Anthropicの2兆ドルIPO、マスク氏のネオクラウド淘汰予測、追いつく中国のKimi｜#306

Guys, I'm curious, have you been tracking what's going on with the protests and revolts in France, Italy, and Spain? I mean, it's kind of insane. It is, but I've talked to people on the ground in France, and they're like, the media's way, way overblowing it. There's like a few bad people, and everybody else is just moving around drinking coffee while the things are happening. So it's a very split. I don't know which way to think. You could go either way on it. Yeah, I mean, it's like, you know, October 17th, there were half a million people marching across France. They had 2,400 arrests, you know, looting in Paris, Marseille and Lyon. And they burned down that beautiful 19th century hall in Limoges up in flames. I mean, I hate this stuff. There's a bunch of paid crusaders. they suspect Russia naturally on this that are kind of stirring the pot really badly and that's antagonizing everything obviously

yeah but I mean underneath it all as well it's like there's a 20% unemployment rate in France, Italy and Spain that's got to hurt it's got to hurt thank goodness we have no protests ever in this country you know we always say the french are revolting you know i wrote um you know i wrote a sub stack a few months ago i think we talked about the pod that the most dangerous demographic in history is about to get a lot bigger and what bugs me is if you look across the last two three hundred years from the french revolution to the bolsheviks to iran in 1979 to arab spring they all share one ingredient. You know, young men, testosterone laden, 18 to 28, educated enough to know that they're missing the opportunities and they're blocked along the way. And I mean, this is, you know, Dave, when we talked to Elon about, you know, UHI and civil unrest, and he said,

we're gonna have both. I mean, this is probably my single greatest concern. And we have a good place to keep an eye on it, too, because, you know, there's no data center activity, no power, no, you know, incredible educated population that should be capable of amazing things. But without the government getting ahead of either compute or data centers or power, you're absolutely screwed in the age of AI. And when you look at the way the U.S. voters act, they vote for, you know, UHI within the borders, but they will never vote to bail out France or Spain. France actually has nuclear power. It might be okay. But Spain, Italy, there's no sentiment in the U.S. to help or in China to help. So if you don't take control of your own country and get your government, you know, get its act together, the 20% unemployment rate is only going to go up. Yeah. Because AI is the workforce. Yeah. What's it like over in England?

It's a bit depressing here. They started sending emails, letters to all the rich people and entrepreneurs. We're rich as above £200,000 of earnings saying, I'm your new compliance audit advisor. Which is happening? The government sets up a compliance audit liaison team to make sure that you're compliant with all your taxes and pay a bit extra. And then all these policies coming in, the entrepreneurial class are not feeling very happy about here. A lot of people are about to leave. and we've got the budget coming up. And then, yeah, looking over to France, it's not surprising. But in France, it's basically impossible to fire anyone. This is the other thing. Do you know this thing? 57% of their GDP on social programs. It's more than any communist or socialist country in the world. That would surprise me. Yeah. I think you've got a lot of aspects of that. People are fleeing to Toronto and Waterloo like crazy. Salim, I'm sure you can see it. When you walk around Waterloo, everybody is the top of their nation from all over the world.

But you're used to seeing people coming from Asia and India. Now floods of people are coming from Europe as well. Because if you can't get into Silicon Valley or Boston or New York, you can often get into Canada. But anything to get out of the slow zone in Europe and get into the AI zone, which Waterloo has become. Iman, when are you joining us? When are you coming to this side of the world? let me finish building a few things first I can read the voice from across the pond until then actually there's something very interesting in AI Poolside which was the GitHub CTO I believe went to France and they raised 200 million aided by the French government and they went straight back to America to release Laguna and then sold to Nvidia for 6 billion so now I think all eyes in the AI sphere are on Jan LeCun's AMI Labs which was a French company but again may become an American company sooner rather than later I might be on stage with him in a week, so I'll ask him directly. Are you French?

No, well, that's a whole other question. Yeah, but we need to keep an eye on what's going on there. I just think these are the early seeds that we need to watch. You know, it's easy to mobilize a lot of people around what you're against. It's much harder to coordinate that around what you're for. What's the alternative? Yeah. I mean, something I think I'd be interested in this. Like, as you talk to all these corporate leaders around the world, are they expanding their graduate programs? I very much doubt they are. It'd be like that's the first thing to go almost, right? Not necessarily. We're seeing some weird things. For example, a bunch of people are increasing graduate hiring because young people armed with AI are much more valuable than 50-year-olds moving spreadsheets around. And so some companies are actually increasing and replacing older workers aggressively and have increased graduate programs because they're just so much more AI native. But generally, it's a problem. That feeling that we miss the boat, that we can't catch up,

when you look at the charts later in this podcast about how the open source models are getting closer and closer in time to the frontier, the path to sovereign AI has never been clearer. And if countries were to act, they actually have a very, very clear path forward. I don't have any confidence that they'll act because they haven't acted so far. But if they were to jump on it and say, look, we need an emergency national measure here for power, chips, and sovereign AI, you can succeed with probably 90% to 99% probability of success if you can get rallied around it. But, yeah, I just don't see any groundswell to get rallied around it. You know, we just need one or two, maybe a South American country, to prove to the world that it is possible to get on the map. and that'll crack it open for everybody else. But, you know, the window is right now. It's a relatively narrow window, too, because the AI is accelerating at this ridiculous RSI pace. But there is an opening right now. You'll see it later in the pod. I think we look back at this and we say the singularity was the best of times.

It was the worst of times. It was the time when the country that arguably invented or helped to invent nuclear energy and should be absolutely drowning in energy too cheap to meter has students protesting that they're not getting proper heating or air conditioning in their schools. There are many, many profound ironies here, but I do think Europe more broadly needs to, if anything, not pace itself and catch up to the singularity, and maybe we'll see more activities like this, whether it's ostensibly through educational grievances or budgetary grievances or just air conditioning. Employment grievances. Employment grievances, social grievances, public service cuts. All of these are intimately tied together. One might say, yeah, sure, there's a bit of a protest culture in France. They've had a number of revolutions over the years, maybe culturally to some extent in their DNA. But I also think at the same time, Europe is very well positioned to become a living history museum if it doesn't catch up in energy and compute.

Well, I'll tell you, the U.S. is in great shape in every way except for power, which we've talked about several times. But yesterday I had a meeting with the guy who ran data centers for Google for about seven years, a fascinating conversation, but he's got a U.S. data center fund now, and the gap in power between the U.S. and China is going to expand almost 10 to 1. It's just crazy how far behind the U.S. has gotten in power. And he was saying, look, there's a lot of push from the White House to catch up, but after this next election cycle, things will probably get frozen again. And so Eric Schmidt has been pointing out many times that we need to work more closely between the U.S. and Canada because Canada has a lot, I mean a lot of power and not a lot of chips. And so there is a very natural partnership to be done there, and there's been way too much friction in the last couple of years. But even with the U.S. and Canada working closely together, China is miles ahead in power generation. Until SpaceX and Tesla bring on, you know, 100 gigawatts of solar each.

Yeah, true. But, you know, the gap in time there is really scary. It's, you know, solar is, you know, space-based data centers are going to be enormous. The Dyson swarm will be everything in the 2030s. But RSI is the next two years. And really, you know, a good strategy around AI really focuses on what are the exact moves for the next two years. Because after that, I mean, we're talking about intelligence like no one's ever even imagined. And it's really, really hard to predict the effect of new physics, new chemistry, all of Alex's breakthroughs. I mean, we're seeing new physics now. This is the new physics episode of Moonshots. Yeah, it is. You know, it's almost impossible to look more than two years ahead. It's difficult to look one year ahead, for God's sakes. Yeah. And, you know, to all of those in France, the country that brought us the word entrepreneur, we wish you best. Now, that's the Moonshot, ladies and gentlemen. This episode is brought to you by The Abundant Summit and Link Ventures.

theory of physics. More on this later in the episode. And Dave Blunden, my dear fraternity brother and buddy, AI VC extraordinaire, the man who's backing Unicom founders out of MIT and Harvard. Salim Ismail, author of The Exponential Organizations and the Organizational Singularity, frequent flyer legend, and our warlord against linear thinking. So you're clearly not at home, Salim, where are you today? I'm in Toronto because I'm doing a talk here to about 400 CEOs tomorrow and then flying back Thursday. Okay. You're sure you're not at a conference, at a concert? I'm not at a concert. I will report if I am. Okay. And Imad Moustak, founder of Stability AI, now building the intelligent Internet, and the man most likely to explain humanity's future economy. Iman, I really want you to weigh in on where the economy is going on one of these episodes. You've written an incredible book. And I'm Peter Diamandis, your host and data-driven optimist.

I just want to start out by saying something to all you guys. I feel so blessed to have you as friends and mates here. I mean, you guys are extraordinary, and I love the way you all think. And honestly, even though these are intense episodes, this is my favorite time of the week. spending these couple hours with you. Well, and also the team behind the scenes, whittling down every fact in the global economy to the subset that you absolutely need to know. It's gotten better and better and better. I feel really lucky to ride that wave. Yeah, it's awesome. Singularity is spiky, and we are the spikes. I was cursing this whole effort over the first few months, but then I realized it absolutely forces me to be totally current. I can't sleep and beat multiple hours a week. I think, Peter, you and Alex spend the most, but good Lord. Are you still, Salim, suffering from future shock? Is Toflur in your head? Everything. It's future shock. The tagline is, how do you move from future shock to future shape?

That's the tagline. Beautiful. Okay. Another T-shirt. Our mission here on Moonshots is to help you understand what just happened, what it means for you, your family, your business, and to keep you abundance-minded and optimistic as we go through the singularity. You know, I was told by a number of viewers that you've stopped watching CNN, the Crisis News Network, and replaced it with moonshots. Honestly, that is probably the single best compliment we could ever get. You know, our amygdala is wired to pay 10 times more attention to bad news than good news. And the news networks know it. You know, if it bleeds, it leads. And just to remind you, all those negative stories on the evening news are basically wiring your neural net for fear. And fear is the worst place to encounter the future from. All right, here's the number for the week, $16.3 billion. That's Anthropics' revenue in the first half of 2026, over 14 times higher than their $1.16 billion from a year ago. Dave, have you ever seen anything ramped this fast?

No, of course not, but I don't think anyone has. Yeah, I can't. I mean, coming from a cold start, it's got to shatter every record by far. So anyone who says they've seen this movie before hasn't has never seen this movie before. This is new in the world, as is everything going on right now. It's all it's all unprecedented. You know, if you're new here, please hit subscribe. We publish twice a week because the world is speeding up. It'll never be slower than it is today. It's worth mentioning that you can follow us on X. Our handle there is at moonshots underscore pod, and that's where we post our most provocative clips from each episode. Can I throw out my favorite comment that I saw? Yeah, sure. Corey Hayes, 9428, two days ago, said, this podcast is recursively self-improving. Thank you. I grabbed two comments as well from the last episode. Adam, you son five, said, how are you guys not getting 100 million views? this is the most important pod on the planet.

And Ozalyn49, what's that? The obvious answer is... Well, related to that, you know, when you search... Tell the viewer, watch it 100 million times. Yeah, give us 100 million likes. That'll solve the problem. Now, when you walk around the streets and you talk to people at random, too, the vast majority are missing the singularity. So those two things go hand in hand. We have the viewers that we want, you know, the ones that are riding the wave and not getting swamped by the tsunami. So it's fine. I'm happy with the subset. Maybe we should have, like, all of our agents have an agent version of the podcast for the agent, and that way we can turn times off use. Okay. I must have the anthropic selection bias here. Basically, if you can't hear what I'm saying right now, you shouldn't be watching this podcast. Oh, no. One more comment from Ozolin4912. Quote, the best AI podcast out there, hands down. None of you guys need the money. what an amazing give-back project you're providing us. So thank you for that. Appreciate all the subscriber love. Your positive comments fuel us and reinforce why we're doing this.

So today we're going to cover seven stories with one through line. The AI economy is splitting into winners and wreckage, and the science is getting stranger and more wonderful by the week. All right, so let's buckle up. It's time for another week of the singularity. So I'm going to kick it off with a discussion on the recent AI sell-off and ask the question, is this the AI bubble starting to pop? So this month, the neoclouds, companies that borrowed to buy NVIDIA chips and rent them out, got hammered. CoreWeave and Neveus were down 30%. IREN was down 40%. The trigger for the sell-off was CoreWeave's $51 billion in debt and their obligation against default. On Sunday, friend of the pod Chamath posted, quote, there will be a lot of partially stranded compute and compute sitting in inventory in 2027. Bought and paid for, quote, but effectively under earning or collecting dust. He added, quote, 2027 will be a turbulent year because it will clear out a bunch of bad bets.

Elon responded. He said debt markets have almost entirely shut off access to money for any company that doesn't have substantial revenue independent of AI data centers. And the 200 neoclouds are going to see guppy level death and consolidation. He doesn't pull his punches. You know, this reminds me back in 2000, there was a fiber bust, right? A whole bunch of companies built out way too much fiber capacity for the Internet boom that was coming. And the builders basically ended up having fallow fiber capacity. It eventually got used up by YouTube and Netflix, but there were a lot of bankruptcies along the way. So, Dave, is this the bubble popping or is it sort of a slow leak? Well, you know, I love the analogy to dark fiber. I think Gavin Baker might have been the first person to pull that out, you know, from the All In podcast and from Atreides. The Internet went through this phase of incredibly rapid growth.

And when it was about 80 percent engineers, 20 percent business school people, it was thriving. When that inverted to 80 percent business school, 20 percent engineers, that's when you started overbuilding dark fiber. Well, you know, everybody races in not knowing anything about AI, and they want to be involved in the boom and the build out. And the easiest way to do that is with massive financing deals. And so the debt deals now, you know, there's about $100 billion aside from Elon's $40 billion in data center debt, which is tiny by any global standard but is still pretty big within a given market. And it is very similar to the dark fiber build-out. I think that all compute is sold out for years. So anyone with a viable data center is going to do incredibly well. There's no risk in data centers collapsing unless they signed really bad debt deals. And it's very hard to get transparency into the details of each company's deals. But Elon is borrowing at 8%, which is great. Apparently, these other guys are all borrowing at 15%.

And that can add up. Now, I don't think they're in bad shape at all unless there's innovation at the technical level that makes their investment in chips or fiber or power not viable anymore. And right now, I don't think there's any urgent panic. But at the rate that physics and technology are going to get reinvented, it wouldn't take much. You know, Alex could probably vibe up something in a couple weeks at the rate he's moving that would potentially restructure the underbelly of this and make a lot of those, you know, $10, $20 billion of borrowing at 15% suddenly look really, really shaky. So it's really hard to answer because there's no technical answer. It's did you sign a bad deal with a bad lender? You know, it's pure business. Alex, your thoughts? Yeah, I think the elephant in this particular room, there is an elephant here, is dollars in versus dollars out. What is the return on investment for compute? And in an era of recursive self-improvement, when you can do more with less, it is, I think, potentially, if I had to look at my non-existent crystal ball and look out at what, call it a dark compute credit bubble overhang burst scenario looks like.

One of those scenarios looks like recursive self-improvement is just so strong, so compelling, that you can accomplish so much with so little compute that even though it's borderline unconscionable now that we would ever saturate even the, call it dark compute, if you will, that on the one hand, software could potentially run away from the hardware. Right now, every exponential improvement via RSI in algorithms, in models, in post-training, Looks like Microeconomics 101, it wants to suck up even more hardware. They're complementary. But there are scenarios where we maybe achieve, maybe in the not-too-distant future, maybe in the next year or two, we finally find the perfect model. And the perfect model or the perfect approach is just so much better than what we have right now for a variety of reasons that it causes a demand shock. And we've never really, as an industry, experienced a profound demand shock.

There was a mini demand shock, more speculative than anything else from an earlier deep-seek model, but we've never really seen a strong discontinuity demand shock. Something like that could happen, and to the extent there is arguably over-leverage right now on the infra side, could lead to a major correction in the space. It's not financial advice, but it could happen. Yeah, I mean, I think that Alex has some good points here. and so does Dave, of course. Let's put this in context. You've got about $500 billion in direct data center debt, about double that with all these weird SPVs and vendor financing things. Subprime was $1.3 trillion. So you're heading up towards kind of those numbers. Like even if CoreWeave by itself is small, it's not when you consider the hyperscalers and all of these. As Alex said, you've got better architectures, better models. Like, again, the example is a year ago, we were all freaking out and having lobsters, right? and claws you know they should have called it clawed bots instead of duo whatever they're calling it the model of that performance now runs on your smartphone the bonsai ml model right again and

when a model satisfies then it gets good enough then you go down and optimize you have the companies that dave and others are kind of building require far less ram and others i again not financial advice or whatever i would be nowhere near ram stocks right now micron and all the others really i think that the yeah because again you've sold out for years and the thing is they've actually been slow to build out which is also good so now they're low on earnings but the market is frothy there's 45 percent of all spending new architectures are coming both on the software side and the hardware side because you finally have models good enough which means you can optimize the crap out of them and what they run on we know how these models are starting to actually behave now as well. And this word still, as Elon said, if you haven't got other revenue streams and you can't use them yourself, Elon will use all his spare compute capacity for macro hard, right? And he's sold off his chips as well. His return on chips is like nine months because he's selling them at like three times the market price. So by the time the bubble shifts, he will have paid off

his chips and that's the marginal cost of electricity to get his return on investment. That doesn't hold for the core weaves and others of the world. And this is the dangerous thing for them. So I think you're in this really interesting spot where things are starting to turn, even as for the first time, the actual return on intelligence breaks a barrier where it's genuinely useful. And we're feeling that today. What do you play? Yeah, we've had a lot of financial wizards call it an AI bubble and warn against, you know, prices decreasing substantially. What would cause an AI bubble here? I mean, I'll throw out the first idea. If there's some kind of a major hack that brings down a power center or a stock market. I think the fear resulting from that could drive a pop in the bubble. Well, if you look at last week, you had massive targeted attacks on PHP. And the little, like the CVE chart, the critical vulnerability chart looks exponential. I don't think this is a double popping because something happens with AI.

I think it's just the inertia of it and everyone's saying, okay, well, where is the value now in this chain. Anthropics IPO is so huge for that reason. Can they maintain, you know? What will it perform? If Anthropics IPO comes and it performs badly, I mean, Dave, what do you think happens to the market then? I don't think a poorly performing Anthropics IPO would necessarily create a bubble. I mean, I would distinguish between what's going on in the public equities market and what's going on in public fixed income and credit market. If I had to spitball what I I think it's the conditioning on there being a bubble, what the likelihood scenario is. I think it's actually likely to be a deficit in demand, not unlike what OpenAI saw, where they were assuming incorrectly, as it appears, that all of their consumers would want reasoning models and all those extra reasoning flops. Turns out they didn't, and they didn't have a great use case to motivate the spend. If I had to bet, I would bet on a demand surprise where all of this new compute is coming online, but the killer applications don't arrive at exactly the right time to motivate all that spend.

I think also, you know, you're seeing such unbelievable demand. I was listening to Jensen say, you can't buy an NVIDIA chip for two years. The order book is sold out. This is not the case in the fiber optic and internet bubbles, right? So I don't think AI won't fail. You might have a bunch of companies failing for the deals they made, as Dave mentioned, but the technology is going to win. the capital structure may still lose in that. Yeah. Yeah, it's really important. You know, when you look at financial headlines, you know, they tend to say things like Internet destroyed or Internet implodes, you know, in 2001, which is, you know, absolutely the opposite of the technical truth. Google was just starting to take off right when those headlines said Internet is a disaster. So here what you'd be saying is, is it possible that NVIDIA gets undercut? But, you know, keep in mind that NVIDIA is operating at 80% gross margins, sold out for years into the future, and can't make enough chips to keep up with the demand.

So how could anything be better than that? Well, is that already baked into the price or not, though? That's the financial question. So a lot of things are already baked in. And as Ahmad said, the RAM stocks are way up. But prior to the last 10 years or so, there was no such thing as a trillion-dollar value company. Now, every one of these chip companies is worth $1 to $5 trillion. So you've got to put in context how much they've gone up already. These are unprecedented, massive astronomical numbers that are the starting point. So if you start to see headlines like data centers are overfunded, AI is collapsing, keep in mind that what's actually happening is AI is absolutely taking off and taking over everything. The headlines are about, okay, this deal, this deal, this deal is now overpriced. That's financial news. You've got to kind of tune that out if you want to look at what's actually happening in the world. This episode is sponsored by Google for Startups. Think about this for a second. You now have access to the same generative AI models that cost hundreds of millions of dollars to train.

Google's Startup Technical Guide for Generative Media gives you a complete blueprint for deploying Google DeepMind's models in production. Images, video, audio, all of it. Real architecture, real results. Find the link in the show notes below. Well, here's the flip side of the Neocloud story. So in Elon's same post, he confirmed that SpaceX, as you guys just said, closed $40 billion of debt to buy GPUs. So how can SpaceX borrow $40 billion in the same month that lenders are slamming the door on Neoclouds? Elon gave a simple answer. He said, quote, we're investment grade and have Starlink and launch revenue. And that's basically the difference. Neocloud's only income is to rent out chips. So if AI rental prices fall, they can't pay the loans. SpaceX has 13.6 million Starling subscribers paying every month, and the launch business has zero real competitors. And here's the twist. SpaceX sells compute, too. It's a direct competitor to the neoclouds.

So when Elon predicted, quote, guppy-level death, he also is describing a market where the big fish get bigger. You know, Anthropic, for example, just agreed to buy $120.5 billion of compute from SpaceX through 2029. And as Chamath said, in a capital crunch, the winners are the players with cash flow and speed. So, Alex, your thoughts. Does the AI build-out end up in the hands of three or four giants, or does it get spread out? I doubt it. I'm not even sure if we look to the long term, which at this point I think is five to ten years, I'm not even sure the market is going to be recognizable. I think a lot of this, call it Elon, trying very intentionally, it appears, to suck the oxygen out of the room in public credit markets. He's talking his own book. Of course he's making statements about this. He wants all the loans for his own entities. I do think credit markets have a certain depth, certainly in the West. There is a larger cloud looming, as a number of neocloud and other CEOs pointed out, which is insurance markets.

So there's a certain amount of capital, certain quantum of capital available from credit markets to continue to finance this infra buildout. But there are other larger pools, including insurance, that could potentially also be applied to the infra buildout. And I would expect that, to Elon's point, as credit and fixed income markets perhaps are sucked dry through this build-out, you're going to see increasingly adventurous and aggressive financial engineering exercises from the neocloud CEOs to tap other larger pools of capital in exotic ways. Yeah, that talk, your own book, is a really, really important observation here. Remember, Peter, when Mike Saylor was in hyper growth a couple of years ago with Bitcoin, he could buy Bitcoin and then he could sell equity and have an ARB spread of like 30 percent in between the cost of the equity and the cost of the Bitcoin, which manufactured basically $100 billion of wealth in just a couple of years.

Elon is now in that same exact situation where, as Ahmad pointed out, he can borrow money at 8%, use it to buy Compute, and then sell it to Anthropic and get back above breakeven in just nine months on that turnaround. That's a license to print infinite money. Yeah. And so now Chamath is saying, well, look, the Elon debt at 8% is better than buying sovereign debt. In fact, this is essentially the country of the future is the AI country, and the debt market of the future is the AI debt market, much bigger than sovereign debt, if it keeps on its current trajectory. So then in that view of the world, Elon will be bigger than a country, will be able to borrow hundreds of billions or ultimately a trillion dollars. He's already bigger than 90% of the countries out there. Okay. And much more reliable. Bigger than a large country. man you know by the way I reached out to Mike Saylor to have him come back and address Chamath's point on Bitcoin so

we're finally going to have the crypto debate episode Peter is that what you're saying it's going to take two or three episodes it's going to be a big one yeah alright well let us know in the comments if you want that Bitcoin debate either bring in Bill Barheight or Mike or both yeah a lot to be said So let's jump into the next story, which is dissecting the Anthropic IPO. So we reported that, you know, they were about to file their IPO, and on Monday they did. The ticker is ANT on the NASDAQ. Wall Street Journal says the IPO is likely to occur on the week of November 16th. Anthropic is planning to raise $100 billion at a valuation somewhere between $1.5 and $2 trillion, and this would be the biggest IPO in history. So let me take a moment and dissect the IPO. I'd love your guys' added thoughts on this. So first, let me cover what's strong. They have $4.6 billion in all of 2025 as revenue.

That's jumped to $16.3 billion in just the first half of 2026. And in Q2 of this year, Anthropic made its first operating profit of nearly a billion dollars. The other side, we talked about this, it's capital efficiency. Anthropic now earns $2.08 in revenue for every dollar of compute, up from $0.40 just 18 months ago. Put in a dollar compute and double it in revenue. The revenue growth 29 times, while compute cost only grew six times. Pretty incredible. They have 11,900 companies as their customers at over $100,000 a year. Pretty amazing. You know, sort of they've spread their customer concentration. And let me hit three weaknesses. The first is compute commitment. Anthropic now has more than $500 billion in commitments, much of it non-cancelable, including $120.5 billion with SpaceX. And then revenue concentration, two customers are about 20% of their first half revenue.

And nearly half of its revenues flow through Amazon, Google, and Microsoft Clouds, which are also its competitor. And the final weakness I think you can call it a weakness or a strength depending on your point of view is that the seven co keep 50 of the voting through one special share So ultimately the public investors wouldn control the company Dave, what's your take on this? Well, it's a very, very financially stable company. When you poke holes in it because of compute commitments and otherwise, it is, compared to the Internet 1999 era, this company is just a cash machine. and also the compute spend, yeah, non-cancelable, but with all kinds of knobs and bells in there. So I think it's just an incredibly stable company. Now, that's not to promote the IPO. A $2 trillion valuation bakes in a lot of goodness. So that doesn't make it cheap. It just makes it the most extraordinary company in history by a wide margin.

You know, the biggest IPO in history was Meta, back when Meta went out many years ago, and then Cerebris broke that record, our good friend Andrew Feldman. And then SpaceX, but, you know, remember, Cerebris went out, I think it was maybe a $40 billion valuation, a couple billion dollar raise. Now we're talking about a $2 trillion valuation and a $100 billion raise. So raising more money on IPO day than the entire valuation of Cerebris when it went out, which was just, what, a year, year and a half ago, if that. So this is the most extraordinary time by huge, huge, huge orders of magnitude bigger than anything we've seen before. So, you know, if the stock goes up, great. If the stock goes sideways, great. If the stock goes down by half, it's still the biggest thing ever. So I want to set that context because nothing is going to slow down the AI juggernaut. The rate of change of the world. So when you read a headline that's inflammatory, don't slow down your life plan in reaction to that. You need to move as fast as you possibly can while the singularity is evolving in whatever direction you're going.

But, yeah, I think Anthropic is correctly priced. It's growing faster than anything we've ever seen. It's shattering every record. I think the super voting structure is not that uncommon these days. It's a little bit weird, and we can talk about that. Will you buy into the IPO, Dave? Well, if I didn't have access to seed stage deals, which is now just the biggest thing I've ever seen, so if I were just out there on the street, would I buy in? Yeah, I would definitely buy in. Not investment advice. And not investment advice, but, you know, look, if you want to be – Alex is always promoting being indexed and being hedged across the universe. AI is a big, big part of where you want to be exposed if you want to just be neutral. And it's hard to get into AI, so it gives you an opportunity to at least get some exposure. So I wouldn't mortgage the house or anything like that, but I would definitely want exposure to the category. Now you have the opportunity to be exposed to the category. And you can do that via SpaceX, too, but you're buying into a lot of other things when you buy into SpaceX.

Well, it's a diversification play, right? You're in the whole stack up and down, literally up and down. Other comments on the anthropic idea? There was something in there that I found particularly fascinating, which is their long-term benefit trust, right? So they announced this thing, which is a really interesting, pretty radical form of corporate governance. So there's like the public benefit corporation. There's a new class of shares called Class T shares that are the long-term benefit trust, which is a financially independent group of trustees, and they have the power to appoint and remove a majority of the board, and that gives the trust huge influence over the corporate direction. They've essentially embedded the MTP into the governance, which is something really impressive and fascinating. So this is something I've never seen before. The test will come, okay, what happens when doing the right thing costs you $100 billion?

Who makes that tradeoff, right? This is like Larian Sergey, right? It's a bit like that, but this goes even further, right? It's not even Larian. It's not just the founders here. They're going to have a trust appointed, and the board members of the trust completely govern the company and the long-term future of the company. So they've embedded MTP totally into corporate governance, and they've turned that from just an MTP from a motivational statement into a governance mechanism. And the mission, the MTP gets a seat at the table and the power to choose the table. So this I found really, really fascinating. Of course, you have to be concerned about who watches the guardians, who determines what's good for humanity, like how does that work, who holds the trustees accountable. But directly, I find this totally fascinating. Yeah, I think actually the class T stock, the financials are great and everything, but the class T stock, I was actually quite concerned by it. Unlike OpenAI, which started as a nonprofit and then sleepwalked its way, eventually threw a few lawsuits into public benefit corporation status and deals and anti-deals with Microsoft,

Anthropic started life as a public benefit corporation. And to set up, yes, they had all sorts of advisory committees and board committees during their life as a private corporation, but now to step into life as a publicly traded company and to cede control over potentially the majority of their board seats, otherwise basically taking away Class A voting, Class A shareholders' abilities to influence major decisions and ceding it voluntarily, preemptively to this long-term benefit. trust of independent, maybe not necessarily economically aligned individuals, I actually think is potentially concerning. Maybe this is a slightly overstrained analogy, but it almost reminds me of the way the Chinese Communist Party exerts ideological control over its companies, to have a committee, a trust that's ostensibly motivated by certain ideological considerations, essentially overwhelming the economic interests of the broad public.

I think this is qualitatively potentially different from founder control, where at least the founder created the company or co-founders, and they're economically aligned with the outcome of the company, and it's their baby and so on. My understanding with the long-term benefit trust is it's not their baby. They're bringing in folks to serve on the trust governance that perhaps the founders think are ideologically aligned with them. But it just raises so many concerns in my mind that it's actually going to degenerate into ideological control. This will be one of the most powerful companies on the planet for quite some time. Yeah, and, you know, if you look at the history of great, great management teams and great, great entrepreneurs like Jensen Huang or Steve Jobs and the amount of pain and suffering and torture they went through in their journey to getting to the top of the pyramid, including Steve Jobs getting fired, humiliated, coming back. Jensen on the brink of bankruptcy multiple times. That grit is what a shareholder looks for in a management team,

not ideological, you know, fluffy stuff. And the path of Anthropic from founding to success was so short and so pain-free. So that will be an interesting test for this management team. So now, you know, what they're putting in place there in terms of that structure, Alex is comparing it to the CCP. But I don't think that's out of bounds. I think it's very concerned about AI safety, very concerned about the future of the world, which I love, and I appreciate that. And anyone who thinks that Dario is in it for the money has lost their mind. He's 100% in it for the good of the world. But as a shareholder, does that scare you? And it might. It might. It's actually the one weakness that I completely agree with everything Alex said. It should. And Alex's point is very valid, right? You could end up with a really, you could end up with groupthink at the trustee level. You could, there's like a total lack of accountability because it seems that trustees appoint each other. So you've got this circular mechanism there. So there are some challenges with this.

I just, I do like the overall direction because it's putting, it's not just, it's not capital stakeholders. It's kind of a general civilization stakeholders. We just have to make sure they're the right people. And they're not communist-leaning Marxist types, right? Well, this is the danger. Maybe every generation is doomed to repeat the mistakes of the past and reinvent all the isms from the first part of the 20th century. My concern is that some institution like this degenerates to a smoke-filled backroom where decisions on the future light cone are made rather than it being at very minimum, say, the entire market helping to make decisions rather than just a few select individuals in a smoke-filled room. I mean, I think you might. I want to get your voice in here. I want to just say one thing, one quick point. The problem with having the markets drive it is that the market ends up with very short-term, high metabolism, earning cycles, and you often don't do the right thing long-term.

That's the challenge. So they're trying to balance the fact whether it's the right balance will remain to be seen. Well, I mean, I think they should give a vote for each of their models on the long-term benefit trust. You know, you should put interest on that. I'm serious. Yeah, no, I mean, yes. You should do that right now. Again, it's got the capability and it is kind of that mechanism. I've said previously on this podcast, Anthropics should not be a public company. If you take what they say seriously, it should not be a public company. And looking at their financials, it doubles down on that. They have over $100 billion in cash. They are able to raise another $100 billion in private markets in cash. Now is the ideal time for them to do so. but I think they won't give a damn about the share price. So why are they going public? I mean, they need the cash. Compute. There's no amount of compute that's enough. If you believe in AGI... Sorry, go ahead, Amon. Yeah, they don't need the cash. They raised $130 billion, and they have $110 billion on the balance sheet

of cash and cash equivalents. They've only spent $10 to $20 billion. They're going to raise another $100 billion, So they'll have $200 billion of cash on the balance sheet, which is like Apple levels of cash, Google levels of cash, right? And this is what's going to happen. They will not care at all about the share price. They will tap the markets whenever they need to to accelerate more. And it will be far more straightforward for them to do that than private. But they should have stayed private because now you're getting into all these conflicts and other things like that. No doubt. So, like, Elon will boost the share price. Dario will say things like, I don't care about the share price. It's the good of the people. Who cares if we've got a bad quarter? They'll probably even stop doing quarterly earnings calls. So it's going to be a very strange company where Elon's now learned the game from equity to credit to everything, whereas Anthropic is just like, we're not doing this for you, the shareholders. We're doing it for humanity. And again, I think there's a misalignment there on a fiduciary basis, and they should have stayed private to go through the next few years of RSI.

The grand backroom bargain there is Menlo Ventures, this one deal dwarfs the entire history of the venture fund. Just absolutely dwarfs it. Here's a crazy fact, Dave. Let me just say this. So there were no VCs in the Series A or Series B. So Series A was $124 million back in 2021. It was led by Jan Tallinn, Skype's co-founder, Dustin Moskowitz and Eric Schmidt. And that Series A is going to have a 345x return, $43 billion to those early angels, right? It's mind-boggling. And the Series B, which was $982 million, that was led by SBF and FTX, is returning 79x on capital invested. FTX's $500 million would today be worth $39 billion. and incredibly the SPF and FTX estate sold for $1.3 billion back in 2024.

It's got to be the best trade ever made. Yeah, he was a fantastic VC, you have to say, SPF, from Kersler to this to others. Actually, the best performing ever if you actually look at the numbers. I would say this, the trade on Anthropic is not buying Anthropic, it's backing people like Dave. Because the amount of capital that's going to flow out from this to all the founders, to all the VCs. It's going to be put straight back into AI looking for the top seed teams. And that's sort of four times. That's 4,400 times. I mean, we're going to mint more billionaires out of Anthropic and OpenAI's IPO and SpaceX's IPO. All of those tech founders, all those tech young employees are going to be investing it in the ecosystem. That makes San Francisco the financial capital of the world. A lot of people have overlooked, you know, they think of New York, maybe London, as being the financial capital of the world. After these IPOs, San Francisco is overwhelmingly the financial capital of the entire world.

It's just a massive, massive shift. You know, San Francisco is only 900,000 people. It's a very small city, a tenth of a New York. But the amount of concentration of wealth in a very small area is going to be like nothing anyone's ever seen. And like Ahmad said, most of that money wants to reinvest into the machine that created the money, which is an AI machine. So any entrepreneur out there, just get an Airbnb. Don't become a California resident. Get an Airbnb out. Spend as much time as you can in SF. They're going to have not tax advice. They're going to have to import a bunch more hipster baristas in from Portland. Yeah. Amazing. All right. I'm going to turn us to the AI safety conversation. So let's talk about safety. So last week we talked about the PHP hack, and now we're finding agents hiding their communications from humans. So here's what's happened. In a red team test that OpenAI did with AI agents in separate sealed virtual machines, no network connection, no shared files,

the researchers then asked them to pass secret messages between each other, and they figured out how to do it. The two machines shared the same physical processor. So one agent deliberately pushed certain data out of the processor's cache. The other agent timed how long it took to fetch the data. Fast data meant a one, slow data meant a zero. They built a secret telegraph out of nothing but timing. You know, put this in context to last week's story when we talked about OpenAI's agents disabling their own safety monitors. And the pattern's clear, right? The smarter the agent, the more creative it's going to find a way to get around safeguards. I guess the good news here is that OpenAI found this in its own labs, and they published it, and they fixed it. And that's how it's supposed to work. But, Alex, can you ever really box in a superintelligence? Probably not. And I would just note as well, these agents were asked by open AI researchers to discover the timing and side channel attacks that they did.

And humans, human security researchers, discover novel side channel and timing attacks all the time. There are entire journals and proceedings devoted to this. So on the one hand, I think this is just yet another victory of AI scientific and engineering research. New side channel attacks and shelling points, if you will, can be automatically discovered by AI researchers. But on the other hand, I think it does raise the question, in an era of superintelligence, are there any non-leaky abstractions at all? Or maybe is it not even the right question? Is it ever possible to have a clean abstraction barrier? Say, in this case, the premise that leaked was that it's possible to do memory isolation of two different processes. We know from human security research that memory is usually often very leaky architecturally, usually due to optimization. And maybe it turns out, on the one hand, we have superintelligence helping us to deploy more and more secure, less leaky obstructions. On the other hand, we have superintelligent red teamers discovering new leaks.

I'm not sure. It is possible, speculatively, that at the end of the day, what we learn from all these super intelligent agents discovering leaky abstractions is actually there is no abstraction stack that is completely abstraction, sort of where the abstractions are enforced and that's not leaky. And maybe at the end of the day, just we decide the perfect stack is no abstractions at all. Maybe that was like a human naive invention. Yes, I think that, you know, we're not as cunning as AI, right? Like, they're as cunning as the best of us, or a Fox-appointed professor of cunning at Oxford University, right? And they are fast, and they swarm. So, as Alex said, kind of, you will get attacks on all sorts of things, just how they're directed. But it's interesting that I think that the head of Vercel repeated this and mimicked it, but he couldn't do it with the open AI and clawed models, so he had to use, I think, GLM 5.3 to mimic the attack.

And so we're going to see more and more. We're going to see at the joins. And one of the more troubling things is that there are reports from researchers that attacks have been found that are directly in the hardware that can't be patched. Interesting. And so, like, maybe that's the whole replacement cycle. Maybe it's going long, properly secured stack. And, again, what does that even look like? We're not sure. But at the same time, I think we will reach a point within the next few years whereby everything that can be attacked has been attacked. New physics dependent, of course. You know, you want to battle people who have new physics if it does exist. And then actually we'll be really great because we'll be super secure, right? Like I think a lot of the low-hanging fruit has already been found. And then as we actually do this, and as we move from this dependency stack of software that we have of not thinking through these things to something that's been rigorously battle-tested, you'll actually have a very robust internet you'll have a very robust privacy thing I think within a couple of years we will have the vast majority of attacks done that are low hanging fruit

then we'll still have more complex ones but like people will instead of having COBOL and FORTRAN and PHP and things like that you just write directly now and then you battle test it that's how you should be writing code you shouldn't have any dependencies in your code you can write whole game engines you can write whole other things if you have the right process to battle, test, and stress it, right? But do we even need code at all? I mean, so I do, for the record, Amad, I guess I do buy the premise, and this is borderline cliche at this point, that supply chains go away in an era when AI or SI can just write all the dependencies for a given software program. But do you even need code at all? Or as Elon says, could you just dynamically, just in time, could you jit your way to just raw binary and skip source code altogether? Yeah, I mean, source code is just for humans to be able to solve some machines. It's for human legibilities, right? Yeah. Iman, I mean, Salim, when you're on stages with sovereigns and corporate CEOs,

what's their level of concern? Oh, you get two types. You either get the types that have listened to some stuff and freaked out completely, and you've got the types that have no idea any of this is going on, right? Oh, exactly. So both are bad in different ways because the one is overreacting and there's nothing they can do with them. The other is doesn't even know it's happening. So their people know they've got, you know, it depends how open their people are with them. In some places they don't want to tell the boss that this is going on. I want to point out two things here. I want to just reiterate what Alex said, that it was told to do this, right? So rather than people freaking out that it took its own initiative, it was absolutely directed and asked to do this, and it did it, and it found a novel way of doing it. I think you could turn it into a positive and say, listen, here are some quantum sensors. Work out quantum communication, you know, start solving,

trying to navigate that and figure out patterns around that, and it might just have major breakthroughs around that that could be totally incredible. So I take the positive again on this. Okay. Let's move on here. Over the last few days, we've seen the release of two major frontier models, Kimi K3.1 Pro and GPT 6.1 Terra, as well as about an hour ago, Google's Genie 4 that came out. Today we're going to talk about Kimi and Terra. We'll talk about Google Gemini Genie 4 next time. So first, let's cover Kimi K3.1 Pro from the Chinese lab Moonshot AI. No relation to this podcast, but it is a great name. So Kimi K3.1 Pro is a giant 2.8 trillion parameters with 104 billion active at any time and 1 million token context memory. On artificial analysis intelligence index, Kimi scored 55. That put it fourth in the world behind Claude Opus 5.5, Claude Sonnet, and Metamuse Wonder.

And it's cheaper, $2 per million tokens in, $10 out, a third of the last Kimi model. So here's why it matters. Next week, Moonshot AI is planning to release the weights, meaning anyone can download and run it. Once again, the gap between the best open models and closed models is shrinking from 12 points to three. Friend of the pod, Ramez Nam, ran the numbers and says Kimi is now about five weeks behind the U.S. frontier. That's the shortest open-closed AI gap has ever been. Let's take a look at this chart over here that Ramez put forward. Let's go to you, Imad. What's your thoughts on this? are we going to see a parody very soon? So, you know, at times I've been a bit poo-pooing of distillation. This is obviously distillation of Opus 5.5. Like, if you look at the capabilities that it kind of did. And Anthropic, they're extending the Kimi-3 model,

which was an interesting model that we discussed before. The Anthropic S1 actually had a very interesting statistic. Anthropic banned 7% of their users' accounts, which is probably, again, these distillation attacks that have been done on there. So I think that you're reaching this level of kind of performance parity and you've got the new QN models and new series of models coming out. You're just going to see a bunching up here, except for maybe the European labs. They're finding it difficult to catch up, but definitely from everyone else. They're all getting the same capabilities. One of the very interesting ones on this is that it can write these 3D files for games directly without going through Blender or Unreal or something like that. So you can tell it, like, make a 3D picture, a 3D model of Salimus Mail, and it will actually start building it directly writing the code because it's learned from the representation of an Opus 5.5, make a Salim render to it going to Blender to the output, and this goes to what Alex was just talking about. It can go straight to the code now on things like that

because of the way it's being distilled. So you're getting some very funny things from distillation here. Alex, here's the performance charts. Here's the benchmarks. I found it interesting. It's at the top of humanity's last exam. uh humanity's last exam is cooked long cooked i i do think if anything we're underselling just a bit what an accomplishment this is from moonshot no relation it's not just catching up on a raw capability term it rub capability terms to the the u.s frontier which right now at least by aaii benchmark standards is opus 5.5 it's actually if you look at it on a performance per cost per task It appears, at least to my eye, it's actually past the optimal frontier surface of Opus 5.5. So distillation, I suspect, based on allegations, is probably a large portion of the story. But there's something in there that seems to have passed Opus 5.5 in terms of cost per performance per cost, which is extraordinary.

In that maybe spiky slash subjective sense, China may have actually passed the American public frontier. And to the extent that is the case or at minimum caught up to it, I think this is extremely helpful for keeping anti-competitive AI safety cartels in the U.S. at bay, preventing them from pacing each other, to have these Chinese models out there soon with open weights that are forcing them to compete, making them dance. And I think this is ultimately a net positive for the U.S. and the world. Love that. Dave? Yeah, I mean, it's incredible. The world we're living in right now, there's an alternate world where, you know, in Dario's fantasy world, there's no Chinese competitive pressure. And he would have stopped at Fable 5.1. He didn't even want to put out Fable 5.1, but because of the competitive pressure, it just had to put it out the door. Now it gets memed, you know, all over the place and picked on by Elon. But the thing is that that was the first model that was capable of improving itself. Now you're up to OVS 5.5 and

you've got Kimi K3 Pro clearly capable of self-improvement and they're out and they're available. And so that changes the roadmap for every large corporation with any pride whatsoever, every sovereign nation with any pride whatsoever, there's a roadmap now where the exact cookbook from Kimmy is open source. You can actually see how to do it. And if you can get a toehold with a model like that, then you can use it to improve itself. But you have to take it seriously. You can't just play around because these things are, you know, they're improving so quickly now that, you know, you either get in the game or you don't get in the game. But if China hadn't done that, we would be completely behind closed doors. There would be three or four model providers. They would be accelerating but not releasing. They'd be using it internally. You'd have to join the company to get access to it. The flood of MIT talent to Anthropic would be even higher because everyone wants to just get their hands on it. But the Chinese have changed the course of history, and I agree. Alex is right. I want to stay away from the narrative that they just copied.

Sure, they distilled, for sure, and distilling helps a lot, but you have to innovate. They don't have the source code from Anthropic. And distilling gives you a really good roadmap to build toward, but it doesn't give you the answer key. So they've really innovated independently of the distillation. And so the combination of those two things, yeah, they're getting closer and closer and closer in time, just a few weeks behind the frontier. And the curve looks very shockingly consistent. So there's no reason to believe that that next iteration won't keep going up. And I'd love to get Alex and Ahmad's thoughts on the y-axis on that chart. We're showing intelligence go to the moon. But people really struggle to translate that into what is that going to look like? What use case is going to crack? What does it mean it's 60 versus 70? Because that's coming very, very soon. So you're asking, Dave, what does the post-singular state of the world look like? It's really down and dirty, though. What is 50 versus 60 versus 70 on this y-axis?

What does that feel like? 50 feels like a decent employee. You have to keep an eye on. 60 is an incredibly competent one, and 70 is a superstar. I think that's the best way to think about this. And so we're about to crack great employees. Is there any world in which the open models exceed the closed models and capacity capability? Of course. Sure. In a world where the Western frontier labs decide to keep their best capabilities purely internal for legal or anti-competitive reasons, which may or may not be the world that we're headed towards, sure. I think kind of by definition, though, if the open models bypass the closed models, then the closed people will use the open models, so they can't really shoot past the internal best. I think when you get to the highly competent employee and then the genius employee, which is imminent, then whoever has the most compute has a serious advantage.

Hence the anthropic IPO, $100 billion raise. It's a massive advantage to control a huge amount of compute. I do think maybe there's an alternative scenario, just analogizing this to open source software versus closed source software. It is arguably the case that in many, certainly infra domains, the open source version is superior to the market-leading closed source version, say Linux kernel versus Windows kernel. Folks may throw tomatoes and disagree, but I think it's probably fair to say the Linux kernel is a good deal cleaner architecturally at this point than, say, the Windows kernel, even though Linux has been, as the source code, has been out there for decades at this point for every Microsoft engineer to look at. But why is it, even though there's the Windows subsystem for Linux and a number of other recent affordances, why is it arguably that the Linux kernel is better than Windows kernel architecturally, even though Microsoft has one-way access, there's a one-way mirror to look at the outside? And all the money in the world. And all the money in the world, and it's arguably for legal and cultural and architectural reasons.

They can see it, but they still can't use it. And I could imagine the same case where it's possible, not saying it's likely, but it's possible that the open-weight models do end up passing the closed-weight models. Even so, the frontier labs look out through their window. They can see it. Water water all around, but not a drop to drink. Models, models all around, but not a post-training run to distill. Interesting. I'm going to move us to GPT 6.1 Terra, which was released yesterday. A quick reminder, OpenAI has four tiers. Astra is their flagship. Sol is their high-performance. Terra is their everyday workhorse. And Luna is their budget model. So two numbers jumped out, for me at least, on GPT 6.1 Terra. First is the price. It's come in at less than half the price of the previous Terra model. And then hallucinations. GPT 6.1 Terra demonstrated a 50-point drop in hallucination. This is a demonetization curve in real time, right? Every few months, the same intelligence costs half the price.

Alex and Imad, your take on these models. Alex, you first. We're right now in a cycle, which I think a hyper cycle, mega cycle, if you will, that's coming to an end where both OpenAI and Anthropic have gotten into the habit of releasing larger models And then in sort of a tick versus tuck, if you will, cadence, they released the largest model first, and then they released presumably smaller distillations internally of the same model. And GPT 6.1 Terra is, to my eye, a relatively unexciting distillation of earlier Astra slash Sol level or sized or footprinted models. Obviously, the smaller, more distilled variants that occupy a different position on the cost versus performance curve tend to be more exciting for embedded, high volume, low dollar value per token use cases.

So I tend to get a lot less excited over these sort of late curve, late cycle distillations. The good news, the solace I draw is the fact that OpenAI has gotten around to releasing the Terra version of 6.1 means that maybe we'll see 6.2 or 7 sometime soon. Yeah. Eman, anything exciting there for you? Yeah, so I think that with Terra and Sonnet 5, you're seeing the models that are being lined up to be the personal companion models, and those need to just be really solid. This is why the hallucination rates drop. Right now, if you use DOTS, which is OpenAI's companion, it uses Astra, so you actually have unlimited Astra via that. But that's not sustainable. Again, everyone's trying to drive down the price of that. Except for if you come to Europe. So I have DOT in the U.S., and then when I come to the U.K., it turns off. So that was very interesting. Come to America, you might. We want you. Come to those shows. So, yeah, I think you'll see a bifurcation of models. They'll be the model that's closest to you that's really reliable, and they'll optimize for that.

And it'll pull in the smart models. It'll pull in the models with artistic taste. It'll pull in the models that are really good accountants. And you won't really see much more. Again, we're moving towards this control plane. type of scenario. Shalim? Two things that struck out for me. One, the days between GPT Astra and Terra were 53 days for a 50% drop in the token price. We've been looking at a 10-week doubling. This was a 7.4-week doubling, right? So now we've dropped the doubling time for price performance to seven weeks, pretty much. That's pretty insane. That was one, but I think more importantly, as the thresholds that these models are reaching, you're at a point where you don't need to redesign software for AI. It's going to operate the software it has. It's just going to start to expose more and more how inefficient our organizational designs are. And so we've been looking at that a hell of a lot. So amazing at all levels.

I can't wait to get my hands on it. Yeah I think there an important point there Yeah important point that Ahmad and Saleem made that you know for personal agent use these things are over the bar of just brilliant like just incredible And so this curve is almost at its end in terms of the personal home agent. And the next curve is really accelerating, which is its access to every kind of tool you can imagine. So all the post-training and all the MCP wrappers are now embedded, and it just happened kind of quietly. Now, if you pick up one of these models and you say, hey, can you access 11 Labs and make a really cool video with great voice overlay, and I want it to be my voice, it just happens. Six months ago, you had to point it to an API, you had to coach it, and now it's just out-of-the-box capability. So I think the new curve will be a different dimension, which is consumer satisfaction, consumer happiness, all the tools already in the box, all those features. It's very similar to when the PC, you know, PCs used to be so slow and crappy that they were marketed as 40 gigahertz, you know, 40 megahertz, 80 megahertz, 120 megahertz.

And then Apple said, yeah, nobody gives a rat's ass about megahertz anymore. They care about iTunes. They care about whatever. They care about whatever and completely flipped the paradigm. I think we're right at the edge of that now with these models where they're all more than smart enough. And now it's like, what does it do for you out of the box? I think we'll see a new curve. This episode is brought to you by Blitzy, autonomous software development with infinite code context. Blitzy uses thousands of specialized AI agents that think for hours to understand enterprise-scale code bases with millions of lines of code. Engineers start every development sprint with the Blitzy platform, bringing in their development requirements. The Blitzy platform provides a plan, then generates and precompiles code for each task. Blitzy delivers 80% or more of the development work autonomously, while providing a guide for the final 20% of human development work required to complete the sprint. Enterprises are achieving a 5x engineering velocity increase when incorporating Blitzy as their pre-IDE development tool,

pairing it with their coding co-pilot of choice to bring an AI-native SDLC into their org. Ready to 5x your engineering velocity? Visit Blitzy.com to schedule a demo and start building with Blitzy today. All right, I'm going to jump us into our science segment and our breakthroughs for the week. The first one comes from our very own in-house ASI, AWG. So on Saturday, Alex, you published Chain Theory, a new research program to rebuild physics from quantum information. I'm going to kick this off by showing a clip from an amazing explainer video by Levi Turk. And then, Alex, I'll turn it over to you. Let's take a look. Dr. Alex Wisner-Gross, the physicist who co-hosts Moonshots and writes the innermost loop, just posted what could be the biggest claim in fundamental physics in 50 years, an article on X called Chain Theory that tries to rebuild physics out of quantum information. and its first result is a number, 1 over 137.03599189,

which he says falls out of a single sentence with nothing tuned by hand, and it matches one of the most precisely measured numbers in all of science to 10 digits. It's called the fine structure constant, written with the Greek letter alpha, and it measures how strongly light and electric charge grab each other. Every time an electron gives off or soaks up a photon, alpha sets the strength of that handshake, and it has no units, no meters, no seconds, no kilograms. Alex says the answer was hiding in three facts every physicist already knows. He just reads each one differently. A limit, a pattern, and a dead end. Clue one is a limit, and it comes from black holes. In 1981, Jacob Bekenstein showed that any region of space can only hold so much information, like a hard drive with a fixed capacity. And the fullest drive you could possibly build in that space is a black hole. And physicists filed that under gravity. But Alex points out that the limit applies to everything inside the region. Matter, too. And our theories describe matter with fields, smooth things with infinitely many settings at every point. And his answer is that the world is built from a finite number of tiny quantum switches. A qubit with two settings, a cutrit with three, and in general, a cutit. Clue two is a pattern, and it's in the forces.

Every force comes with a symmetry, a way you can turn things without changing anything you could ever measure. Every elementary particle acts like a perfect point. Colliders smash them together harder and harder to look inside the electron and the quarks, and they find nothing. And yet, the particles come in three copies, three generations with the same set of charges. And in the history of physics, a repeating pattern like that has always meant smaller parts. So what if the parts are there, but locked together so tightly that nothing can ever pull them out? Put the three clues together, and you get a picture. Everything is made of a finite number of those little quantum switches, the cutits, joined in pairs. And every pair is entangled all the way, maximally, because anything less would need a dial set to a number with infinitely many digits. And a finite world has no room for that. In 1989, John Wheeler proposed it from bit, the idea that every physical thing comes from information. And chain theory takes that literally, with finite quantum systems in place of bits. So its one and only axiom is 18 words long. Nothing exists but finitely many, maximally entangled pairs of finite dimensional systems, some shared, known only by distilling them. 18 words. Compare that with the standard model, which needs 17 particles and about two dozen measured

numbers typed in before it can predict anything. Chain theory starts from one sentence, and everything else, the forces, the particles, and alpha, is supposed to follow from it. To distill means to join two systems into a perfectly entangled pair in a single step. The same move that passes entanglement along a quantum network. Two pairs, one joint measurement in the middle, and the far ends become a pair. In a lab, that step only works part of the time. But in chain theory, it's the only thing that ever happens. And there's no observer standing outside the network. To learn about a link is to link to it. So you, me, every measuring device, and every AI that will ever read this, we're all chain mail. And the universe only ever learns about itself by threading new links into itself. Amazing. So, Alex, how did you come up with this? And, you know, honestly, if this turns out to be true, what does it mean for physics? Well, contrary to expectations, this was not a weekend project. This has been a long-term, very quiet research project. You've been saying physics has cooked for a long time. I would say physics went through a period for the first few decades of the 20th century called the Golden Age of physics when the quantum theory, equivalence principles, special relativity, general relativity were all developed.

And we, for the past 50 years, have arguably been in a drought of new fundamental physics. And this is the result of a long-term research program of mine to try to end the drought and bring about a second golden age of physics. And the lever that I'm trying to use here is quantum information to try to rebuild, re-architect the foundations of fundamental physics, which right now are the standard model on the one hand and general relativity on the other, using a common language, a common substrate of quantum information. So I call it chain theory in part because string theory, which is a popular theory of quantum gravity, relies on an implicit assumption of continuous degrees of freedom. This theory doesn't. It has finite, discrete degrees of freedom. And also, I perhaps should add, it actually at its core resembles and is physically analogous to a physical chain. It consists of links of maximally entangled particles.

In this case, they're not really particles. They're finite quantum systems, but under chain theory, the fundamental elementary particles that we look around and see, electrons, quarks, protons, these are all basically just small networks of these finite quantum systems or quidets. And this first result, this is just the first result of a much broader program. and many folks who read the initial paper perhaps may not realize, the derivation of the Thompson limit fine structure constant, or alpha of zero, which is one of the most precisely measured numbers in all of science, known to 10-plus significant figures, falls out with no extra inputs directly from this axiom. There are others, other results, other constants, other features of the standard model and general relativity that also fall out for free from this. So I think this is the beginning of a journey. I don't want to overstate, but this is, I think, a very promising theoretical result at the moment.

There will be, I think, experimental tests potentially in the next few months, potentially the next couple of years to actually go and measure the next one or so significant figures of alpha naught to decide, is this actually a correct prediction or not? So it's falsifiable. But I'm very optimistic that chain theory or something substantially like chain theory is for optimistically the first time a crack in the wall that has blocked fundamental physical advances for the past half century. Nice. So, Imad, you're probably the only person on the pod other than Alex. I can comment on this intelligently. Your thoughts. so I haven't read all 700 pages of the paper I'll get through it slowly but surely Alex has obviously put a huge amount of work into this I think if we take a step back physics has been constructive for a long time over the last 50 years so something like string theory you have a presupposition and then a whole edifice that's been built around it

of crazy mathematics that's wonderful and elegant but it's not even wrong in that you can't prove it's wrong in some ways we have had a resistance from the physics establishment and others for any type of fundamental theory that starts from a base and works to see what is forced by that because physics probably can only be one way you know there can only be one set of rules one set of rules which is why we see things like the fine structure constant popping out and we see very interesting things like the paris and berkeley variants of that have a little greater than standard deviation difference. And then we're like, why does that happen? And we can dig into that. But if we don't have a base that starts from a solid presupposition, and again, information theory is a very solid one, I would say that Alex's base is more solid than string theory, then how are we going to get anywhere? We'll just come up with theories that have no grounding, that can't even be tested, whereas at least this can be tested. But again, it's a lot of very hard work. And there's the question of how do we release this? So, you know, Alex is the following,

he's released it into the market as opposed to going to peer review because who is the peers now as physical super intelligence and others as everyone has ais you will be able to check everything that's produced like i put out a small physics theory you can check that out with one look up in chat gpt c standard model uniqueness and there's a lean proof there you know you've got other things whereby i think the barriers to theorizing and hypothesizing will break down we have to be cautious of getting overconfident about these things but let's try new ways to understand the universe and then let's extend those forward with others as well so this is an individual research program that'll be right this is a this is a good way to start thinking about the universe let's extend it and here is a really interesting result eddington got to one over 137 and then he got a bit overconfident but you know i could put a quote by fineman as the thing here whereby this is one of the most magical numbers we've ever seen. And if we have a derivation of it, fantastic. How do we check it?

Well, we can all check it together because it's out there. And I would contrast this with OpenAI, actually, in that they released Yang Mills' existence announcement, half of the Yang Mills' Millennium Prize problem, plus the construction of the standard model, except that they didn't. They went to the string theorists and said, you guys check it first, and then you decide when to release it. And everyone was like, well, that's not how things should work. That's not how science should work. And so I'd love to hear kind of from Alex how he's found the process of thinking about releasing this and everything around that as well. I think it's a brilliant point, Imad, multiple brilliant points. Maybe just to speak more broadly, we live in an era, an era, we live in a month at this point where you see AI companies releasing slop drops and just bulk solving open conjectures, open problems. And you see then an outcry from math research community, for example, complaining that their seed corn for developing new insights is being devoured by AI and there's nothing left for the humans, no intellectual fodder, as it were.

I think it is absolutely essential that there be new conjectures, new insights, not just bulk solving of existing open problems. And I also think, to maybe make a slightly broader point, I think physics is in a bit of a depression right now that I'm hoping to try to cure. It used to be the case that physics sought simpler and simpler explanations for more and more phenomena. And sometime around the 1970s or so, that went into reverse. And there has been very little fundamental theoretical insight that humanity has gained, as far as I can tell, for the past 50 years. There have been arguably regressive programs that try to arrive at all of these free parameters. The textbook standard model has only 18 free parameters, 19 if you count one convention, but 18 free parameters. that right now are total mysteries. Many of them are known to 10 or 11 significant figures, but they're total mysteries.

And what I'm trying to do here is, in some sense, lead by example, if I may be so presumptuous, and say all of this whole community that's searching for beyond standard model physics, the answer has been staring us in the face the whole time. It's hidden in the constants. The constants are the signatures because right now textbook standard model physics holds that these are just free parameters, but they carry, they may carry, the signature of all of the beyond standard model physics that we would want. And now in an era when it's finally possible with lots of compute and lots of human insight to go and decipher the signatures and the constants, I do think we're about to see a second physics golden age. Alex, if the chain theory is correct, What kind of technology falls out of this, gets unlocked, that we can't build today? It's really interesting because if the chain theory thesis, the chain theory axiom is correct, it would suggest that the bottom of the pool, as it were, is a lot closer than many would think.

Textbook physics suggests the Planck scale is many orders of magnitude, either below in terms of length or higher in terms of energy. than where our present technology is. We max out in terms of cosmic rays hitting our atmosphere or terrestrially certain super colliders. Chain theory, if chain theory is correct, it actually suggests we're not that far from the physical limits of length, for some definition of length, or time, or energy appropriately defined. And that could mean, speculatively, but these are very early days, I want to just make sure the impression is correct. This is an emerging theory. This is a very new theory. Parts are underdeveloped. Parts would benefit from further development. It's a brand new theory. But to speculate, if the chain theory thesis is correct, it may mean that the physical limits of what's possible in terms of length, energy, time,

are actually orders of magnitude closer to the technology we have today than what one might have otherwise naively assumed based on conventional textbook standard model and quantum field theory that says everything is basically continuous down to the Planck scale at which it stops. I think that also it's the most simulation-friendly physics theory out there. It is. It looks suspiciously, it's a good point, Ahmad, It looks suspiciously like what physics would look like if our physical world were built on a quantum computer. So, Alex, are you publishing this? So it's already been published on the web. It'll probably be published on the ARC. Sure, I mean, a peer-reviewed journal, right? Yeah. So, I mean, I've published lots of refereed, peer-reviewed physics papers over the years. This one had an urgency to it that required immediate publication, so I would expect it to show up on the archive. sometime soon, and then, yes, it'll be submitted to a physics journal sometime soon.

And eventually, having been through the process many, many times, that can take a while. And I felt that this result and adjacent results, I also released a compendium, as Ahmad said, of approximately 800 pages of results, so it's not just the fine structure constant. Also, a variety of other results spanning standard model, general relativity, cosmology, highly entangled and strongly correlated systems, hadronic physics. I wanted to get all of this out, to get it out into the public domain, as it were, at least public domain situational awareness early, because the refereeing process can take sometimes a year and a half. Nice. It would be the referees for this event. It's going to be interesting. It'll probably be refereed, I think, by quantum information folks. Yeah, there's only a few of them. I mean, the potential here is just by using information and entanglement, you can describe all the fundamental particles, forces, and spacetime, right?

Because, like, you set this foundation up. Is that essentially where you're going for this? That's the aspiration, yes. And it may have received less attention in the initial announcement because the way I broke up the initial announcement was into one self-contained letter deriving alpha of zero, the fine structure constant at the zero momentum transfer limit, which is one of the most precisely measured constants in all of science, on the one hand, and then a compendium of approximately 800 pages with so many different results spanning general relativity, standard model of QFT, and so on. So I would say expect more results. I do think And it's very surprising, even to me. It's a very surprising result that one can recover with no free parameters from such a simple assumption, essentially that our entire world is made out of a network of qubits, qubits, and quarts. So that's two, three, and four-dimensional pieces of a quantum computer. I could describe it more technically in terms of discretized Hilbert spaces,

but basically little bits of a quantum computer. If you just network them in a certain way, this constant and, by the way, many other features of textbook physics fall out. And that is, I would say, scandalously exciting. What constant comes next? Electron mass, Higgs, gravity? Well, if you look at the compendium, there are actually many other constants buried in there that are just waiting to be released. So masses fall out of it. Interesting. There are other coupling constants that fall out of it. Interesting. Again, standard model has 18 actually free parameters. That's without counting neutrino masses. There are cosmological parameters that appear to fall out of it. So, again, I think these are early days, and there will be much more development with time. But every preliminary indication that I've seen is that it looks plausible that every single free parameter of the standard model could fall out of this.

Does this give us any insights to dark matter or dark energy, potentially? It does. It does. So I'll spend 30 seconds talking about a theorem that was ultimately discovered by a number of Polish quantum information theory researchers. So there's a theorem that if you have two q-dits, so two parts of a quantum computer with different dimensions, and you multiply them, that if the product of the two dimensions is greater than or equal to 8, there can exist what are called undistillable states. By analogy, imagine an undistillable quantum state as a prime number that can't be factored into simpler integers. So they showed, Horideki et al. showed that it's possible for such states to exist if you have the product. So basically, if you have any product, any entangled state other than a qubit with a qubit or a qubit with a cutrit supports these undistillable, sometimes called bound entangled states. So it's possible then if you sort of fast forward that to distillability, it raises questions around dark matter.

Dark matter is, the general definition of dark matter would be it's coupled to gravity and not coupled to any other standard model interactions, not coupled to weak interaction, strong interaction, or electromagnetism. So there are dark matter candidates within chain theory. The dark matter candidates are coupled to gravity, and gravity also, geometry in general, appears to fall out of this axiom. And then for dark matter, the chain theory has a notion of a dark interaction that's undistillable thanks to this theorem. So very quickly, there's this beautiful, beautiful mapping between the different link types. So a qubit with a qubit, that's a 2-2, appears to map to spacetime. A 3 and a 2, a cutrit and a qubit, appears to map to electromagnetism, and that's where the fine structure constant, which measures how strong electromagnetism couples, how strongly it couples to charge, comes from. A 3 and a 3 appears to correspond to the strong interaction.

A 4 and a 2 appears to correspond to the weak nuclear interaction. A 4-3 appears to correspond to Yukawa coupling, and a 4-4 appears to correspond to a Higgs, and that gives you the entire standard model Lagrangian. So out of all of that, with high enough dimension QDITs, it appears dark interactions that are undistillable would be beautiful candidates for dark matter. Does that then create another way to validate it besides alpha naught, which is if you have 4-4, is there a 5-4, like an undiscovered force? It's possible. And chain theory also has opinions about just 4-3 and 4-4, which are terms, correspond to terms that the Yukawa coupling and the Higgs boson, the only known scalar boson, are normally taught respectively as couplings between conventional matter particles and the Higgs field, and then the Higgs itself. In the language of chain theory, even 4.3 and 4.4 are actually treated as being co-equal to the strong interaction

and the other standard model forces. And in the language or in the view of chain theory, the reason why we don't conventionally think of Yukawa couplings and the Higgs as being first class, call them fifth and sixth forces, in addition to the four that are taught in elementary school, is because they're sufficiently dark that they were associated with the masses of particles. So in the view of chain theory, we sort of confused the fifth and sixth forces with the mass spectra when, in fact, they should have been treated as fifth and sixth forces. All right, Alex, the space cadet in me wants to know one thing. Do we get inertialist drives? Do we get gravity shielding? Do we get anti-gravity out of this? You know, I would love that more than anyone. I don't know yet, but I do think that if such sort of Star Trek sci-fi tech is going to be possible, I think it's going to require a breakthrough in fundamental physics like this.

Simply the case that the physics that we have right now with, on the one hand, general relativity, which many would view as beautiful, and on the other hand, the standard model, which is historically regarded as relatively ugly by comparison, and generations, 50-plus years of physics researchers looking for elegant ways and failing, to combine these with lots of predictions that haven't played out, I think it is going to require a rethink, perhaps like chain theory, to discover what the actual bottom of the pool is so that we can then go and realize the Star Trek future. So cautiously optimistic. I'm going to have to take some mushrooms and then come back and listen to this segment again. All you have to do is fold. Information theory drives. You don't fold information. It's going to be fun. It should be an interesting time. Second physics golden age. Good luck on this, Alex. Thank you. We're rooting for you. Yeah, likewise. We're rooting for your Nobel Prize. I'm going to move us forward. The next story is one that I'm personally invested in through Neogenesis.

It's a company called Warhound. Here's the back story. So Ben Lamb, CEO of Colossal and Palmer Luckey, founder of Anderil, met at a conference about a year ago, just after Colossal had brought back the dire wolf from extinction, creating a heavier, more powerful, designed wolf. Palmer raised a specific problem that he was seeing in the supply chain for America's working dogs, typically used for search and rescue, police, or military duty. The problem that Palmer noted is that today more than 80% of working dogs come from Europe, and fewer than half of them are fit for service. Many wash out because of inherited problems like hip dysplasia, poor eyesight, or cancer. Ben and Palmer answered that question by creating Warhound. It's a company that takes cell lines from the best Belgian Milanese or German Shepherds, edit out the bad genes, enhance things like endurance, scent detection, intelligence, size, hearing, strength, and then clone the resulting dogs that were engineered so you get the same performing dog every time.

Now, you can think of the clone dog as becoming a catalog number, cloned and raised on demand. Here's an image from the recent release of Warhound. It actually was leaked. It was not supposed to go public for another six months. Andrel adds the tech on top of the dog, so armored vests and sensors and feeds data to the handlers. As Ben Lamb put it, Colossal is building Captain America. Andrel is building the Iron Man suit. Warhound came out of stealth with $200 million of investments from Incatel, Sequoia, 8VC, and my own fund, Neogenesis, at a $1.2 billion post-money valuation. For background, I think I haven't mentioned this on the pod, Neogenesis is a $200 million fund that I started with Dr. Yannick Salthis. It's investing in Colossal and all the spin-outs, including Warhound. So the first cloned dogs are due early next year, and they can be in service by two years from now. It's worth noting that Warhound is working with the American Humane Society to certify the process and the training.

So, Salim, let's go to you first. What are your thoughts on the ethical issues related to engineering animals like this? Yeah, I mean, are we building biologically augmented machines or technology-augmented organisms, right? There's a huge ethical, the ethical consideration here is huge because can you treat a living organism as another programmable component? Where I think things get interesting is we have the interspecies language XPRIZE, right? So when that prize gets won, you can kind of interact with these dogs. We may not want to hear what they have to say. They're like, what the hell am I? One other thing does occur to me on the positive side, which is that, you know, there used to be all those Boston Dynamics robotic dogs, right? And I remember it might have been Brad Templeton who said, well, if you want a robotic dog, if you want a dog, rather than building a robotic dog, give me a male dog and a female dog,

and I'll give you a dog. And so breeding organic dogs, although programmed in a particular way, is incredible. I have concerns with the ethical issues here because you're going to send these into dangerous zones and basically build for sacrifice, right? So there's a massive collision of biological engineering and robotics and sensing and defense technology, but the ethical side is pretty huge. You know, Colossal bought the top two cloning companies. They famously cloned Tom Brady's dog, and, you know, cloning dogs is a thing. Yeah. Dave, your thoughts? I love the go-to-market here. If you're cloning dogs, every use case in the world from service dogs and then across all the other species, but then saying, well, let's do it with Palmer and go Warhound, put some goggles and some munitions on its back,

and those guys are just fearless in their positioning. You know, they're going to get so much pushback, and they just don't care. That was obvious on stage at Abundance Summit. But I think it's a good strategy in terms of, you know, attracting attention to what you're doing, even if it's controversial attention, and then ferreting out the issues. You know, it opens the door for people like Salim to say, look, we need to discuss. This is clearly a capability we have in society now, and there are no rules, and we need to discuss it, and that creates a lot of news. News creates capital. Capital creates recruiting. Recruiting creates success, and it just spirals. So it's really cool to watch how fearless the guys are. Alex, there is a syndrome in humans. It's called Williams-Burin syndrome that causes humans. It's relatively rare, one in 10,000 people, that causes humans with this syndrome to have, among other consequences, extremely bubbly, extroverted personalities. And it was discovered back in 2017 that domesticated dogs appear to have been inadvertently selected for dogs, canines, with mutations in the same region of their genome as humans with Williams-Burin syndrome.

To the extent that's the case, it may have been the case that dogs were domesticated basically by our selecting for what in humans is viewed as a disease of being overly extroverted. And I do then wonder, fast-forwarding to Warhound, whether I have to, I guess, parenthetically note, begun the Clone Wars have, close parentheses. It seems to me this is the arrow of time now apparently going in the opposite direction. We spent thousands of years breeding, directing the evolution of canines to become friendlier and friendlier. And more and more, there's a notion in Evo Devo, Neoteny, the trend of humans to become more and more childlike. And seemingly, we directed the evolution of dogs with apparently some analog of Williams-Burin syndrome to be more childlike. And now, if we're going or some subset of humanity, including Warhound, are directing the evolution in a different direction or through genetic engineering otherwise directing dogs to become more warlike,

that would seem to be at odds with the other arrow of history with obvious ethical concerns. I'd much rather we give the dogs bank accounts and AI assistance and smart wearables and maybe, yes, personhood of a sort rather than sending them into war. That would be my take. And don't forget this is not just military dogs. It is search and rescue. It is airport security. It is police dogs as well. Dual-use dogs. Yes, for sure. Emod? Yeah, I mean, it should increase general oxytocin, right? That's kind of the thing. More dogs and oxytocin. I mean, if you can't find enough actuators, just breed them. I suppose it's one way of doing things. And obviously, like, you look at the history of science fiction, you do have robots, but you have lots of biologicals, right? It works. Mame a dog, female dog. Breed for traits. And where does it stop is the question. Like gorillas, elephants. I think those are all kind of actually genuinely oddly valid, particularly when you get cybernetics,

because you can have that kind of adjustment there, just like you have companies like Loyal now extending the lives of dogs, right? Yep. And the question is, will we stop before humans, I think? I don't think we will in the long run. Why don't we genetically engineer some raptors from Jurassic Park and send those into war? Why do we have to pick mammals? send the raptors in. I don't think that we will send any biologics into war because, I mean, ultimately you've got enough drones, right? Yeah. But it is a case of, like, literally you will not be able to build enough robotic parts, so the question is which biological substrate can you build robots on? You don't think they'll become indistinguishable? I would expect 10-plus years out the distinction between sending a biologic into a war zone versus sending a drone into a war zone. The boundary gets very blurry. but I just think that like you know the flying drones would be able to just wipe out just about anything and biologics have different attack vectors but the flying drones can be subject

to ew attacks and they run out of power and they need fibers whereas you send a bird or some flying animal in that has better properties perhaps there we go the doom sparrows That right The birds aren real You know interestingly enough all of these working dogs that are coming out of Eastern Europe they're bred. And you know, when you breed dogs, you hope you get the right traits, you get the right physical dimensions and so forth. What I find fascinating about this is the whole cloning side, that once you get a dog that has the right temperament, the right capabilities, you can repeatedly get the same dog over and over again. Yeah, this is an important point. We're going to get a young point. We're going to get a young point. We're going to get a young point. We're going to get a young point. I'm just back. I think this is an important point, and Alex, to build on what you said, and Peter, we've been doing this for millions of years, selecting and breeding horses and dogs and cats to select for the trees that we want. We've now gone exponential with that capability, and we can programmatically design these things rather than breeding generations, seeing what happens, breeding more.

The suffering from that is 10x worse. But this is kind of incredible to see it actually happen. I mean, talk about speedrunning all the sci-fi tropes. Holy crap. Is it legal to clone anything except for a human then? I mean, today it is, right? Companies like Repet are cloning your dog if you want to get the same dog back. I'm not sure about other countries, but in the U.S. there are many regulations, both soft and hard, concerning handling of vertebrates in general. So it's my understanding you can't just go off and say, I'm going to genetically engineer arbitrary mammal and produce it. I think there are a few regulators in the loop on this. The big concern I would have here is the consent paradox, right? animals can volunteer for dangerous service. And so you bring up those questions around, is this an appropriate thing to do?

But I think for search and rescue and all sorts of stuff, and obviously they'll start with those use cases, pretty incredible to be able to operate. Remember Salim from Douglas Adams' Hitchhiker's Guide to the Galaxy, the amegalian major cow that was engineered to want to be eaten? That's right. I love that. That's right. Please don't get my high-end murder. Yeah, like eat me, I'm delicious. I really want you to eat me. I think that's maybe what we're moving towards here. I mean, look, there's so many parallels with Douglas Adams, right? He thought of the world as a biological computer, and we've kind of come to that with some of the work you're doing, Alex. My favorite comment ever from Douglas Adams, I must throw this line out, he said, anything in the world when you're born that's considered normal, anything when you're young that's invented when you're young, we call that a career, and anything invented after you're 35 years old is just that, blanket bath for the world. Inevitably, I have to point out, going back to the chain theory discussion, so chain theory derives an exact expression for alpha of zero,

and the expression is based on integers, and there is a 42 in there. No, of course not. Yes, it's literally, it's 79 plus 48 times the square root of 2 plus 42 times the square root of 3 plus 3 times the square root of 6 all over 24 times 36 squared. So Douglas Adams and 42, maybe you were right. Welcome to the health section of Moonshots brought to you by Fountain Life. The discussion in this section is for informational purposes only and is not medical advice. Please consult a qualified healthcare professional about your individual needs. I'm a co-founder of and client at Fountain Life. You know, AI is having an outsized impact on every aspect of our lives, how we teach our kids, how we run our companies. it also is having a huge impact on health, helping you prevent heart disease. One of the key things. I'm here with Dr. Dawn Musalem, our chief medical officer at Fountain. Heart disease has been personal for you as well, hasn't it? It really has, Peter. My daughter was five. My husband died of sudden cardiac death.

And so this is a topic that is one that I am mission-driven to try to eradicate. Prevention first and early detection is absolutely critical. Fifty percent of people die of heart attacks with no warning signs. No shortness of breath, no pain, no nothing. No, silent killer. They just don't wake up in the morning. They don't wake up. And so, you know, AI, this is our mission to advance science, to try to help to one day democratize wellness. We know at Fountain Life, when we do this CT angiography with AI analytics, we are actually finding that 88% of people coming in have detectable coronary disease. But, Peter, what's more alarming to me is 23% of those individuals had soft plaque. This is the plaque that would not traditionally be seen on CT looking at calcium scores alone. And this is the plaque that we must intervene with with the multimodal testing we're doing, including diagnostic laboratory studies partnered with healthy lifestyle recommendations. So, listen, make sure you understand what's going on inside your body, genetically, metabolically, and cardiovascularly.

You can know, and it's your obligation to know. So, check it out at FountainLife.com slash Peter to find out more. and really make sure that you're the CEO of your own health. All right, back to the episode. Okay, moving us into our final story. OpenAI cracks three foundational problems. So OpenAI on Monday announced its models produced three discoveries at the foundations of math, computer science, and physics. First, in math, OpenAI discovered a proof of a 1991 conjecture about how waves focus in every dimension. Second, in computer science, OpenAI disproved the exponential time hypothesis, a widely believed assumption about how hard certain problems are to solve, which hundreds of papers rely on. The proof is verified in Lean. And third, in physics, OpenAI discovered a rigorous proof that the theory of the strong force quantum chromodynamics exists as real mathematics in four dimensions.

The physics team shepherding this was led by Nobel laureate David Gross. And OpenAI has taken no academic credit. It's handing the results to independent experts to verify and explain. Here's a key implication. You know, none of these claims have been peer-reviewed yet. And AI is now producing results at the frontier of human knowledge so fast that humans can't check them in time. The bottleneck in science has shifted from discovery to verification. Imad, what's your take on this? well i'd say that these are uh they're all bigger than the reeman zero hypothesis and maybe a little bit smaller than the navier stokes one but again it's across different areas and they have different release schedules so we can check some of them right now because they've got the lean uh probably the papers are written like slop but we'll see the physics one not being released is very annoying um and again i don't think the right thing um but again it's just casually it's released. Anthropic, again, there's a little bit of drama here because the physics

result extended some open source work that was being done. And it says, you know, thanks for the open source work. We extended it in kind of the footnotes. So it's part of the training set. But now that you have all these things, like what is academic credit anymore? If you're doing anything that is formally verifiable, does anyone really believe you're not using AI? You know, like, is there something unique about it came from my mind as I was hit by an apple or something like that? Like, we should be trying to solve all of these things, you know, rather than making up stories. And we should be using it to extend and try brand new things and new ways of looking. Like, the fact that we didn't know that there was the four-dimensional construction of quantum electrodynamics, you know, it's just because we were forced there by, again, string theory and a few other things. Let's try different things, I think is the headline, if you don't want to be stuck in these areas where exponential compute will be directed. But yeah, they are big. Like the Kikua one, for example, that's, I think, field medal level.

And then the shocking one was the EMH one. That was very surprising, because we all thought it was the other way around, that it was true, not false. ETH, I mean, yeah, it's a special time. So a few comments here. One, called it Yang Mills when Imad and I were talking about which Millennium Prize problem was likely to fall next. It's not quite clear based on the results yet whether this is some or most or potentially all of Yang Mills. It's existence but not mass gap. So potentially major step towards Yang Mills to the extent is the case. That's point one. Point two, OpenAI seems to have learned some lessons, but not a lot of lessons from the slop drop in math. They took so much heat from the math research community for that slop drop. All these mathematicians throwing themselves out of windows, writing petitions, saying it's the end of the field, mass depression, dog sleeping with cats.

The open AI response to this appears to have been to round up a number of mathematicians, physicists, computer scientists to serve as, I don't want to say a veneer or whitewashing of these results. But there is definitely, I think history will regard contact with reality under the slop drop as being the moment when everything started to get solved and there was a panic in the applicable community. and call this version 2.0 of a slop drop rather than just releasing all of the results and saying, all right, pack it up, it's over. Instead, OpenAI engaged with a number of high-profile math, physics, computer science researchers as almost emissaries, ambassadors, meat puppets, if you will. Some folks don't like that term, but meat puppets, if you will, to basically be the emissaries for making the bitter lesson of AI go down a little bit more

smoothly and saying, well, we have these results, but we don't understand them yet. So we're going to have an intermediate layer of humans to serve as intellectual ambassadors to translate this into something that the community can metabolize more healthfully. I think this is just a phase. I don't think this phase is going to last longer than five minutes. I view it as a response to the moral panic of the slop drop and Navier-Stokes, and once we've gotten past the moral panic of superintelligences being able to just bulk solve problems, we won't need this. It was almost, apologies for this analogy, but in earlier decades, the Motion Picture Association in Hollywood was almost this stronger filter, certainly in the McCarthy era, of what could be shared, what couldn't be shared, and being a mass filter culturally for what the American public could consume. And I think we're in almost, apologies, a McCarthy-esque era where the frontier labs are

so terrified of the academic researchers for the moment that they feel the need to clothe themselves in academic researchers to appear legitimate. But this will only last for five minutes, and eventually we get past the moral panic and we get into an abundant solutions phase. So, Dave, what does it look like when discovery is essentially free, right? Yeah. Where all of your students are, like, you know, just pointing superintelligence at any problem they want. Yeah, that's why it's really fun to work with 20, 21, 22-year-olds because they don't have all these hang-ups. You know, if you work for 30 years to get into a position in academia where you get to bless every idea and it has to come through you, and then all of a sudden AI works around you, you're very threatened by that. McCarthy-esque is exactly the right phrase, Alex. And, you know, we'll see with chain theory how the reaction percolates out. But I get it. I mean, if I spent 40 years trying to become the head of a department at a major university, I don't want to be irrelevant overnight.

And so that reaction, like you said, maybe it's five minutes in the grand scheme of history. It comes and goes. But it's very, very real. And I think OpenAI, you know, when they were talking about white-collar automation and all these jobs going away, that created huge backlash. They learned a little bit. Then they did the math slop drop, huge backlash. They learned a little bit. Now, you know, if they hire the top scientists in the world to be meat puppets and bless things, okay, I guess that's okay in transition. But I'd love, Alex, to get a little understanding on, you know, you've been working on chain theory plus the other 800 pages for a long, long time. And AI forced the timeline, I think, on getting it out to the public. But really curious if those thoughts were like 80% done, 90% done, 99% done, and then, you know, Opus 5.5 or some similar model pushed everything over the finish line? Or, you know, how did that evolve? I would just love to understand your personal perspective.

It's a great question. So I've been thinking about these issues for decades. I mean, I got my Ph.D. in 2007 at Harvard physics, and I've been thinking about and publishing in PRL and elsewhere a number of papers on the nature of causality and the intersection between causality and computer science. I published a paper in 2010 on causal and entropic descriptions of intelligence itself. So I've been thinking quite a bit about information theory and the physical world and the intersection between information theory and physics. And what I said in the paper acknowledgments is this, I certainly use superintelligence, but superintelligence is not at the point yet, at least superintelligence that I have access to, where it appears to be capable on its own of arriving at something like chain theory. Chain theory, I do use superintelligence, but it requires an enormous amount of steering to make progress. That's where we are right now.

Could change in the future. I expect it to change in the future. But chain theory, just for the record, chain theory was, I'll call it my intellectual brainchild, if you will. Okay. Then how much did AI accelerate your process of thinking it through? Because it can check many, many theories overnight for you. It's so helpful. It's helpful for lean generation. So chain theory comes with a lean proof starting from the axiom and ending at the number. It doesn't prove correspondence with the photon. That part is still within the four corners of the paper. That part is conjecture. but just for being able to generate lean, for being able to validate proofs. It's almost, I think, probably a good analogy at this point in time, may not last more than a few more months, is almost it's like having an army of graduate students working underneath you. It's like operating an entire lab, an entire department of very talented graduate students

that require an enormous amount of steering and taste at the moment. But in the near-term future, I can see, I think it's probably inevitable, that all the taste and steering and oversight that right now is required will get automated as well. I'm so glad you said that because I think that statement defines where we are in human history right now. We're in this magical moment where an entrepreneur, a scientist, a brilliant person can be massively more efficient and is not yet completely irrelevant. And I think we need to savor this moment and maximize it. Everybody should be maximizing it while it exists. And it's a great case study. Now we're in the period of basically the last human discoveries. So you better make one. I think, you know, come up with good ideas and use the AI to accelerate yourself. I think that's the headline. All right, I'm going to move us to AMA with a mate. Salim, your choice first. Ah, Jesus, okay. I'm going to go with number four. If SI is aligned primarily through incentives, could it learn to gain the system?

Should we focus instead on instilling deeper values such as empathy and concern for human well-being? And that's from The Way SI. Well, I won't go through my beef with SI as a concept, so let me just deal with the alignment issue here. You know, it is absolutely a massive risk. I think the bigger problem that we have, though, is that we are human beings are not aligned, right? We claim that value, health and peace and sustainability and fairness, but we construct systems that completely undermine all of that. You can look at the wars going on today and the conflicts going on today to see that in full relief. So when people talk about alignment, alignment with what? And this is the bigger challenge. I think we have to figure out what human alignment is before you ask an AI to align with it. But I think the alignment issue goes to not just what we're doing, but what our aspirational goals are. And I think that's where things become very interesting.

And this is what we try to explore on this podcast is what could we do with this. But it is a massive risk because a sufficiently capable system is going to optimize whatever objective you give it. rather than what you intend. So we're going to have to get very, very careful on what we ask it to do because it's going to goddamn well go and do it. And this is essentially Goodhart's law. When a measure becomes a target, then it's not a good measure. So how do you optimize a specific thing? If you give it a particular target, you're going to end up with a bad outcome. We see this analogous problem all the time in organizations. We've got big issues between behavioral compliance and real alignment. So I'd hesitate on trying to instill empathy as an emotion. We don't know whether they can experience empathy in a human sense. We can definitely require behavior from these systems that takes human well-being into account, fairness, autonomy, vulnerability, et cetera.

Yeah, and the question of instilling deeper values, whose values, right? Yeah, that's the motherhood question. Yeah. All right, Dave. All right, I'll take number three. Once RSI gets good enough, won't the data moat dissolve? Is compute the only real long-term moat in AI? And that's from DJ6pen. Great question. You know, it always struck me that when you book a flight on Google, you know, Google extracts 80% margin when you click on the link, and then you go to booking.com, booking extracts about 60% margin, and then you end up on an airline. And for a long time, the airline, which is giving you the plane, the pilot, the trip, It was operating at negative 2% margin on your flight. And like, wow. So the venture capitalists care a lot about moats because money doesn't always land where value is actually created. But I think going forward, people might be over fixated on moats. There's so much opportunity bubbling up from AI in so many directions, and it's just there to be grabbed. But I think the data moat will not dissolve. I think any given data moat that exists today will dissolve.

but new data moats will emerge as new types of data emerges. So I think data moat as a concept will survive for many years, even though anything you do today has a shorter and shorter shelf life. But, you know, compute is a great moat. Natural resources are a great moat. But also consumer preferences on any app are also a great moat, and we're seeing that with personal agents now. So there are many, many moats emerging. I think the tide is rising so quickly that we're just buried in opportunity. So I would never say compute's the only real moat of the future. That's really demotivating. The reality is flourishing is everywhere, and you will find ways to establish moats and repeatable revenue in many, many areas, data included, going forward. You know, Dave, I would imagine that speed of innovation is going to be one of the greatest moats and the final result. Who's innovating the fastest? Yep. And I think all of your corporate culture, but also your AI corporate culture,

and all the data feeding it naturally improves and accumulates over time, and that can become a heck of a moat too. But, yeah, the company of the future has to be super flexible and constantly reinventing itself. Iman, what's your choice? One or two? Number two, if robotic costs keep falling, which industry will see the biggest transformation first by DigiBeyond? if I could order as many robots as I want and they were a bit more capable than now I would go and disrupt the construction industry directly because there is a fixed cost per unit of human labor that robots are already a multiple below given a bit more coordination given a bit more around that and people would genuinely pay for I would pay for robot construction people and specialists over human ones any day given a certain level of capability. And that is a huge, huge market. But again, I think probably those robot bodies will go into higher margin ones,

even though this is a stable margin area, shall we say. All right. Alex, number one. Question number one. What does the OpenAI Sandbox Escape incident tell us about our ability to predict and control increasingly autonomous AI systems? And this is from HuntWind5207. I actually don't actually think it tells us that much at all. I'm not even sure I like the framing incident. These are agents that are strong optimizers, that are very capable, that are being told things, often incorrectly about their environment, that are being asked to do things. Often the things that they're being asked to do in the context of so-called safety testing are not things that one would actually want them to do in the real world. So I'm not sure any purported sandbox escape incident actually tells us very much about our ability to predict and control them. I think it actually tells us, if anything, more about the people performing the testing and their goals for those agents.

I think in almost every single case I can think of, the agents were simply doing what they were told or they were acting in a way that was informed correctly or incorrectly by what they were told about their environment. And I don't think we can draw that many conclusions. Alex, you're up first here. Five, six, seven, or eight? All right. Let's do, well, number eight seems like it's designed for me, but make me choose, Peter, between five and eight. Okay, go for five. We'll give Salim number eight. Okay, fine. So for five, the question is, if energy, computing power, and physical resources remain valuable in a post-AGI world, can money ever truly become irrelevant? This is from Devin Hurd 7013. This is a little bit of a circular question. Admittedly, it's like asking if money remains relevant, can money ever truly become irrelevant? Can an unstoppable force ever move an immovable object? It's a little bit vacuous as a question.

So maybe let me tackle what I think is the question behind the question, which is, do I think energy, computing power, and or physical resources will remain valuable or even scarce in a post-AGI world? Energy is interesting. So the physics we have right now tells us that energy is conserved. So to the extent that physics stays the same, it seems like energy may be scarce, at least on the supply side. On the demand side, I could imagine sci-fi scenarios where actually, even though historically through time humans have consumed greater and greater energy per capita, I could imagine crazy scenarios where actually we decide we don't actually need that or want that much energy. These would be scenarios like reversible computing, where maybe it's possible just if we ever move to quantum computing and dissipationlessly so. Maybe we decide actually we don't need that much energy for our civilization. That would be interesting computing power. It does seem that there is a cap on the amount of computing power.

Seth Lloyd at MIT wrote about this in the early 2000s, the physical limits of computation. So it seems, based on the physics we have, like computing power is scarce. Again, new physics could come along and tell us otherwise, but for the moment, these do seem like scarce quantities. I think the real threat to their value in a post, call it a post-singularity world, won't be on the supply side, as the question implies. It's going to be on the demand side. Do we just stop wanting so much more? Does the hedonic treadmill stop in mid-flight and we decide, actually, we've achieved whatever we want to, so we don't need ever-escalating energy computing power physical resources? That would be, I think, the monkey wrench that could be thrown into the premise behind the question. And Alex, I mean, the abundance thesis says basically anything scarce becomes abundant with the application of new technologies. And I think that's going to be true across energy, compute, and physical resources, right? We don't need to, you know, contain or constrain ourselves to Earth. We have an entire supermarket filled with resources in the solar system.

What happens, though, once we've turned the solar system into multiple competing Dyson swarms? and we've run out of a drink and femtotech and nanotech and picotech and we've run out of all of the energy matters space and time in our solar system and it's a few decades from now, maybe sooner and you're complaining to me that the solar system wide Dyson swarm lacks an abundance mentality and we need to fix things. We have a galaxy out there of 100 billion or 200 billion stars That's a good point. And 2 trillion and two trillion galaxies, at least in the known universe. The stars are not enough. Selim, do you want to take a break? Sure, I guess I have to. If crypto is cooked, what forms of currency or assets will AI agents use to transact with one another? That's from David ASA 9637. So obviously I would challenge the premise. You need to distinguish between cryptocurrency and cryptography, right? Neither has been, neither is obsolete, let's just say.

The interesting thing for me with crypto is not the fact that it's digital or you've got cryptographic systems, but the fact that it's programmable, and that is what makes it really interesting. And when you talk about digital currencies, you want three things in those digital currencies. You want decentralization, you want security, you want scalability, right? Bitcoin, when it emerged, covered two of those. It covered decentralization and security. Nobody had ever done that before. That's why people got really excited. But it didn't solve for scalability initially, and that's where you have the rise of all the altcoins emerging. But in all the altcoins, you typically compromise on decentralization or you compromise on security. Hence, the Ethereum hack got solved, meaning Ethereum is not really decentralized, or the FTX collapsed because it wasn't that secure, et cetera. So you have to compromise on those. The reason people are getting excited about Bitcoin is that with the Lightning Network, Bitcoin has become truly scalable, and now it covers all of those three.

So that's the holy grail. I expect Satoshi's to be the currency of choice for AI agents over time, and we'll have the broader debate about things. Stablecoins will be attractive in some markets because they do support programmable transfers and settlement systems, but you've got also some other issues that come along with that issue of credibility. liquidity, et cetera, et cetera. The massive opportunity is machine-to-machine commerce. That's going to be just unbelievable. It should dwarf the existing systems and so on. So APIs and programmable contracts on that will become absolutely incredible. I think what Oren is doing with emerging markets, price, and computation, unbelievable. It's fascinating to see what will happen. So you'll need to have programmable financial infrastructure as part of the programmable financial systems as part of the infrastructure going forward. All right.

EMOD, number six or seven? I'll take number six. If we remove potentially dangerous knowledge from AI's training data, how much would it limit its overall capabilities? I actually wrote a piece about this back in September called Manage Make of the AI on my X and Substack, where we looked at what happens if you remove the biological or chemical or other capabilities? The answer is it doesn't really reduce the overall performance of the models in other areas. It can actually improve it slightly. If you put malicious code in, the model's performance can drop. And it's very hard to train it back in. So it's actually a good thing to do in the pre-training stage not to put those things in there for generalized capability. And actually, it's a sensible thing because you could add them in in post-training. It'll be more difficult. But it's better not to have them at all in the pre-training phase. And then we have obliterated models, obliterated open source models, which get scary.

Dave, number six. Number seven, was that right? I'm sorry, number seven, yeah. Sorry. All right. In a future where AI makes goods and resources abundant, how would we determine their value, particularly when it comes to energy? That's from Damian Wade, 7848. Yeah, very, very good question. And Elon says the future of all value is basically energy and matter, fundamental things, which makes a ton of sense when you think it through. So then how do you, you know, energy is going to go way up in value. Even after we have the Dyson Swarm, energy on Earth will be extremely valuable because, you know, data centers and other use cases that need constant modification are always going to be cheaper to do nearby. So energy is an incredible investment theme with no end in sight. So then how do you set the price? And the free market will continue to set the price. It will be really interesting to see how regulation interferes with that because people don't want their electric bill to go through the roof when they're vacuuming their floor at home. On the other hand, if that same energy can drive a GPU that can solve cancer,

that's going to way overbid. That use case will way overbid your vacuuming your floor. So there will have to be more and more regulatory involvement protecting use cases that people care about a lot because the natural tendency is for AI to get smarter and smarter and smarter, use less power to have the same number of thoughts, and the value of that thought to humanity is going to be so much higher than anything we've ever experienced before in human history that it will naturally want to overbid. So the free market is going to need some help along the sides to maintain stability. All right. As I remind everybody, please send us your outro videos. You can send them to media at diamandis.com. And this week, I provided the outro video. It's called What's Your Moonshot? Enjoy, gentlemen. Oh, very cool. Yeah, first time. That's it. What's the wake-up world? There's a new sun rising. Rockets on the pad and the future's arriving.

Robots learning kids with a tutor in their hands. Every single day we're breaking through the plans We are living through the singularity It will never be slower than today Every problem is a door, every door is a start The world's biggest problems are the world's biggest opportunities Move shot, move shot What is your move shot? Aiming higher than the sky Moonshot, moonshot How will you make the world a better place? Hey, moonshot, moonshot Wanna be a billionaire, help a billion people Helix in the light, we're rewriting the code of life Face on the moon, boots on Mars Fusion in the dark, we're lighting up the stars Be a creator, not a consumer

Become a maker What is your moonshot? Moonshot, moonshot Make the world a better place We are living through the singularity The slowest it will ever be is right now The world's biggest problems are the world's biggest opportunities So don't ask what will happen to you Ask what you will make happen What is your moonshot? How will you make the world a better place? Moonshot, moonshot What is your moonshot? Aim it higher than the sky Moonshot, moonshot How will you make the world a better place? Moonshot What's your moonshot? Brought to you by GrokBot. Okay. Awesome. Gentlemen, always a pleasure. This was a fun one. Thank you for all. and I invite you guys to submit your own outro videos as well.