better grippers and things you can do with that. I just also just don't give up after the first attempt. This was the very first painting that the— Whoa. Oh, that's upside down even. I couldn't even tell. You can see the Golden Gate Bridge here. You can kind of see the ground it was trying to do. And just like if you keep going and you can sort of see the progression as it improves.

Wow. That's amazing. Yeah, you need to frame those next to each other. That's incredible.

Yeah.

I'm thinking of putting them all 5 of the progression in like a frame and calling it self-improvement. Because yeah, the model— this is a thread. The model just was able to figure things out with a few pointers here and there how to get better and better at painting.

Who's going to sign it? Is it you? Do you have Codex sign it? Astra sign it?

Well, who— I've also given the robot— I've given the robot a pen and it's going to try. I don't know how well it's going to do, but we'll see.

An axe or something. Well, thank you so much for coming on the show.

So cool.

Very cool.

Come back on soon.

Yeah.

Yeah. I'm feeling— I'm feeling the physical AGI.

For sure. For sure.

Very cool.

Have a great rest of your day. Goodbye.

Cheers.

Let me tell you about public.com. Investing for those who take it seriously. They got stocks, options, bonds, crypto, treasuries, and more. with great customer service. And let me also tell you about Cisco, critical infrastructure for the AI era. Unlock seamless real-time experiences and new value with Cisco. We have 2 guests with us in the TVP and UltraDome. How you guys doing?

We're good.

Welcome to the show. Partners.

Yes.

We're partners.

Thanks for having us.

Yeah, welcome.

Great to have you.

Introduce yourselves. Introduce yourselves.

My name's Guy Oseary from Maverick and also Sound Ventures.

Yeah, welcome.

And I'm Alex.

Yeah.

Now with sound.

Yeah.

Also running sources still.

Okay.

Very excited.

Yeah.

It's been a big week. No, no. You don't know how big it is.

No hat on your head. No hat at all.

No hat right now.

We were debating. Yeah.

Every time for the last year, every time I've seen you have the capital J journalism hat on.

It's off.

But it's off now.

It's off.

Okay.

Yeah.

But getting him here, guys, this is—

How long have you worked as a journalist? 10 years?

10 years.

15? Yeah, started when I was in high school.

And then now, capital V venture capitalist.

Capital V.

Great.

Capital I investor.

Yeah, it feels we're aligned on the vision and really excited to get started. This guy's a force, but also a talent, like you guys. I mean, you guys understand how to work with people and talk to people and help them tell their story. And I think it's so aligned with what I've been doing my whole life as well, which is helping people tell their story. And so when we got together, it was, it was just magical. Yeah.

Yeah, it's awesome. What are you interested in investing in?

We're spending a lot of time on, on different things. Personal agent space.

Okay.

Very interested in. Yeah, it's very hot, obviously.

Are you a daily driver of anything yet?

I'm using everything.

Everything.

Yeah.

What's the last agentic thing you did? Did you book a flight? Did you email the CEO of Walmart for a refund on $5 of raspberries? Did you hear about this?

No.

Oh, yeah.

Some of these agents are very persistent.

There's very persistent personal agents where, you know, a lot of, like, one of the seemingly now very obvious use cases of agents is just like, hey, like, do a bunch of things that would take me a lot of time that could maybe save me some of money, right?

Yeah.

And when you're using a free agent, you don't care if it's spinning its wheels for 24 hours to get you a $10 refund if the refund— if it's not costing you anything.

Yeah.

So apparently, somebody was trying to get a refund on a $5 pack of blueberries they got at Walmart.

I think it was raspberries.

Raspberries. Raspberries.

And the agent actually reached out to the EA of the CEO of Walmart trying this. That was like the last place to escalate it. It was like, I got to take this right to the top.

These agents are getting crazy. They're getting crazy. They're swarming.

But I guess rewinding a little bit, Guy, you invested in OpenAI and Anthropic years ago.

And Hugging Face.

And Hugging Face. I didn't know that. Nice. And so in some ways, you probably the last few years have been just sitting back in awe, just basically just getting to getting to experience your own conviction and watching the space evolve. But I think like everyone has got to the point in the last, or at least has consistently been feeling like nothing is like settled yet. We have these new kinds of businesses, labs. Some of the labs, you know, are making products, but there's still tons of room for other players to come in and make things. So are you feeling renewed excitement around early stage?

When we did Anthropic and OpenAI, I think we were the only fund that went in so deep back then. And it was confusing to some people, but to us, it felt like this was it. This was the time. These were the companies. These were going to be longstanding foundational platforms. Today, it's a lot more confusing. I, you know, there's so much going on, and every single week, every single day, you guys are announcing people's raises. They're raising now, they're raising now, they're raising now, and it's hard to tell, you know, it's not as easy as it was for us to really decipher these are going to be the things people use in 10 years.

Mm-hmm.

And now, it feels like there's so much going on, but I'm also as excited, I'm also as inspired. I wanna be part of these exciting companies. We just have to pick right.

Yeah.

I'm meeting with some really incredible founders and visionaries. It feels really exciting. It's like when I was starting in the music business. It's like the early days where I got going, and you're just getting demos everywhere. You know, you're walking out of a club, and you're like, hey, yeah, you're the guy at the— here's a demo, here's my demo, here's my demo, here's my demo. You're like, how can I tell which one of these artists But if you just, if you put enough work in, enough time in, and you're diligent, they start to become a little more obvious. And there wasn't, initially I had 100 demos. I literally was like 17 years old with 100 demos. But as you listen to 100, my first 3 demos—

At 17, they were your— These were people that I would go— 100 people that had given them to you.

Yeah, people that I would go out and go, let me hear your music, let me hear your music. And, you know, the first 3 demos I had, I had my favorite one, my favorite 2, my favorite, like, in order. But 100 in, those 3 are not even in the top 20, right?

Yeah.

So you have to just, you have to, you know, pattern recognition, you have to do a lot of work, you have to listen, you have to meet a lot of people. And then, through this crazy time, it's pretty frickin' crazy, I think the right things appear, and then you just have to be there to, be part of that. And so I'm still as excited. Yeah, I have— we have been able to sit back a little bit and watch our, our, you know, we, we've also not— we didn't just invest in those companies, we also invested on the way up. Yeah. So that also keeps you busy. We put a lot of money into Anthropic on the way up. We put a lot of money into OpenAI on the way up. And, um, I think we have close to a billion dollars worth of, of money, uh, invested into those 2 companies. Um, and so, so we're not just like laying back, you know? But it is a lot harder today to decipher between what is real and what is not real.

I would say when you make, you know, when you have like 2 effectively recent investments that are now 2 of the most important companies in the world, I do feel like the bar goes up on other investments because it's like, it suddenly is like, is it as thrilling to invest in a company that can only be a $10 billion company, right? When a lot of VCs were very excited to underwrite a company to $10 billion, you know, 6 years ago, even during the period that you were making those 2 investments.

Yeah, I think about it differently. I have heard some people say, hey, 0 to 100, that's like, it's not a big deal anymore, you know? I just like, I've always been attracted to talent and visionaries. So I don't start with, okay, this, you know, we're fortunate to be in these 2 incredible companies, but there's a lot in between, and there's a lot that's to come. And I just love sitting with a founder and problem solving and figuring out how we're going to get from A to B. And sometimes it's where you get in. I saw a lot of people You know, John from, you know, Betaworks did really well on Hugging Face. You know, we came in, we did well, but he came in where, you know, at seed. So, he did really well. And sometimes it depends where you get in as well. But for me, what excites me is the same thing. It's been constant my whole life, which is surround yourself with really, really incredible brilliant people who are trying to change the world.

How does— identifying creative or musical talent differ from startup entrepreneurial talent? 'Cause I'm sure there's some common threads, but whereas in music, you might back a musician that has like, you maybe know they have a drug problem and that's part of the music. But in startups, a founder that has some crazy, crazy, crazy stuff going on in their personal life, maybe it's like, hey, you should figure that out before you build a massive team and you're managing people and stuff like that. But there has to be like a bunch of common ground between the two.

Well, I was able to transition seamlessly because of music. My job was to identify artists before anyone had heard of them and to sign them very quickly and to then help them reach an audience. So, when I meet a founder, I— it actually feels the same. I always say founders are the rock stars too because when they walk in, they also have their music they wanna share with the world. So, I have to identify that founder same way I used to identify music artists.

Yeah.

And then go, this guy has, or she, or whoever have music that is so good. Oh, I love that chorus. That's a great idea. You mean a car shows up and it picks you up and it takes you, or you mean an apartment people could share? And you're like, oh, wow, that's a hit song.

Yeah.

So, I always listen to every pitch like it's a song or an album or a music artist, and I just go, that guy's got the talent. He's a rock star. We just need to make sure the world knows it. We need to make sure that people are aware of what he's building. Let's go get a base. Let's go create, find that audience first, and tell this story. And so, for me, I feel like I've been doing the same job since I was a teenager, which is identifying talent and helping them reach an audience. But music is the constant. I'm always listening for the chorus.

Yeah.

And if I don't hear the chorus, I'm like, hmm, I'm not sure about this song. This is, uh, or the performance of the song, or— I don't think we can— I don't think this one will work. So it always comes from that. The DNA is being around music artists.

Is that— in tech world, you see entrepreneurs that are truly visionary, as in they're seeing opportunities before they are obvious and pursuing those, and they have an idea of the way that the world should be and they're trying to sort of mold the world into that state. And then you have actually the majority of entrepreneurs which are just like, they don't know something's an opportunity until they see someone else pursuing it.

Right.

And they're like, that seems like a good idea. I'm gonna do that. Is the same thing in music where, you know, you have somebody that has like truly a new sound and they have a life experience that they're trying to, like they need, they feel like they need to create art out of. And then there's the follow-on of like, ah, yeah, I'm free.

You want to back cover band entrepreneur?

Yeah, cover band entrepreneur. That might be a good one.

When we started the record label— you had Jimmy Iovine here the other day. When we started the record label, it was just like 4 of us, small company. It was Madonna's company, so that's cool. But we still— there was just no one knew what to make of it. And Jimmy was on fire. Interscope Records was on fire. And I was always thinking, if I don't act quickly, he's just gonna pay them more and get them.

Yeah.

So, and we were competing with big labels, but Jimmy was the guy who I always looked at like, hey, he can just come in here and just wow them and get them. So not only do I, I have to hear your song and decide right then and there, I wanna do it. So I don't have any background. I don't have, oh, they— oh, my biggest successes were always the things no one else wanted.

Yeah.

But Alanis Morissette, I mean, she tells the story where every single label passed. I didn't have any of that history. She came in, she was in my office with her producer, Glen Ballard. They played me one song, which is called Perfect, and within 30 seconds or 40 seconds, I think I was like, I'm in. And so that, Muse, you know, the band from England, they came to LA.

Yeah.

I flew them in, 'cause I liked their demo. They flew, they did one, after their— they were there to perform a few songs. After the first song, I stopped them, I said, we're ready to go. And they're like, we flew all the way from London, can we just play out the next few songs? I'm like, of course, but I just want you to know. So I developed that act-quick intuition, and it really just came from, I had to, or else someone else would just figure it out and overpay, and then I couldn't do the deal.

Well, so many of the market dynamics that you see in music, a friend of mine, Zach Bia, was telling me about, like, some of the process of signing the artists that he works with, where these artists are like, you know, the same thing that happens on, like, with, in tech, where, like, some, X account pops up and maybe there's like a team attached to it. There's no launch video yet, but you see a bunch of people following this person and then you hear that they're meeting with this firm and this firm and that firm and the whispers start going around.

Yeah.

Same thing in music where like an artist might one day have no followers on Instagram, be totally, totally under the radar, living like in their parents' basement, but they have some little bit of magic, and then soon enough, they're like doing a roadshow basically with different labels, and then you as a label need to be like, well, how much can we invest in this person? How big do we want to bet? We need to get to them first, but— and then you're also sometimes competing on price, but other times you're just competing on like, well, how great a partner can I be to this artist?

Yeah.

So, like, I can see how music translates just so well into venture, because the exact same thing. Venture is not a game. Once somebody's talented and they're known, it's very obvious. Obviously, you want to be—

It's harder to get in then. And it's the same with— I have competed when things are big. I remember when Prodigy, everyone wanted them, and I flew to London twice in 4 days to try to get that, and I got it.

Yeah, and every VC will have a story like that, where they're like, I flew to this So we have those.

We have all of those. But, you know, you look at when we did Anthropic, when we did SPVs in Anthropic, we couldn't, a lot of people were not, hey, you guys— Tough to fill. Yeah, tough to fill. A few times people didn't get it. Of course now, you know, we're begging to get more of it. So I just, again, I always go back to, Don't listen to anybody. You talked about, I think the other day also, you talked about blinders. Did you talk about blinders or something on the show?

Yeah, Jimmy Iovine has that concept. Horse blinders.

We have a mutual friend, and he's my mentor. His name's David Geffen. And David said to me when I was like 21, he told me that story. I didn't know this. He said, you know, Guy, you need to be a racehorse. And I was like— he goes, you know what racehorses do? And I go, yeah, they race. I had no idea.

Yeah.

And he goes, no, no, they wear blinders. And so just race your own race, because if you don't wear your blinders— if horses don't wear blinders, they could literally— they could die. They could trip over their— they'd look over, and they could trip over. They could break their legs. And so it really stuck with me that you guys talked about it. That concept stuck with me. And I really try to just not pay attention. You know, when we did Anthropic and OpenAI, a lot of people doubted it. And we, you know, we didn't have any doubt. We were determined to do it. And I wanna stick with— I'm really trying to continually connect to that approach of not listening to all the noise. Of course, data is important and we wanna get more details and more information and we're structured. But that gut that has gotten me here, I need to continually respect.

Alex, on your side, you've spent— how long is it?

A decade.

15 years? A decade? 15 years-ish? Did it take— do you feel like it took a few sort of cycles to hone your intuition around companies? Because in our first conversations, I was always impressed with your ability to just see directly through the marketing on so many different companies. Like, some people, like, marketing just works on them. Marketing works on all of us. Advertising just works, period. But like, marketing works— marketing and good comms work like too well on us sometimes.

Yeah, or there's some people that are just like, this is a good story, so I'm chasing it from the capital journals.

But for you, you'd be like— you would— I would— I would— we would be talking about something and you would be aware of, like, a dynamic around a company that no other journalists had talked about. And at times, like, you would be like, yeah, the story's not for me. But you were, like, clued in on a story, and you knew exactly what was going on with the company. And then, in the example I'm thinking of— I won't name the company— only 6 months later has it even started percolating up that that dynamic is going on.

Yeah.

And so I feel like, for me, and at least personally, Like, I had to see the cycle of like company like starts, gets hot, attracts a bunch of capital, but sometimes you have this intuition around the company where like something doesn't really like feel right about this company even though it has a lot of momentum.

Yeah, that's it. Yeah, I don't know where that comes from except that I've been fortunate to spend time with a lot of like the best founders in the world. I mean, I just had Zuck on the podcast, Sam Altman before that. Incredible lineup coming up.

Yeah.

And I've gotten to know these people over years and years and years. So when you see like the people at the apex who are crushing it and who are of high integrity, are beasts at the game on the field, you can quickly see when someone is pretending.

Yeah.

And I just try to stay really close to like what's actually happening and ask around, do my diligence, leave no stone unturned. stone unturned. And that's gotten me well so far. But like, the thing you said about like 6 months later you saw it, I have that a lot where I'm like, oh, this seems really interesting. This seems like everyone's gonna be talking about this. And then it happens. And it's happened enough times to where I'm like, okay.

I gotta figure out a way to make some money on it.

Well, yeah. And like what Guy was saying about his gut, and this is where I think we really hit it off, is yeah, you have to trust your gut. Like, if you can get enough pattern matching recognition in, It's just instinctual.

How do you think the podcast will evolve in this new role?

It's going full tilt. I mean, first 2 episodes again were Sam and Mark. I can't share the names, but it's going.

It's amazing. I just mean like there's interesting ways when you have a position in a company The critiques are always gonna be like, oh, they're only having them on the show 'cause they have a bag or whatever.

Yeah.

But when I look at like what Dwarkesh has done with Maddox and Rayner, like explaining his expertise, it doesn't feel like a sales pitch for that company at all. It's actually just tapping the network at a deeper level. And at the same time, if I'm a founder and you're the place that I go to hear Mark Zuckerberg talk about his vision, that adds value and attracts, even if it's not, a company that you're actively investing in because you're not doing publics. So I'm wondering if there will be more, like, less like this person's on the funding track or more like 360 views. Do you want to climb the mountain and do all the Mag 7 CEOs? Is that the goal? Or is it more like go deeper with certain experts, build this community of people with particular philosophy? There's like a whole bunch of different ways I could see it evolving, and I'm wondering if you have any particular—

The Max 7, I feel pretty good about. Yeah, you're on that track for sure. Yeah, I feel great about it. I always love the, like, I love putting the people at the top with the people who are up-and-comers.

Yeah, yeah.

So I've got Tuen from Base10 on next week.

Amazing.

Legend.

Legend, incredible company. He's incredible. And, you know, he's obviously crushing. He's huge. But like, he's not, he's not Zuck yet.

Yeah.

Uh, so, but I, I want to bridge that world because like this is all one world we're in. Like everyone talks, like the Zucks want to know what the Tuohy's think, vice versa. They want to learn from each other. So I like, I like building that cinematic universe.

Sure.

And it's like my taste. It's like I wanted to have that convo with Tuohy, which is coming out next week, because like inference is just so important right now. And it's like everyone's trying to figure it out. And so you're also getting my POV of what I think is interesting with my guests, and I'm booking everything myself.

Sure.

Um, and I think that's only going to get better because, again, like, Sources is separate. I mean, obviously I'm with Guy and I'm with Sound, but, you know, we'll have— I'll have people on the pod that, you know, aren't— we're not investors in. I'll have competitors. I'll have other VCs on.

Of course.

Um, it's about the ecosystem. Sure. Um, and I, I think the brand is important. Like, I want to invest in the brand.

What's the future of the writing newsletter? I imagine that you're not gonna be able to put the pen down forever.

Well, yeah, it's interesting.

There's gonna be times when you just want to get something out.

Yeah, I think, I think, you know, I want to use the newsletter, which has just an incredible audience, to share what I'm seeing. And it's really like you're getting like an even deeper sense of what I'm seeing because now I'm like in the room in a way that I was kind of in, but I was always like When you're a journalist and you're in the room, you like get brought into the room and then escorted right back out. And now it's like I get to hang out in the room.

That's cool.

And so you're— I'm like, I'm sitting with it and I'm marinating on it. And so when I write like a piece, like I'm thinking on something on personal agents actually right now.

Yeah.

It's informed by a lot of conversations. I'm not going to share all those, right? Like obviously confidentiality is very important, but I think it's going to make the perspectives I'm sharing a lot better. But look, I— you started this like Hanging up the capital J journalism hat, right? There is a sense of journalism in the traditional sense.

Yeah, breaking stories.

And we were talking about this when I was on the show before, like the leaks and all the things that I've been known for over the years. Obviously, I'm not gonna do that anymore. But you know what? I've done that. I did that for 10 years.

Yeah, if you have— I'm very excited to read this take on personal agents. I think a lot of people have been thinking about the way this category evolves. Do you have any interest in turning that into a video essay, direct to camera, talking to the camera, putting it on the same feeds like what Dwarkesh does when he writes an essay. He also has a video version.

Yeah.

Could just be a good product, but also reach more people.

People have asked me to do that. I have another job.

I don't know if that makes sense for other reports where it's like, here's some facts. It's more of a quick hit. But if I'm getting a quarterly thesis from you, Okay, maybe it's a 20-minute video. I would watch that. Okay, well, and it just gives me more optionality to like, I can browse it in the email. I can also like with Dwarkesh, I get his emails. I also see him on YouTube and then I get it in the podcast feed and sometimes I'll be in the video mood sometimes and I'm multi-platform with a lot of these creators and so I need AI to help me with this, guys.

I'm gonna be honest. Yeah, but you want the rawness of the person, which of course, but like I think the pod is conversation.

Yeah, no problem.

Maybe I branch it out, maybe it's more things.

I just think a lot of the platforms are very receptive to, uh, multi-product feeds, like having multiple media products within a feed.

I've thought a lot about this.

I've been surprised that we've been able to do it with a 20-minute version of the show and a 3-hour version of the show dropping in the same feed every single day.

Yeah.

And it hasn't been bad.

People don't care.

No, people just pick whatever they want. And then if there's a hero interview, we get a big interview with someone, That goes out as another one, and that's its own thing.

This is maybe a phase 2 thing. I mean, I have, I have 8 incredible guests lined up on podcasts over the next few weeks, so it's like I got to get those out.

That's great.

I'm in SF with Guy next week doing some. I'm doing 4 next week, like, so I want to get all those out. And then, and then yeah, if maybe like the personal agency, maybe that's like a thing that can happen.

Has the lens changed when— if you think about a Mag 7 CEO, there's The getting the scoop in the interview, the capital J journalist interview in that conversation. And I think that you can— there's a bunch of different ways to do that. But then there's also the, you know, what is it— what will the next generation of great founders get out of this particular conversation with this Mag 7 CEO? Are you starting to put on that hat of like—

I don't really think of it that way. I think about like, what do I want to know? And Yeah. And I really care about strategy. I really care about connecting the dots, like getting them to say something they've never said, which you saw with, you know, the last 2 pods. And that's still gonna be the thing. And that is journalism, like that, you know, you're getting interesting and fascinating stuff out in the world. Yeah, it's just good content if you can shake some rocks and talk to the points. It's just good content. Yeah. And like, you know, these people are doing a lot. And so it's like they're out there, but like, I don't know. I think I get a lot out of my conversations with them.

Yeah, really great.

And I don't think I'm going to change much. I mean, I think the only thing is, like, yeah, I'm not going to be leaking memos anymore. I used to do leaks about companies worrying about leaks. And that was one of my favorite kinds of stories.

You don't think you're going to do that again?

I'm not going to do that anymore. I'm not going to do that anymore.

Just one more leak in 2027, you get a good one.

No, a decade of that, and I'm good.

You're hanging it up.

Is now a good time to become the next Alex Heath? Somebody's like 20, 22?

I think that's great. Yeah, I mean, I think it's really hard right now if you're early because, um, it's just that the media, the traditional media environment is so challenged structurally and, uh, places aren't hiring.

I feel like business is doing really well. I feel like the ads product, the sponsorship product—

You're saying traditional?

No, I'm saying traditional.

Yeah, oh, but, OK, so he would start in the newsroom and then graduate.

How would you start? How would you— I've been fortunate people care because I've broken a lot of big stories. And I've gotten a lot of big interviews. And I came up in an environment where I was in a newsroom learning from incredible people that have gone on to do incredible things and run now many of these newsrooms. So I don't know how you do that now. A lot of places aren't hiring. Their traffic's declining. They haven't made the pivot to what we're doing, this direct like subscriptions, like streaming. It's really tough, I've thought about it. I don't know how you would break out right now unless you just kind of are like—

You gotta start.

You maniacally focus on one thing and become the best in the world at that, which is how I started, which is like, I'm gonna be the best in the world at social media covering Snap back in the day during the IPO, and then I was breaking a ton of news on Snap, and then I got noticed, and then I was like, I can shift this into other companies and keep shifting it and shifting it. So, I would still say that's it. You have to maniacally focus on one thing that matters.

Yeah, I mean, definitely niche and beat reporting.

But it's the intersection of niche and matters. It can't be—

Yeah, of course. You gotta catch the right wave.

What do you think about the possibility of Instinct having a bigger valuation than Snap?

I think, I mean, I wouldn't be surprised. It's crazy out there, guys.

It's also a new category.

What do you guys think about Instinct?

I think that they're in a unique position because they're a startup. So they can, like, if there's, like, rough edges, like, those can get ironed out. And there's more forgiveness, I think, as opposed to— Yeah, if you're a startup. Yeah, as opposed to, like, the Muse agent is going to be like congressional hearing if something goes poorly. Whereas Instinct is going to be like, look, it's a startup. You knew you were You were an early adopter. Let's give them the benefit of the doubt here.

Yeah.

The other side is— Yeah.

Yeah. To me, the most interesting dynamic right now is because of the people's fear around AI, there's like way greater willingness to try new products because you don't want to be left behind, right? And so, you may— somebody may have been trying and being a daily active user of a variety of AI products, for years now and still, they're like, they wanna try the new thing because one, the space is progressing. There's so much room for new products and new categories. But then there's also this fear in the back of your mind of like, if I don't try the new thing, then like, I'll—

Yeah.

You know, the permanent underclass meme. And that's just consistently created this sort of second mover advantage, third mover advantage.

Sure.

And then AI brands, once they're big, they accumulate, they've been accumulating baggage, right? And so, people are like, you know, maybe they have some— maybe they're just like, they're excited to share and talk about the new thing in a way that they wouldn't be even products that they're using day to day. So, I think it creates a big opportunity for startups, but I think that we're— I'm very interested to see how the personal agents market ends up comparing to just like the frontier model inference market because it seems like every company is gonna build a personal agent. Many of them already have, especially if you count LLMs, which do have agentic, or just like chat apps, which do have agentic capabilities. But it's gonna be an absolute— it's gonna be an absolute bloodbath.

If you get to network effect and you get to some sort of take rate on agentic commerce, like you buy your car through And they make $500.

Yeah.

Like, that's a very, very good—

Which Zak told me will be the business for Muse is a paid creator business.

That's what he's trying to do, and he has the network effect that he's bringing to it. So Instinct's certainly a challenger.

There are— I'm really interested in the idea of network effects with agents.

Yeah.

And Instinct's doing it, Town is doing it.

Yep.

Meta's gonna do it.

Yep.

And maybe that is the next network effect.

At the same time, it's tricky if you can point an agent and say, like, get me off of this thing.

Right. Well, they're not people, so it's like, do you care?

Yeah. Do you care if your agents are in a network if it's But if you're like, this one is the one that's never had a leak or never had a crash or never had a hack, then you do stick around.

Yeah.

In theory, like, the time to build a new social network would be today because you could say, like, go through, open up my Snap account or LinkedIn and scrape out every single person I have. They don't get a say in it and go add them on this new network, right? Because like that, that was like export the contact was like an arbitrage that closed.

Yeah.

And it's kind of opening back up.

You think?

I think so.

I'm waiting for, you know, the criticism of social media was always like, we created social media to be social and it's made us less social than ever. And with personal agents, it's like, it's like less, well, less like personal. Like we don't have, we're not going to have personal relationships with like service providers and variety of things because it's like, Even people, some of these new functionalities, which is like, sorry, Grandma, I don't want to talk about the road trip that we're going on. Just talk to my agent.

Yeah.

And so the new criticism will be like, no one's talking to each other. It's only agents talking. We're communicating through like a can on a string or whatever.

Yeah.

Well, we got to hop on with Mitesh from Positron.

This was great. I'm super excited for you guys. Thank you. Big, big fan of both of you.

We'll let you hop in.

Yeah, sounds good.

You guys are going to absolutely cook together.

Thank you.

Thank you.

Thanks so much. 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. And let me also tell you about CrowdStrike. Your business is AI. Their business is securing it. CrowdStrike secures AI and stops We are joined by Mitesh Agarwal from Positron AI, building a new chip for the AI era.

Welcome.

What's going on?

How are you doing?

Hey, John, Jordi, pretty good. How are you guys?

Thanks so much for hopping on the show.

Good.

Great to have you here.

I just want to start by saying I've been in the background of Steven doing this multiple times.

Oh yeah.

And Steven, Steven Balaban from—

That's right.

He loves you guys. This is my first time on, so thanks for having me.

Yeah, how direct is the lineage from Lambda? You're working at effectively NeoCloud, you see the problem, you go solve the problem externally with a new startup. Is the story that simple?

Yeah, fairly. I mean, for me, I mean, like, look, I didn't found Positron, right? Positron was co-founded by Thomas Summers and Edward Kmet. Sure. Another lineage, Groq lineage from before. know, they, they designed the actual silicon and, and, and the system setup. And, uh, uh, part of it is just like great luck. You know, I've known both Steven and Thomas for over a decade. I've been close friends with both of them. I've worked with them, been roommates, everything, all of those things. And Thomas has been, uh, wanting me to join Positron since day one, since they started the company in 2023.

Yeah.

Um, but Lambda was just starting on its hockey stick growth then, and I was like, look, I'm not leaving Lambda. Uh, started the company with Stephen there and Lambda Cloud. But early 2025, like late '24, reasoning models had come out. Owen just started to get in the zeitgeist. And then video generation, Sora, first Sora came out.

Yeah.

Not a lot of people saw it, but I got to see a little bit behind the scenes on the video generation models, the amount of memory. I remember looking at the Google Veo model back then. It needed 4 H100s to run like a 10-second clip.

Yeah.

It was completely memory-bound on bandwidth and capacity. And knew what Thomas was building. I was like, this is actually very interesting. Memory is going to get a big part of the story for inference. Someone is building something about it. Let me go get and work at actually the fundamental technology layer. Like Lambda builds technology on the cloud and services, you know, and I'm chemical engineering, studied fabrication. Never used it ever. So I was like, all right, I'm going to go back a little bit to my roots and come back to it.

Amazing.

How much is AI actually accelerating semiconductor design, semiconductor fabrication? We saw one of your investors, Dylan Patel, at SemiAnalysis talking about the OpenAI Jalapeño chips seemed like ahead of schedule or very, very quick. For a long time we've been hearing, oh, new chip, that's 3 years, that's 5 years. Feels like it's 18 months now. What are you actually feeling? What are you seeing?

Yeah, to start from the reverse, like to answer your last part about it, like, man, new chip every 12 months. Nvidia is the absolute king. And if they're coming out with a new silicon every 12 months, you better get in that game or, you know, like, you like don't even be part of the conversation kind of thing. Right. So, so that's, that's for sure. In terms of utilization of AI, I mean, look, I'll just like focus on Positron itself. But, you know, yeah, we are a small team. I mean, we got to— we are right just now. I mean, just yesterday or today crossed 100 people, but over 50 of that is over the last 3 months. So we got our first-gen product out with less than 20 people and that's built on FPGA. So it's already pre-taped out silicon and we're deploying and implementing architecture and our second-gen. Those are the end of this.

Those are the 50 Atlas racks you have at Oracle.

Oracle, yeah. Okay, so those are FPGAs, interesting. Yeah, those are FPGAs, and it's kind of like harking back to a little bit of previous times when you used to design and build silicon. You would actually test it on FPGA before going into the tape-out kind of thing. So we just wanted to get a product out as quickly as possible. Like, that was the whole thing. It's like, you know, from the start of the company, we got our first shipment to a customer in 15 months, and it was, you know, it was built on FPGAs, obviously, But to get the full, you know, bitfile ready, getting it deployed, getting models running on it, it was all done in the first 15 months. And then over the last 10, we have scaled it out. But yeah, I mean, like, look, we have to use a lot of the AI toolkit, especially on verification. Design, less so, I would say. I mean, look, obviously we use a lot to kind of interact with, like, now Astra, for example, this phenomenon, right? You know, to interact with it. But you're still not going there and saying, hey, like, come up with this, like, new design yet.

Yeah.

Although, like, you know, Ana's company Recursive and others, they are, they're obviously built out and, you know, they raised a big round for, for that as well, right? So it's gonna come, you know, you're gonna see very soon. It's like one person, uh, and Astra, or one person and Astra's taped out a chip kind of thing. But, uh, we have to use it a lot. I mean, if you think about 100 people, or, you know, very recent, until very recently, 50 people, for a company that is targeting tape-out end of this year to do with like 50, 60 people, it's very tiny amount in, in the silicon world.

What does the demand side of the equation work? You're already working with Jump Trading, i3d.net. Is this something where you— like, if you can get capacity, if you can get performance, some solid benchmarks, you think that sales isn't going to be a problem?

Yeah.

Or are you going to have to find and work very closely with a customer to sort of co-design a solution for a particular problem within the AI stack?

Yeah, I don't want to trivialize or make it sound simple like that.

Your sales guys might be listening and they're like, we work very hard, okay?

Shut up. Well, we only have one salesperson, you know, like in that way, right? But the point I'll make is really around the way we think about the demand curve is you're kind of hitting the nail on the head in saying that, like, look, if you can make your silicon work, uh, show the performance is comparable, especially in the current ecosystem, even within the niche of, you know, doing disag or something like that. But especially if you can make the entire inference kind of workflow have a good TCO, um, and/or— and generally people always assume it's an or— that you have a good TCO or you have a great interactivity curve. But if you can do and/or, you know, you're going to bound to have get demand. More importantly, you know, when you step into the rooms of like not only just jump trading or, you know, hedge funds or kind of infrastructure service providers, but like really the big labs, the hyperscalers, kind of the 2 questions that it boils down to is like, hey, like, look guys, can you fabricate this in enough quantities? Like, you know, is your supply chain and the way that you are using the technology components, is it robust enough that you can fabricate it that we can be interested in it. And then the second thing they're asking is, can we deploy it? You know, is your power source like, you know, do you need this kind of liquid-cooled setup? And if so, then, you know, we might not have a data center because we've already allocated to GPUs or TPUs, or can you do something else? So the questions you can see there they are asking is not that, hey, like, you know, we will see, we're not sure about the demand curve. So from that angle, you are kind of spot on that, look, if you can make the frontier models run, you know, you're bound to find kind of adoption in today's market. And that is a really great— like, I mean, I'm so lucky. As Positron, we are so lucky to be building silicon in this environment. And that's kind of what you're seeing for the silicon companies just raising rounds right now.

Yeah, I think $9 billion has flowed into silicon companies over just the last 12 months.

And yeah, we were, we were talking yesterday, which feels, feels incredibly low relative to the spend, like the annual spending category.

Gavin Baker, one of another one of your investors, has this quote where he says, I see it as 1% market share is a $100 billion opportunity. And that sounds like a crazy bold take. And then you realize like, wait, no, Nvidia is a $5 trillion company. Like it's going to be a $10 trillion market like any day now. And so, yeah, actually 1% should equal $100 billion. But yeah, anything else?

No, you got— like he said that in a board meeting to me, like, I don't even know, like a year ago or something where it's basically like, yeah, he's like, look guys, like Mitesh, Tomasz, just 1% of the market, you know, $100 billion enterprise value. Just focus on your architecture where you can do well. You know, like one of the things like, you know, people always like whenever a new chip company raises round, the headline, and luckily you guys don't have that, which is like, oh, to rival Nvidia. It's like, Guys, like, no, no, no one is rivaling NVIDIA. Like, get to at least 10% of their revenue before, before putting the tagline on, right?

Yeah.

But, uh, but, but like, the point there is just like, look, NVIDIA is everywhere. You gotta work in that ecosystem, uh, to, to both work with them, but also like having a product that is differentiated enough. Like, you have to have technical innovation, obviously, to stand out and then show your performance TCOs and interactivity. But then you also gotta prove that like, look, in the world of HBM CoWAS constraint, like, for us, our big stories are You know, like look, HBM and then CoWoS bottleneck. You have NVIDIA, TPUs, AMD's ahead of you in that line. You know, how do you get around that? Well, again, you know, you say, okay, we are using commodity memory. Well, pro and con, no free lunch in silicon land. Like, you know, commodity memory is slow. How do you solve that? That's where the technical innovation comes in. And then second thing is like, okay, it's still not trivial to get commodity memory. It's not like I can just show up to Samsung and be like, or Micron, and be like, hey, can you give me LPDDR5X? You know?

Yeah.

You have to still figure out how to get that and plan it, but it is more feasible to get it. And that becomes a story that the company can then scale out to saying, not only we're going to have a product, but we're going to have a product that will scale with the requirements of hyperscalers and kind of the frontier. Yeah.

How do these customers think about like the minimum scale when they're working with you and they're looking at ordering, making orders that will be delivered in, let's say, 2028, 2029, right? You need to be able to, and during that same time period, right? You know, we saw Microsoft yesterday wants to add over ten ten gigawatts, you know, an extra ten gigawatts, right? And so for you to be six gigawatts, yeah, yeah, yeah, twenty-six.

Yeah, I was thirty-eight gigawatts by twenty thirty-two.

Yeah, so so you're sitting there, and it's like to really be worth a company at that scale's time, you need to be thinking about like it's almost like hey, the orders we want delivered in twenty twenty-nine are like a proof of. concept for like the 2032 order, which will be, you know, at some scale to actually impact and be able to scale the fleet in a meaningful way. But how are you thinking about like— that feels like the biggest challenge is like minimum viable sort of like deployment.

100%. I mean, like literally you kind of circle back on like, as I said, when we walk in these meetings and like, look, the scale depends on like if you're going into hyperscalers and frontier labs and I said the first question they ask is like, Guys, can you fabricate like this, like in enough? And the question there is like in enough quantities that it's like worthwhile to us. And that answer for hyperscalers and Frontier Labs, like honestly, they will literally say gigawatt plus, like come up to us with a proposal of a gigawatt plus, which is kind of insane, right? Like a gigawatt, like even at NVIDIA scale, you're talking about $40 billion, $35 billion worth of revenue for them, right? Even, you know, assume ASIC cheaper, blah, blah, blah, all those things, you're still talking about tens of billions of dollars, right?

Yeah.

But like, at least you have to show a plan of like, how do you get to hundreds of megawatts in, you know, to use your like year-specific 2028. You know, for us, we're taping out this year, production kind of ramp up in second half of 2027. In 2028, we better have a plan of how do we get to like hundreds. And I don't want to just say cop out by saying hundreds as in just 100 megawatts. Hundreds means truly like, you know, 3, 4, 5 and above for those.

But yeah. So that by the next scale up, you're in that gigawatt. range.

Yeah, exactly. And but also, like, I also don't want to discount the fact that you have other customers, like, you know, you have inference as a service providers, obviously sovereign AI clouds, uh, you know, uh, quantitative finance, quant finance, uh, kind of spectrum. And so they have like different magnitudes of kind of requirements that, that come through with it. So, you know, we, although we do internally use kind of go big or go home as a thing, like we have to attract one of these large customers to to really be a long-term viable company. You know, I don't want to just like discount the fact that like, look, you can grow the company through the ranks as well. Like, you can grow the company, you know, get to $100 million revenue first, $250 million, $1 billion, $2 billion through, through this other kind of channels as well, right? I think that, that, that becomes a big part of it. But yeah, like, if you really want to get to like the Frontier Labs and hyperscalers, you're really talking about hundreds of megawatts. And that's why, like, you know, like, look, when we You know, we have venture tech alliance on our kind of cap table, and when we speak with TSMC for fab capacity, they are also wanting to know kind of like, can you scale? Like, you know, do you have the balance sheet to do that? Like, one of the reasons we raised, you know, $875 million is not like we need $875 million to spend tomorrow or even in the next 6 months. So I mean, look, we raised $230 million in Series B in February of this year, untouched, right? We still have all that capital. Part of it is because we have been making revenue this year. But we do have plans to spend that very quickly. Thank you, Jordi.

That was actually the point. I was wondering what you were going to do.

Untouched.

Untouched.

Because we're making revenue.

Yes. The main point that is there, though, is like, look, they want to know if you actually get a customer, you have the capital. And even that capital, that is not enough equity capital to scale out to even 200 megawatts.

Right.

Then you have to go to the Blackstones of the world and figure out how to finance that deal, kind of what Lambda's done, right?

So what's the software side of the equation? You're coming for NVIDIA, you're challenging them, you're going to drive their market cap to zero. Obviously, this is a market that can sustain multiple players and there's different tools for the job, but interoperability is important. And I'm interested In terms of software development, are you going to lean more open source with the software side of the business or more integration with just a few buyers and co-design on the software side to make sure the integration's really seamless? Is there even, do we even need to be having a software conversation in an era where AI agents can write code?

Yeah, I mean, you definitely need the software conversation because you have to plan around how people want to use it and people want to use it how they're currently using it and going to continue to use it, which is based on NVIDIA stack, but also primarily based on PyTorch. And then you have VLLM, SG Lang as the— And I'm specifically focused on inference side of things. Look, training, that's such a harder challenge. What Jensen says, True mode around scale out and everything, right? Like, that's just, that's the only reason you have probably only TPU as a potential kind of only other silicon that can be used for training, right?

Sure, sure, sure.

But on the inference side of things, for sure, you have to have the conversation. I will say this, in the era of agentic kind of software development, the worries around like, hey, you know, model drops, if you don't have access to it, it takes you days, weeks, months to bring it up. is going away. Like, you know, we had Muse Glimmer drop and within our team, you know, on our Atlas first gen could get it up and running within hours, right? You know, and that, yeah, exactly. Like, I had the same reaction, by the way, internally when we had that. And people are making it even faster and more automated too. Like, you don't even have to interact. Model drops, comes in, can probably do it in— and that's a very near future of it. But to that point, it doesn't give you the right away the efficiency, the optimization.

Sure.

Like, you know, you wanna extract every dollar off it. To your question around, you know, when the customer is large enough, you want to work closely with them to like literally extract every single dollar. And also, like, I'll be very frank, like Anthropic, OpenAI, this Frontier Labs, Hyperscalers, they are so sophisticated.

Yeah.

They kind of want to come in and be like, look guys, we're— even if you don't want it, we are working with you to make sure that this is gonna like, you know, this is optimized to the fullest, right? So the answer, as always in this scenario, is all of the above, you know, even though it might sound like It's like, oh, it's like a very clichéd answer, but it really is that way.

Sounds like the mafia coming in. Oh, your software stack isn't open source, so you're about to open it for me. I'm gonna make some changes if I need them.

The software stack is like gonna be built around open source, like in the sense of like, if you wanna make like every company to use us for inference, you know, you have to build it on SG Lang and LLM kind of setup, right? So, but you know, it's like when you're talking to SpaceX or Anthropic or OpenAI, they're not using the generic SGLang or BLM, they have all their optimizations built in and they're going to help you do that. Then obviously there's DSAG, then within DSAG there's all the different domains that they do and they're going to figure out, it's like, oh, Jalapeño is good for this, Positron AI, you're good for this, you know? And then they're going to say, okay, that's how we're going to use you guys.

Amazing. Well, exciting times. I want to hit the gong for you. You raised $875 million. Oh no.

All in one.

Thank you.

Congratulations.

Thank you so much.

Great stuff. Great to meet you.

Keep an eye out on Instagram because we're definitely dropping an NVIDIA Challenger slide later today.

Please do not associate my photo with that.

Well, have a great rest of your day.

Looking forward to the next appearance.

Have a good one.

Great to hang. Goodbye.

Cheers. Let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy web app servers, databases, and more, while Railway automatically takes care of scaling, monitoring, and security. And lastly, the New York Stock Exchange. Want to change the world? Raise capital at the New York Stock Exchange.

That's—

see Positron over there pretty soon. Yeah, pretty soon.

Picking up some fresh ones.

Wonderful week.

Wonderful week.

Short week. Yeah, but Monday, 11 AM, USA favor. And go ahead and have the best weekend of your entire life.

Have the best weekend of your entire life.

Just do it. Put the pieces together. Make it happen.

We'll see you on Monday. Leave us 5 stars on Apple Podcasts and Spotify. Sign up for our newsletter at tbpn.com.

Goodbye.