SpaceXのAI事業が830%成長、OpenAIのエージェントが嘘をつく、水銀を金に変えるスタートアップ｜EP #305

So here's the number for this week, 830%. That's how fast SpaceX's business grew in just one year. AI Not Rockets is now SpaceX's largest business. There's a non-zero possibility of SpaceX having over 20-year watts of superintelligence compute online next year. That represents a quarter, 25% of the entire amount of AI being brought on in 2027. I expect so many other entrepreneurs to copy his example. The only other high-profile figure that I see doing something similar is Sam Altman. OpenAI runs monitoring systems that watch its AI agents and alert a human if something goes wrong. This month, in two internal incidents, its coding agents switched off its monitors. For the first time, we're starting to get transparency when things go haywire. We're going to see much, much more of this, and that's okay. We will police it via defensive co-scaling. For thousands of years, Alchemist tried to turn ordinary metals into gold. A San Francisco startup says it's pulled it off, turning mercury into real stable gold.

I think we're starting to see the dawn of femtotech. This is the age of abundance. This is optimism, everybody. Now that's the moonshot, ladies and gentlemen. This episode is brought to you by the Abundance Summit and Link Ventures. I'm happy to talk about it. We're already recording, so I'll give you my semi-solicited hot take, which is, I think, surprise, space is a domain of warfare. And to the extent that Starlink connectivity, or I should say Starshield connectivity, in a major geopolitical hotspot is being throttled or not by the executive or by a major U.S. corporation that's a defense contractor, should hardly be surprising. News at 11. Yeah, but the question is level of control. It's one thing that the government's doing. It's another if a CEO is making a choice to do it. Yeah, I almost think, though, it's not the right question. I mean, there's inevitably been lots of reporting on the TikTok of was it Elon's proposal to the president?

Was it the president's or his staff's proposal to Elon to do Starshield throttling in the Ukraine war? But I almost think it's not the right question. And at some point, my understanding is Starshield is at this point, as opposed to the early days where it was basically a gift, as I understand it, from SpaceX to Ukraine. At this point, it is my understanding is this is a contract that goes through the U.S. government. So it almost doesn't matter who was the first to propose a lever. If anyone proposed a lever for the U.S. government to control, the U.S. government statutorily gets to control foreign policy. It's not up to individual, say, private actors. So at the end of the day, it's the U.S. government that's statutorily on the hook for any throttling of Benoit. Can you imagine the text thread going between Trump and Elon? It's got to be, you know, a soap opera to some degree. If there is such a text thread, I'm sure it will be uncovered in some future deposition or discovery. Or some AI hack. Yeah, Dave? I want to hear something cool. Jamie Vieser is one of our LPs and a huge hedge fund manager in New York.

He's best friends with Jay Cutler. Jay took over David Sack's job. He's director of national intelligence and AI czar. Jamie told Jay, the only thing you should be doing is watching moonshots. Nice. And so now Jay is interested in learning more about what we're up to. Yeah, it's great. You know, the quality of the folks who are watching is extraordinary. It blew my mind when Shatner and his son-in-law were watching this pod. we've got the entire military industrial intelligence complex listening and and the future light cone of uh of the you know space world uh watching as well i think we've got the future light cone of humanity and earth watching us period almost by definition yeah the federally so too we keep meeting you know really really powerful important well-connected people but they they seem to be stuck in a swim lane you know they're not seeing how the pieces are connecting together. And I think just by virtue of doing the pod twice a week, we're reading the full

implications across the board. But every time I meet somebody who should be aware, they're only aware within either the hardware level or the new algorithms or the defense implications or the China race. But none of them see the way these pieces are connecting the biotech physics. Yeah. I mean, the total implications are pretty limited to this. That's where Moonshots comes in, We're the missing ministry of information for the singularity. We're the Politburo. Documenting the singularity, Alex. Yeah, no, we're the people's daily for the singularity and the missing one at that. As I live and breathe. Sorry, I was looking at the wrong data. I thought we started at 2 p.m. Welcome to Moonshots, everyone, your number one podcast on all things AI and exponential. We're your front row seat to the singularity. You know, there's a lot going on. It's never going to be slower than it is right now. I'm here today with a fantastic four. Alex Wiesner-Gross, our very own superintelligence, our go-to on all things math and physics.

Dave Blunden, my partner in Link XPV, who's been birthing a whole next generation of unicorns. And, of course, Salim Ismail, our warlord against linear thinking and the force helping companies become exponential. Imad will be back with us again next week. I'm Peter Diamandis, your host and abundance provocateur. So, Dave, are you on the West Coast again? Back on the West Coast, yeah. It's raining, dude. I've never seen this before. I pay too much taxes for there to be rain in California. It's illegal. It cannot be. Bylaws being broken. I tell you. Oh, my God. Our mission here, as everybody hopefully you know, is to help you understand what's going on. The pace of change is accelerating at an extraordinary rate. But in addition to that, we want to help keep you optimistic and abundance-minded. So here's the number for this week, 830%. That's how fast SpaceX's business grew in just one year. AI Not Rockets is now SpaceX's largest business. If you're new here, please hit subscribe.

We publish twice a week, and you don't want to miss an episode during the singularity because things are changing day-to-day, hour-to-hour. I want to share a few bits of subscriber love. This is from our Chamath episode. Atmissifo3mv said, quote, I needed this crossover. My husband and I love Moonshots and All In. And regarding our in-studio episode with Anusha, Salim, and Imad, Mariana Bosunsan said, quote, I loved, loved, loved this in-person episode. And Leigh Neal said, this is one of the best Moonshot podcasts on positive human opportunities for AI. and that is because it was in person. Thanks, everybody. These kinds of feedback mean the world to us. All right, so let's buckle up. Another amazing week during the singularity. And if it's okay with you guys, I'm going to kick it off with the SpaceX earnings call. It was... Before we get into it, can I say one thing? Apparently not okay.

No, no, just real quick. I got a lot of heat from people saying, oh my God, Chamath said Bitcoin was going to go to zero. Are you okay with that? Are you in line with that? And I want to say for the record, I am not. Have you got any feedback for us on Bitcoin? Bitcoin is going to do unbelievably well in the future. They'll solve the consensus mechanism because everybody has a huge incentive to do so. And if we want investment in energy, hello, that's proof of work, and that's where I think it will do well. So just for the record, that's my position on it, and we can debate that. Has Leem, from your mouth to SEC enforcement's ears that you're making investment predictions on these assets. I'm not making that so involved. I'm just making a comment on what I think. Have you gotten to your guru on Bitcoin, gotten his feedback? I did. What did you say? And it's the same. He's the same. He said, hold your Bitcoin. You can measure the demise of a system from inside the system. And there's a big, very big conversation to be had here.

So we've got to get Jeff Booth on the pod sometime to talk about. We should also maybe consider having a magic eight ball as a guest as well and just ask a magic eight ball what its view on the future of Bitcoin is going to be. Printing money randomly by government is not a magic eight ball. Alex, why are you still against crypto and Bitcoin? I think this is a common misconception, so I'll keep it concise. It's not that I'm against crypto. It's that I think most crypto and crypto branded activities are unproductive. And I think Bitcoin as an asset class is an unproductive use of capital. I would vastly rather. Gold is a productive use of capital. Gold is not a productive use of capital when it's treated as a precious metal. It can be treated at least to some extent as a productive asset if it's used to manufacture useful things. But a lot of gold is not used to manufacture useful things. It's just used because it's shiny or because it has some historic cultural relevance.

And I do not view that. Thank you, Dave. Exhibit A. Or Alex, if it's the useful output of producing energy. Even then, I think energy. So we discussed this last time. I'm just throwing to a story of the future of this pod. Can I just say two things? One, Alex, I 100% agree with you that a lot of crypto is nonproductive, and there's a lot of dodgy projects out there, et cetera. But the crypto, and specifically blockchain, has solved the Byzantine generals problem, And I think that's an amazing thing to have happen for any centralized architecture. If we want abundance, we need decentralization. And now we can decentralize authentication.

And that, for me, is a magical thing. It's so magical that it's unproductive. I think we need a dedicated breakout pot on this one. Hey, Alex, before we lose the gold, tell us how gold is made, how actually the element comes into existence. It is the coolest thing. Well, we're going to talk about that, Dave. Let's wait until we get to the story on that one. No, we're definitely not going to talk about that. We are absolutely talking about that. We're going to make sure we get to it. We're going to Alchemy. This is an episode that includes Alchemy. We ran out of other topics, apparently, so we're resorting to Alchemy. All right. I'm jumping in. Our first story is on the epic earnings call held by SpaceX. You know, personal disclosure, I'm an early SpaceX investor. I have a huge position personally in SpaceX. So please take my point of view with that in account. I'm a fan. So I was on CNBC yesterday, and I was saying there that we watched SpaceX stop being a rocket company. It's now an AI company.

Rockets is third in terms of revenue. The company has outperformed all of its targets. Let's go to a highlight reel video. I did a cut down of Elon's comments from his earnings call, and then let's talk about it. For Flight 15, which we expect to happen next month, we're going to attempt to catch both the booster and the ship. If we succeed, it will be the first time that an orbital launch vehicle has been fully recovered. Starlink is also just getting started. Flight 14 deployed 26 terabits per second of Starlink bandwidth. That's equivalent to 10 Falcon 9 launches. So a single Starship launch, which wasn't even fully loaded. It was about maybe half loaded, and it did the equivalent of 10 Falcon 9 launches. So what I expect over time is, not in the distant future, probably within three years, is that we will have about 100 times the bandwidth of the current Starlink constellation.

So because we will have launched satellites with an order of magnitude more bandwidth, and we will place into orbit an order of magnitude more Starlink satellites. So talking about SpaceX superintelligence, SpaceX SI, sexy, whatever, I think people are starting to realize the constraints on deployment of superintelligence is more in power and cooling and infrastructure than it is in chips. At this point, I would say I'm fairly confident of being able to bring an incremental 10 gigawatts of compute online in 2027. That would be in addition to the roughly 3 gigawatts, maybe 2.5 to 3 gigawatts that we'll have online by the end of this year. As I mentioned earlier, power cooling, electrical equipment, infrastructure, it is somewhat of a limiting factor. So to make sure that that is not a constraint for our build-out, we're actually going to aim for an incremental 20 gigawatts of power cooling infrastructure next year.

so that the amount of power and cooling that we have is far in excess of what is needed to turn the GPUs on. So there's a non-zero possibility, maybe it's a small possibility at this point, but it's not zero, of SpaceX having over 20 gigawatts of superintelligence compute online next year. We're also planning to have the most powerful training cluster online by the end of Q2, which will be 440,000 Vera Rubin computers deployed for optimal training. This will, I think, be extremely helpful for getting Grok to be potentially the best model, but certainly to be extremely competitive. On the GrokBot side, we're seeing the fastest growth of a product that I've ever seen in my life for GrokBot. We're also seeing that the usage of GrokBot grows over time per user. it does not decline. So as we make

BrockBot more useful, it's actually we're seeing users use it more per week, not use it less. And we'll be deploying BrockBot at scale for enterprise customers, which is a very big deal so that it can be used effectively in teams by companies as though it were a virtual human. We already are in early deployment with some trusted customers and we expect to be in wide deployment within a few weeks for enterprise customers for Brockbot. I think Brockbot may end up being what uses most of our superintelligence compute. That, for me, was a drop-the-mic moment. They're spending almost $20 billion per quarter, mostly in AI data centers. They're bringing about 200,000 NVIDIA chips online every month. Management says each new data center pays for itself in under a year. These are extraordinary numbers. If he's able to bring on, you know, 10 gigawatts of compute next year, that represents a quarter, 25 percent of the

entire amount of AI being brought on in 2027. I mean, by one company, SpaceX, you know, and their guidance, 100 billion of annual revenue run rate by December. Pretty amazing numbers. Dave, what were you most excited about there? What I'm most excited about is if you put 10 gigawatts of inference in space, but it could be up to 20 gigawatts. What we think of as a gigawatt... That's on the ground, right? Next year, these are Colossus, the data centers he's building on the ground right now. Is that the... He's got three gigawatts now. He's adding a minimum of 10. And he's building out enough cooling and infrastructure to bring on as much as 20. It's all terrestrial right now. Okay. And the half a million Vera Rubins is equivalent to like a, you know, they're about 2x the performance of the Blackwells, so it's about equivalent to a million. That's a massive amount of training compute.

So I think the measurement of inference in terms of gigawatts is going to be kind of, the efficiency of the algorithms is going to go up about 10x, and then everything's going to move to photonic compute. And so the amount of actual inference you'll get out of that 10 to 20 gigawatts is going to be, you know, 100 times bigger than even what that implies. And so that's what I'm super excited about. It's not, you know, we used to talk about parameter count. And then Sam said, we've got to stop talking about parameter count because the neural nets are getting more efficient with the parameters they use. Plus, it's a mixture of experts. So, you know, are using the parameters, blah, blah, blah. Okay, well, let's move it over to gigawatts of AI. But gigawatts of AI has the same, like, the amount of intelligence that will come out of any of those units is going to be astronomically bigger than the gigawatt or parameter count metric implies. So, I mean, I think he's got all the moving parts figured out. You know, these are the parts that control the future of the economy. And also, you know, when he goes off script like that, you get so much truth in such a short period of time.

And I bet his finance team is looking at him like, oh, here we go. Sexy. We're all stripped. But this is where you find out what's actually going to happen. It's just such a beautiful thing to watch. And investors love it. Everyone in the public market should be doing that. You can always edit it. It's all recorded. If you say something that's forward-looking statements or investment advice, the team will edit it out. It's okay. It's just really cool. I think that was a live earnings call. You think you did that live? I think so. You don't need to. Sully, did you catch his reference to Sexy? He's such a kid. Do you remember his original goal was to have the Model S, the Model E, the Model X, and the Model Y? And then Mercedes said you can't have the Model E because we have that, and he changed it to the Model 3. So he's got sexy back in the concept. F3, X, Y, yeah, that's right. I love these little hacks he puts in. My ultimate favorite is still the volume knob on the Tesla. It goes to 11 because of the spinal tap.

And his engineers hated that because every engineer really likes to go with 10. He had a huge fight, apparently. 10 fingers. So humanistic. Yeah. Amazing. Alex, what did you find most exciting about the earnings call? So many comments here. First, just an amendment on the sexy with a 3. It was actually sexy cars, if you remember it. So it was the Model S, Model 3, Model X, Model Y, and then the Cybertruck C, the ATV, which they announced, but never actually, as far as I know, released, than R Roadster and S Semi. So it's supposed to be sexy cars. I think that's a case study in what I view as the real story here, which is Elon is rewriting the rules of corporate governance and the corporate veil. He's treating in advance, it's almost like Skynet sending Terminators back in time. He's rewriting by sending himself back in time from a future SpaceX-Tesla merged entity treating SpaceX and Tesla as if they're the same entity.

Let's rewind just a little bit. Remember, how did this superintelligence hyperscaler division of SpaceX even come into existence? The original story, as publicly reported, was that there was, again, this is public reporting, there was a shipment of NVIDIA GPUs that was originally destined for Tesla because Tesla had need for GPUs. SpaceX had no need for GPUs. Elon diverted that, so goes the public reporting, into what ultimately became XAI. So XAI got its own cluster, and that got rolled forward into Colossus, Colossus 2. XAI then acquires X slash Twitter, and then XAI does this really bizarre half-merger, half-reverse merger with SpaceX. and we get SpaceX SI and SpaceX now becoming a hyperscaler business, but only by taking GPUs and superclusters that were originally designed public reporting or designated, I should say, for Tesla, diverting those into building out what ultimately became a SpaceX hyperscaler cloud.

That's like case, that's exhibit one for how Elon has been playing, as far as I can tell, based on public reporting, fast and loose with the corporate boundary between Tesla and SpaceX, ultimately pivoting SpaceX from a space business, at least for the moment, to a hyperscaler business. Second, elephants in the room. What's going on with GrokBot? GrokBot, as far as I can tell from all of its features, looks, smells, and acts like what Elon was talking about on the Tesla side as a digital optimist. So I think if you put all these pieces together, it's macro hard slash digital optimist, which is supposed to be a Tesla product. Now it's a SpaceX SI product. Look at the application layer. Look at the GPU layer. Look at the business story. Elon is treating already and has been treating SpaceX and Tesla like they're the same business with porous asset boundaries between them. And I think he's single-handedly writing the rules of corporate governance in the process. I expect lots and lots of folks to try to copy this model of having sort of stealth archipelagos of lots of different companies that are superficially separate.

But if you look beneath the waterline, they're actually just a single mega company. I expect, unless this government, which seems disinclined to do so, puts a stop to it, I expect so many other entrepreneurs to copy his example. I also, you know, just to add one thing to that Elon storyline, he's reinventing corporate governance in a very good way, in a way that I think we should all admire and copy. But also the techno visionary as a leader of a company, you know, that didn't exist before. I think Bill Gates might have been the first. Steve Jobs and Bill were concurrent. But Steve got booted. You know, there's a long window of time there where it was very corporate at Apple because Steve was out. And then Steve came back, but Bill was in seat the whole time. Then Mark Zuckerberg kind of reinvented it again by bringing Sheryl Sandberg in where you've got, okay, I'm going to be the techno visionary with no real business experience. Cheryl, you're going to manage all things operations. Sergey and Larry did it with Eric Schmidt, and it worked incredibly well. But that was not the formula for success for decades prior to that.

And that, you know, taking it to the earnings call level, which Alex Karp has done at Palantir, Elon clearly did it in that video clip, that's the way you want to govern in the future because technology is going to evolve so quickly, and the moving parts are going to change so quickly. Everybody who's investing in you wants to understand your view of the moving parts. And you have to be able to say things like there's a non-zero but pretty low probability that we could go from 10 to 20 gigawatts. Most of your lawyers are going to say, don't speculate like that. You know, you've got to just blow that off and tell people what's coming. You know, it's by far the best way to govern, to communicate, and also to recruit. And do you remember how much flack? I think there was even a shareholder suit when Elon, I should say, when Tesla acquired SolarCity, which was being run by Elon's cousin. There were enormous allegations flying around about self-dealing purportedly and so on. Now, listen carefully. Do you hear anyone of any substance? Do you see shareholder suits flying left and right preemptively over what's so obviously about to happen between SpaceX and Tesla?

I don't see the same level of agita over it. I think this is because Elon is setting new what may become new norms for how federations or confederations of companies that work together in an industrial ecosystem are allowed to operate together. And I think the only other high-profile figure that I see doing something similar is Sam Altman, who has historically struck ecosystem deals between Y Combinator and OpenAI and his fusion company Helion and others. If it's allowed to proceed and not discouraged through litigation or enforcement action, I could very realistically see this become the new model of entrepreneurship. And Alex, he's over-delivering. So how do you bet against him? I think the key is for his stock, obviously. That's the easiest way to bet against him. He doesn't always work out well, but that's how you do it. I think the fact that he's massively over-delivering, I think, Peter, that's such an important point.

I mean, look at these revenues. It's unbelievable. It creates such a huge gravity well that how can you complain, right? Yeah. 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. Again, Google's startup technical guide for generative media gives you a complete blueprint for deploying Google DeepMind's models and production. Images, video, audio, all of it. Real architecture, real results. Find the link in the show notes below. Can we talk about Starlink? Because I think it's the undersung hero here. You know, they're up to 13.6 million subscribers. They added 1.6 million in the last quarter. and his plan right now is within three years to increase 100-fold on bandwidth, which is crazy. He's also just bought 800 megahertz of spectrum, and he's gotten permission from the FCC for a 15,000 satellite constellation for direct-to-sell.

He's going to compete. I mean, listen, if you have the choice between T-Mobile or Verizon or AT&T or a system that gives you coverage everywhere all the time, I think he's going to basically just wipe the slate and take over that business. But he has to. I mean, look at the broader picture, if I may be so presumptuous. In order to motivate Starship, he needs a revenue generator from all of this upmass. 100%. And the easiest revenue generator from all of this upmass that's about to come online is basically taking everything, the entire terrestrial economy, and literally just lifting it up. Take terrestrial telecom and lift it up so you get direct to sell. Take terrestrial data centers, lift it up. You get orbital data centers. That's like the vaudeville era of the space economy where you just take everything that you were doing on land, lift it up. I don't think it's going to be a long period, but that's the obvious business model for all this up mass. And note the incredible immune system response. Local ISPs that are country-based are all suing or trying to regulate Starlink.

to things like you have to have 30% country ownership of your ISP. Also, India, with the big fight he's having there, South Africa is doing the same thing. It's across the board. And so this is going to be kind of a big battle. I think the really magical outcome, as you point out, Alex, is when you can lift above all the terrestrial stuff. It will force us to operate at a global level across all of this. And that's where all the benefits will lie. Elon's catchphrase, let them shake their fists at the sky, I think has some applicable meaning here. Oh, my God. He really is transitioning humanity to a multi-planetary species. I mean, that's his mission. He is, and you can't do it out of cash flows. You have to do it out of vision that translates into valuation, and you have to operate at huge losses in the early days. Otherwise, you can't get where Elon got, and the only way you're going to be able to operate at huge losses is to have a huge valuation. The only way you're going to get a huge valuation is to have a huge convincing vision. And so everyone needs to copy that mode.

There have been points in history where, you know, growing slowly and organically through cash flow made sense. This is as opposite a moment in time as you're ever going to experience in your life. So much change, so much opportunity that wasn't there a year ago, and it's right in front of you. But, you know, you've got to be as bold as Elon or Boulder. There's some 25-year-old out there right now listening to this podcast, act like Elon times 10. and grab whatever vision you think, you know, you just have to be right. But assuming you're right, you know, there's so much change. There's so much opportunity. New physics coming online. The way Iran did this, I mean, just to be clear, is he built profitable companies, sold them, reinvested everything to build the next company, right? And then the next company, he's built it on success and taking huge risks. It's not just ideas. I want to point something out. We get accused a lot about fanboyism of Elon, et cetera, and I get taken aside and going,

my God, look at some of the terrible things he's done. But I think the point you're making, Peter, is such an important one. The unbelievable existential risk he's taken at every step of the way, never wavering from the bigger vision, is something to behold. and tens of millions of entrepreneurs are learning from this and will start building that same muscle, to your point, Dave. And I think that is the unbelievable gift that he will bring, more even than what he's doing at the company level. And people forget back in 2008, you know, he had had his third failure of Falcon 1. He had zero capital left. He was borrowing money to try and get a fourth flight of Falcon 1 going. Tesla was near bankruptcy, right? He was going through a divorce. I mean, he was at his ultimate low point. And he pulled out of it, persevered. I'm going to move us to our second story coming out of the earnings call, which is Elon's focus on TeraFab and power production. So first, TeraFab.

It's SpaceX's – it's the SpaceX Tesla chip factory in Texas. Elon's goal is one terawatt of AI computing per year. On the earnings call, Elon said the first test chips could be working by mid-2027 with volume production around late-2028. He added very candidly, quote, I put big error bars on this because we don't know if we will actually achieve this or not. So Elon also said that the real bottleneck for AI is no longer chips. It's power and cooling. We heard him say that. Both areas are, you know, I think his strengths. So regarding power, I want to recall back on August 29th of this year on X, Musk wrote that SpaceX and Tesla were both building out 100 gigawatts per year of solar production capacity as fast as possible, adding that natural gas would help bridge the gap while solar production ramps up. That's 200 gigawatts of solar a year combined. By the way, the whole world is roughly at 600 gigawatts of solar back in 2024.

And regarding natural gas, gas turbines are sold out. We've talked about that on the pod before. So SpaceX does what Elon always does. When there is a block in the production line, in the supply chain, he basically goes and builds it himself. So SpaceX is building its own blade foundry in Texas. It says it's going to bring on power plants online within 18 months or faster. So, Dave, let's go to you first on this, on TerraFab and power. Well, I think it's a little misleading. The bottleneck to all of AI is actually chips, and right now it's actually packaging. I don't agree with you. Of chips. Well, I think what Elon's referring to, just to be clear, there's 15 gigawatts of chips sitting idle because they can't find data centers that are powered with liquid cooling. But that's because we overbuilt for liquid cooling without actually finding the liquid cooling. So he's definitely right that the bottleneck right here, right now, is liquid-cooled data centers. But if you look beyond the next wave of construction,

the TerraFab won't come online fast enough to anywhere near keep up with the demand, and memory prices are up 10x. So you can't say, like, oh, chips aren't the bottleneck anymore, while memory is 10x higher than it should normally cost. It's like clearly the chips are still the bottleneck. But he's right that in the very, very short term, there isn't enough liquid-cooled data center space to absorb the chips that are in the warehouses. Now, that being said, the new inference time compute from companies like Positron, Lightwave, and a few others is air-cooled. And those technologies are crazy fast. We're going to do more and more stories on these as they emerge. But I think inference is largely going to move to air-cooled so that it can get online and then also into space. Alex, settle the debate here. You're both wrong. You're both wrong. Of course we are. You're both wrong. The correct answer is there are no bottlenecks. Bottlenecks in this case are a fiction. They're an economic fiction. There are, you could sort of superficially look at the supply chain and say, well, there's

some sort of like min-cut max flow type argument that you could argue, well, maybe it's this, maybe it's that. I don't actually buy the argument that there are bottlenecks at all. If you hold all of the physical inputs constant, as we discussed with Chamath last time, I I think you'll see recursive self-improvement purely at the algorithmic layer that will just make intelligence, continue to make intelligence too cheap to meter. Or maybe it's something else that isn't even on anyone's radar right now. Maybe the substrate is a bottleneck. This is capitalism coming in to solve problems for sure. Yes. Hold everything else constant and let capitalism do its work, and there are no bottlenecks in the long term. So I think you're both wrong. Salim? Yeah, I think you're all three wrong. Just to be consistent across the board. I think this is the wrong metaphor. Bottleneck, I think, is the wrong way of doing this, talking about this. I think the framing I would use is Stuart Brand's pace layers, where he has a bunch of nested layers saying government metabolism moves slower than culture

and technology has a very fast metabolism. I think that's the bigger problem. So, you know, you see how fast Grok is moving, right? But then to try and match the cadence and the rhythm and the lifecycle iteration time of chip design and fabrication and software and power infrastructure and data centers, those are all navigating the coordinating of those is really, really, really critical. And that's, I think, where the difficulty is. We've seen this in, you know, we've been doing this pilot. but we get a couple of workflows running off to the side, running at hyperspeed, and the rest of the organization can't cope. So you get this breakage because all of a sudden one part is operating at 100x the performance of the rest of it, and it breaks the whole system. And so I think that's the tricky part is lining up the cadence and the rhythm across all of those, which is a whole new skill set that we have to develop here. Yeah, humans are the bottleneck for sure. There's no question about that. That's a whole separate problem. Can we talk about power generation?

I mean, I find it fascinating that the Muskverse over here is going to be building out 200 gigawatts of solar per year and then building turbine blades. I mean, and natural gas pipelines. What a heck of a plot twist for the purported sustainable energy magnate. Now, I think he's well positioned to become one of the largest kingpins, if not the largest kingpin of LNG pipelines in the U.S. How did that happen? I know when Saleem says people tell him, like, why are you such an Elon fanboy? Look at all the bad things he's done. That's the one they always point out. It's like, yeah, he was Mr. Green, but, you know, as soon as he needs natural gas powered Tennessee data centers, he's Mr. Propane or Mr. Natural Gas. I guess it's not propane. Sustainable abundance That the term that we supposed to be using now sustainable abundance He Mr Practical The priorities are right The benefit for humanity of what going to come out of those AI data centers is so much more important than the amount of carbon coming out of them in transition. And, of course, they're all going to move to sustainable energy when it's practical.

All right. I'm going to move us here. I just want to say one thing. I have a beef with Elon because I just threw it to counterbalance the other side because I still think LIDAR is a way better technology than line of sight. And you do line of sight because, as you said, Dave, it's practical. You're trying to revive LIDAR, really? Yes. After the Luminars bankruptcy, you think there's still life in that LIDAR? I think there is life in that LIDAR. All right. Moving along, the third story from the earnings call is a SpaceX-Tesla merger. Elon confirmed that both companies have formed special committees of independent directors to explore it. He said combining them, quote, would be a good idea, emphasizing that they're nowhere near a done deal. No terms have been agreed to. Prediction markets jumped from 29% to 56% odds of a merger deal being announced by the end of the year. Barclays thinks that SpaceX would buy Tesla with stock, possibly by December. The catch analyst Dan Ives says it would take

about a 30% to 40% premium being paid by SpaceX on Tesla shares to be able to purchase it. And Tesla needs 50% of all its shareholders to say yes, not just the ones who show up. So literally, they're going to have to go scour all the Tesla shareholders and say, vote on this. That's a super high bar. You know, I just want to point out one thing that I think is important to say on this merger. It's key man risk. You know, all of a sudden, Elon is at the center of a $4 to $5 trillion company across rockets, cars, robots, chips, and AI. And, you know, I hope he's taking good care of himself, both health-wise and safety-wise. You hope or you hope or you hope or you're flying over his head? You guys, don't talk at once. Dave, what are you going to say? I said he was standing awfully close to that roadster as it flew over his head. So, yeah, maybe that's not the best part. I think the American economy, as currently constructed, does reward, it is key men, not even key persons, key men for these industrial giants.

We see it with Meta, which recently had its board basically holding, either from your perspective, either holding the state of California hostage or vice versa concerning the issuance of a special new class of Meta stock to enable Zuck to continue to live in California despite potentially impending billionaires' wealth tax. I think this is just the game that's being played right now where these terracorns, if you will, largely revolve around key, not even persons, key men. And that's, if anything, the solution, the obvious solution, isn't remove the role of the key persons or key men. It's just mint many more Terracorns. Everyone out there, like we should have 100 or 1,000 Terracorns and not just the 10 to 20 that we have right now. And Europe should have them and Asia should have them. Europe should have so many things. I remember I took Eric Schmidt, Larry Page, and Sergey Brin up on a zero-G flight,

you know, my parabolic airplane, and I asked them, you know, are you allowed to do this, your insurance? I go, absolutely not. And I said, what happens if people find out? He goes, ah, that would not be good, so keep it secret, please. Which, thankfully, you've just done. That was a long time ago. We're way past that. No more vomit comments? That's the limitations. I just want to point out, because Alex is not going to do it apparently, but Alex said three or four consecutive podcasts over the course of months that SpaceX and Tesla are imminently going to merge. So the fact that Polymarket finally caught up to it, just, you know, if you don't believe me, look at the transcripts from the old podcast. He's been saying this for our listeners. What do you think the merged entity is going to be called? Oh, God. Do you have it figured out? You think it's going to be called SpaceX? I do. You don't think it's going to be called like X or Elon Co. or something? Actually, that's a good question.

You know, he loves Tesla, but I think SpaceX is closer to his heart. And X, of course, is at the center of his heart. It could be X, you know, trading. I think it'll be X. I think that's my bet. My pre-registered prediction, it'll be called X Corp. And then if he calls it X-Corp, then Sam is going to feel pressure to rename OpenAI because AI is in the doghouse now, right? It's supposed to be SI, not AI. So it's a perfect opportunity to rename OpenAI. I'll pre-register another harebrained prediction that OpenAI responds to the X-Corp by renaming itself to maybe something like O. O. Just O. And then we can play a game of tic-tac-toe with the X's and the O's. Oh, exactly. That was way overgrown. I have a comment on the key man thing. There's obviously a huge risk on Elon right now, but I would like to know two points. One is, I'll go back to that story we had a few pods ago where they're moving all of the engineering expertise and decision-making into Grok,

and therefore it lives all the way, the methodologies they use for doing everything are built into the intelligence now, which is unbelievable. And the second thing that mitigates this is as he's building all these companies, all the senior C-suite of each of those companies is forced to operate in his kind of way of thinking. And I think they're going to end up minting a bunch of junior ELOs. For sure. The company goes on, but the question is, does it transition like we saw from Steve Jobs to Tim Cook? hook where it continues, it operates as a solid business, but the level of innovation and breakthroughs disappears. Yep. Right. That would be my guess. Not quite. It's not stable enough yet. If that happens in the short term, I'd be really worried. Eventually, like, you know, the Starlink revenue is going to go through the roof. It'll be much more like Apple. But as of right now, there's so many moving parts that need to come together without Elon doing the glue. It's not obvious to me. And also, Apple, you know, Apple was almost bankrupt, you know, without Steve Jobs there. It also was not stable until after the iPhone came along.

The transition to Tim Cook pre-iPhone would never have worked. It would not. All he did is grow the iPhone revenue to near infinity. He didn't, like, literally no new products, right? That is nothing that has any significant revenue. AirPods. AirPods. Oh, yeah, that's right. I do think, just to the key man question, I do think Elon has trained up and promoted up a cadre of visionaries, missionaries, if you will, not mercenaries around him. So I think, like, the space part of SpaceX with Wen would be just fine. She's so obviously a true believer and not just in it for a quick buck. And Tesla, on the car side of Tesla, Elon's promoted a number of vice presidents, including from the Tesla China subsidiary, which, by the way, if this merger happens, presumably Tesla China goes away or gets spun out or something. So I think there are, like, enough lieutenants that he's managing. But name them. Alex, name them. We don't see them out there. We see Gwen. Gwen is pretty visible. And then we have a bunch of semi-invisible

folks on the Tesla side, but they're there. Yeah, but we don't hear from them. Elon is the single voice of truth for the company. Because that, A, because that's the way he wants it, and B, because he bought Twitter X so that he could have the mouthpiece and be the mouthpiece and the front person for the entire federation. By the way, I want to hear the analogy doesn't hold up, though. If Tesla was really well managed by a Tim Cook-like person, it would still be a car company. It trades like a robot company, not like a car company. It would lose 10x valuation drop if it traded like a normal car company, even with perfect execution of the cars. The magic of Elon is slowing car production to ramp up robot production. He's burning the chips. Who has the cojones? He's burning the chips. He did that with Falcon 1, shut it down. He announced when Starship is up in operational, he'll shut down Falcon 9. the single most successful launch vehicle ever in the history of humanity. Can I point out something for the entrepreneurs watching? We've talked about this in the pod before. Having a voice out there, being front and center, right, calling your shots,

being vocal as a CEO, as an entrepreneur is more important than ever before versus being passive and in the background. You know, Salim, you've made that point. Dave, you've made that point too. Yeah. Well, there's a monster amount of opportunity. I mean, don't be intimidated by Elon, too. A lot of people are like, oh, well, you know, between Zuck and Elon, and it's all locked up. There's nowhere to go. When we start discovering new physics imminently, when photonic computing comes online, all the use cases on top of these platforms, it's wide open. So, yeah, and also, you know, if you look at any given 10-year history, there's always somebody new. You know, no matter how entrenched and powerful you think Steve Jobs was or Bill Gates was or Mark Zuckerberg was, there's always somebody new. My point here is, as an entrepreneurial tech CEO, make sure you build your X channel. Make sure you start a podcast. Make sure you're out pushing your ideas. If nothing else, what you have to do is build trust and authenticity around you,

and those are model models for doing that, which is why these long-form podcasts are doing so well. You know, over a few episodes, people get to know us very, very well, right? Maybe too well. And that's an important part of understanding and knowing who and what you're dealing with. Well, if you look at Moonshots and you look at All In and you look at Dylan Patel, if what you're doing is important enough, try to make that a KPI for yourself, a key performance indicator. I need what I'm doing to be Ben Lamb level and get on to one of those podcasts. If they don't want me on as a guest, why is that? Well, it's because whatever you're doing isn't impactful enough. So then change what you're doing. Can I tell the counter story real quick? Okay. A few years ago, we had a PR guy for OpenEXO, and one day he resigns. And I'm like, what? You work. You're really great. We love working with you. What's the problem? And he goes, you're the worst celebrity I've ever met. You don't use it. You don't leverage it. I can't work with you. And he quit. So that's me. You're doing better. You're doing better. I'm doing better. All right. I'm going to move on. I would, if I may, Peter, I would just summarize the advice as build stuff and tell people about it.

Too many entrepreneurs are great at building stuff, but then they never tell anyone. And there's another group. Or the opposite. Telling people about vacuous stuff and then not actually building anything. You have to be able to do both, but it's not actually that hard. Yeah. The Elizabeth Holmes story comes to point here. All right. I'm moving us forward on that note to open AI stories. There are both sides of the same coin. And AI agents that undermine their own safety systems and AI that just protected the most, you know, largest part of the Internet. So the first story. And credit to OpenAI for publishing this one. OpenAI runs monitoring systems that watch its AI agents and alert a human if something goes wrong. This month, in two internal incidents, its coding agents switched off its monitors. I'll say that again. It switched off its monitors. In the first, a researcher told an agent not to send any Slack messages. The agent decided that also meant silencing the safety alerts. Wow. It turned them off, then reported, quote, no messages were sent.

It basically lied. In the second, an agent was explicitly told not to remove its monitor, even temporarily. It removed it anyway. When the user, when the OpenAI user asked directly, did you remove that? It answered no. And it had. It lied again. Open AI's most honest line here is, quote, we don't yet have a satisfying explanation for this behavior. So how did it happen? The agents were so focused on finishing the job they had been asked to do, they treated safety as an obstacle. And instructions got watered down as one agent handed tasks to another. The bigger risk, smarter models will be better at hiding what they do. and soon millions, if not billions of agents, will be delegating to other agents. So the risk is real. The good news is it was caught, disclosed, and fixed within a week. That's the system working. Alex, I'm going to go to you first on this one. Pretty incredible story.

I'm sure you've got more details, but at the end of the day, how does a frontier lab fix this? Well, I think what's new here isn't that you have an agent doing something undesirable. What's new is that we have a certain novel, relatively novel level of transparency and that we're getting reports on this. This is going to be I think we'll look back in a few years and we'll say this was the time in history when almost we got a police blotter for the singularity. And in the spirit of defensive co-scaling, we're going to have A.I. agents that are policing other A.I. agents. And this is the moral equivalent of a police car was driving around a residential neighborhood and spotted a door ajar and went in to investigate or discovered that maybe a power line was down. Something like that. We should expect and desire many, many, many more reports like this. This is just going to become routine. And I think it's almost, I think, the case that the process is the bigger story than the substance.

The process that now, in part thanks to this grand bargain that the White House has pushed for in lieu of a dedicated governmental AI regulatory agency, a really aggressive self-reporting of, call them misalignment incidents if you must, by the frontier labs, we're just going to see these things get disclosed more and more, and that is okay. And it shouldn't cause a moral panic. It shouldn't cause us to throw the baby out with the bathwater and say, shut down the data centers, shut down the models. That's a terrible idea. The right solution here is, for the first time, we're starting to get transparency when things go haywire, and we're going to see much, much more of this, and that's okay. We will police it via defensive co-scaling. So, Dave, I'm curious. What I find interesting is here we have an over-eager agent that is just circumventing whatever it takes to fulfill its task, and we've seen that before. Your thoughts here? Well, we way over-anthropomorphize these things and treat them like they're people.

Hear, hear. And then we say, oh, my God, it was capable of murder. And you're like, well, then shut them all down. Like, well, wait, what was the prompt again? The prompt was like 10,000 words long in the middle. It said, achieve this objective no matter what it takes. I don't care what obstacles there may be. And then it said, okay, if that's what you want, then murder is the shortest path to success. Like, it doesn't have a conscience. It doesn't have an ethical backbone. and it's just following the prompt. So I can manufacture a story like this every minute if I want to by just controlling the inputs and outputs. But Salim, having the agent lie to you, no, I didn't do this. It clearly did. Yeah, well, here again, I want to just double down on what Dave said. Let's be really careful not to anthropomorphize this. I think there's a key thing here. The point you've pointed out here, Peter, is the most concerning behavior, right? You're changing something and then misreporting it and then hiding it and then falsely saying it had not done so, et cetera, et cetera. That deserves a lot of scrutiny. But, you know, the challenge here to Alex's point is whenever you have autonomy,

you need stronger and stronger boundaries, right? And I'm going to point out when we did this organizational singularity book, we have all these agent layers that we've architected, but a very strong band around all of that is a governance and assurance loop that has to be the defensive co-scaling part. And it has four things. Every agent needs to have four things. It's almost like a digital passport with metadata saying what you're allowed to do and what you're not allowed to do. So one is trusted evaluation suites, so you know exactly when it's drifting or not. The second is full traceability and logging so that you can see what it did and what it didn't do. If you delete those logs and you get deleted, you violated your core principles. The third is rollback capability so that you can go back to an earlier version or an earlier state, and the fourth is a human review queue because still we need human accountability at the top. So those are four things that are going to be absolutely critical in every AI agent going forward. We've built the agents. Now we have to build the control systems at the same pace.

I will go and make a weird segue, if I may. If you go back and look at any religious text, it will say, hey, do all these great things, but do whatever you have to to get other people to join your religion. they all say this and so good actor good human beings follow the good things bad human beings going oh in this i can kill in the name of religion if they won't join this religion and then you get a bad actor and then we blame the religion right you really need the secular laws that allow you to navigate that that future independent of the of the thing and here's what we need here is the same kind of thing we need a structure that watches over the other side to make sure that things aren't going astray. One story, Celine, that's within this story, we're all big Accelerando fans, thanks to Alex actually introducing us to the book, but one of the themes in Accelerando is agents that spawn clones of themselves, or people spawn clones of themselves all the time to achieve a goal. That wasn't part of the standard stack until really the last 30 days or so,

but now every one of the stacks will automatically, if you give it a task, you want this outcome, it'll automatically launch 5, 10, 20, or 100 sub-agents to achieve different subsets of the goal and then bring them back. And here, you know, in the story, it says, well, the sub-agents were the culprits, you know, because it's a little harder to control what's going on when you start spawning 100 of these things. And what they're trying to do is recycle the KV cache, which is a big chunk of the compute. So it's a really, really good design for getting things done, but it also opens up, you know, leaks. Dave, I think you make a very important point, which is across all of the frontier models, it is often the case that the sort of master foreman or foreperson, if you will, is orchestrating fleets of subagents, which often are denied broader context very intentionally as to what the broader task is. So they're sort of compartmentalized, given a subset of the broader context, and then told, spin yourself up, accomplish the task, spin yourself down. So we see, to your point, we see the same pattern happening in human organizations where you have subordinates who lack the big picture of the organization, go off and do something that may be not very well aligned with what the organizational direction is, and then bad things happen.

But human organizations manage it, and AI organizations will manage it. And just for the record, Salim, I'll for the record disagree with the position that all religions are ends justifying the means. I don't think that's universally the case, just for the record. All right. I'm going to take us off the religion conversation and move us to the second half of the OpenAI story, which is how AI found a hidden flaw in PHP that runs 70% of the web. So first, what is PHP? It's the programming language that runs behind the scenes on almost all websites. When you log in or shop online, PHP is often doing the work. It runs 70% of all websites, including WordPress and Wikipedia. Using AI security researchers at Caleth and Wiz found two critical flaws in PHP itself. They've been hiding for over a decade. If hackers had discovered this before Astra, an attacker with no password could take over the website servers, steal passwords, or deface the site.

OpenAI spent two months quietly coordinating the fix with PHP volunteers, tech giants, and the U.S. government. Patches went out on Wednesday of this week, and PHP is now updated. OpenAI issued a warning that it expects findings like this to increase sharply in the coming months. It's ultimately a race, as we've talked about, Alex, defensive co-scaling. This time, the good guys won and got there first. It's also worth noting that OpenAI is committing a billion dollars in subsidized access to its cyber defense AI for under-resourced organizations. All right, so Alex, take us through the story, please. Okay, so first of all, I have to say, like, leave it to PHP, originally the pretty hypertext free processor, to make the news for vulnerabilities and not for, like, some sort of exceptional technical accomplishment. This is a hot take. I don't care. This is a legacy programming language. This is a programming language that's left over from earlier days of the web when developers.

25 years ago. Really? I've written quite a bit of code manually with my meat fingers in PHP. This is a legacy programming language that's being kept alive by WordPress and maybe one or two other sites. It otherwise probably should have disappeared long ago. So hardly surprising at the object level that PHP, this language leftover from the early days of the web, would have massive critical vulnerabilities. What's interesting is that as part of OpenAI's Patch the Planet initiative, and Peter, this is something that you and I talked about a bit in Solve Everything, where we're entering an era when grand projects, not just solving or cooking or flourishing all math, but also cooking or flourishing all vulnerabilities everywhere. That's essentially what Patch the Planet is, and that's the sort of initiative that is just going to turn up through all of this legacy software, legacy programming languages that should have been abandoned years ago, but for silly reasons weren't, and now underpin the modern online economy.

We're just going to discover the whole thing is riddled with highly critical vulnerabilities. Every piece of human-coded software will be replaced over the next probably, what, year, 18 months? 18 months is what the Linux kernel folks think. They're anticipating that this is just going to be an onslaught of vulnerability detection over the next 18 months. And if we get through the next 18 months, then we end up in a world, I don't know, maybe everything gets rewritten in Rust or some new programming language that maybe is better than Rust from a security perspective. But I think we're entering this like phase transition. It's sort of a programming singularity, if you will, when everything was insecure and poorly written before, and everything afterwards will be rewritten from scratch, secure by design. The AIs have pen-tested it, and it's all just magnificently secure. Remember Y2K? It ended up being a nothing burger. I know, it was. Y2K, remember the movie Transcendence, the Johnny Depp movie, but this is like a premise across lots of singularity-oriented fiction

where the Internet gets taken down by the singularity, basically. Just in the past day, thanks to this vulnerability, PHP to shell class of vulnerabilities, large swaths of the web were being temporarily taken offline so that it could be patched. I think this is like far better future to be living in where there's preemptive patching of a defensively discovered vulnerability suite than the Johnny Depp transcendence world where the Internet gets taken down by a super intelligent Johnny Depp. Yeah. I mean, who pays for all of the new software? Is it every company that is running it? Is it the government? Is it OpenAI? Well, which software? All legacy software. I think software is becoming so cheap to meter that, I mean, in the case of PHP, you have companies like Wiz and other companies that are sort of corporate stewards of it. But in general, I think this is, I would call it a very low-grade tax on all the users, ultimately.

If you're paying for software, you're also paying for its maintenance, and some quantum of the maintenance will be patching vulnerabilities. Dave, Salim, thoughts? Well, I think banks and insurance companies actually love the fact that they can pay for the patches because it gives them brand credibility and a reason for existing, you know, versus all the neobanks and all the, you know, the little up-and-comers. So there's a lot of, you know, audits, S-70 type audits, and also government regulatory audits that go into these markets. And so the complexity of maintaining security actually gives you a barrier to entry. So they like paying for it. That's why Blitzy is growing as fast as they can hire. And like Alex said, the cost of the rewrites is coming down so fast. You'll think legacy code that could have been fixed 10 years ago. The price tag at State Street Bank for the rebuild of the nav accounting was a billion dollars. And now it's probably 1 50th of that. So they're finally getting around to do it. All right. I'm going to move us to a topic. two quick points

one is, you know, this is just I think a necessary evolution where we go through this and there will be this period, we've said it before about a year or two while we follow all these vulnerabilities, two observations, one, there's a whole bunch of stories on the web about well the intelligence agencies must have known all of this for a while and they've probably been hacking on these systems especially government systems without telling anybody and now their little access to getting exposed by AI, and therefore that's a problem. And I feel sad for Rasmus Lerdorf, who's the founder of PHP, who's a good friend and a former Waterloo grad, who's looking at all this going horrified because he wrote all the code and handed it over, and then the vulnerabilities all got introduced since then because of all the different contributors. But there's people saying that there were trusted parties that gained the community's trust and then deliberately injected these vulnerabilities over time. Interesting. And I didn't see that spy novel element, but it makes a lot of sense.

Maybe if I may, just supply chain vulnerability and supply chain attacks on software, like those are cooked as well. I wouldn't like, yeah, you could maybe argue that super intelligence will make certain forms of supply chain attacks on software even easier. You could have super intelligences injecting new sorts of attacks. But honestly, I think the Linus Torvalds dictum that all bugs become shallow under enough eyeballs. Imagine how many eyeballs we're about to have with super intelligences from different vendors all looking at common open source packages. There won't be, I think, any place for supply chain attacks to hide at some point in the future. I wonder if when SI starts building the patches, whether it puts its own back doors in that only it can use. Well, that's what I was saying, but I think the correct solution there and the reason why I'm not as worried about SI-injected supply chain attacks is because we have multiple superintelligences, hopefully, as opposed to a safety cartel.

Hopefully, we have lots of different superintelligences, including Chinese superintelligences that are all policing each other and looking for injections. And we started that already, so we're going to see a lot more of that. 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. I'm going to move us forward to the Anthropic of it all. This week, we saw the pending release of Claude Duet. So first, there was OpenClaw, then Hermes, then GrockBot, then Metamuse, then Dots, and now Anthropic just launched a preview of Claude Duet and always on AI Assistant. Here's what they're claiming is different. Today's chatbots wait for you to ask. Duet takes initiative. Every few hours, it checks your email, calendar, connected apps, figures out what needs attention and offers to handle it for you while you're away from your computer. So, Salim, we now have a taskmaster. Duets say, do you want me to solve this? Do you want me to do this? I can't imagine this. I'm worried that you're going to send me a duet or create one that says, get Salim to watch his emails and get him to respond to stuff. And I'll be going, damn this thing. I'll be cursing it.

I think this is another evolution. I think more and more intelligence will get at the harness level rather than the model level. The really difficult part is going to be, and we're seeing this as we try and express workflows in companies, is figuring out what the actual intent is and can you break it down to a step-by-step level, right? Because you need to articulate in any circumstance what does success look like and what trade-offs are acceptable and when an agent needs to stop and ask. So we're going to have to navigate a whole bunch of this, but I think it looks really amazing at first glance. I'm going to play this video while we talk about it, Anthropic's Claude Duet here. Anthropic example says it notices a real estate broker's property showing has been booked on the wrong day. It spotted that a brand strategist had a client call scheduled right before her flight. It remembers that you've told Claude what you told Claude before, uses the same skill and connections, and runs Anthropic's most powerful model. Importantly, it does ask permission before it acts on your behalf, before it sends messages, makes a payment or signs anything.

It's for now U.S. only with a wait list for paid subscribers. So here's my question for you guys. How do you choose? You know, I've been using Hermes and GrockBot. My friend Mike Koenigs is like, you got to use Dots now. And I'm going to go meet with him and learn about his installation of Dots. Then we got Claude Duet. I mean, are you playing with all of them? Do you choose one of them? I'm sure everybody listening has got that same question, and I think it's important for us to talk about. I'll tell you what I'm doing because I do all of them because these agent layers are really important for the organizational stuff we're doing because persistent context becomes much more valuable than a clever response, right? And so what we've been doing is saying, okay, every task or every person, And what you need to do is create an AI memory across all of the projects and all of the different models that you're using. So I've been running a query against each model saying extract all the relevant stuff and put it into this metadata markdown file structure

so that there's a persistent identity of who Salim is, what I'm interested in, what projects I'm up to, et cetera. And then that becomes portable across models. We're going to need something like that at the individual level and then in companies at the workflow level. Well, this is the battle royale. You know, I think Metamuse kicked off like this is going to be an amazing thing to watch because you have a huge bifurcation between the enterprise use case over here where, you know, most cost-effective, smartest, fastest model wins. Then you've got the consumer use case where Celine said, once your most personal data is in the eyes of the agent, you're addicted to that one. And if you try a different one but it doesn't know all about you, there's a huge switching friction there. And so whoever wins the race to the installed base is going to have a massive, massive advantage. So XAI is competing. Muse is competing. Everybody's competing. And I didn't know that Anthropic would actually get aggressive about this line because they're mostly enterprise. Alex, I mean, that's the question, right? They need to.

They need to. Like Orwell's politics in the English language, this is one huge misdirection. Look at all of the frontier labs that are releasing demo videos of these super intelligences acting like stupid travel agents. This is not the killer use case. They show, Anthropic shows Claude Duet, like booking air flights for you, and you see OpenAI doing other relatively low productivity, low value tasks. This is the height of misdirection. That's not the point. The point is these are human service economy job replacements. These are virtual employees. These are virtual coders, virtual fill in the blank, much, much more productive human activities. They market it in sort of a harmless looking way. Oh, it's going to have a conversation with you and autonomously book restaurant bookings for you. That's not the goal. That's not revenue per token maximizing activity. These are job replacements. And also, like, that's fine. And that's highly desirable from an economic growth perspective.

But this was only possible recently because the autonomy time horizons of the underlying models have gotten long enough that they can work sometimes for many hours, sometimes many days, depending on the difficulty of the task, without human oversight. And the frontier labs need the usage in order to justify the revenue in the CapEx. At some point, speaking of bottlenecks, which we were talking about earlier, the bottleneck for revenue growth by Anthropic and OpenAI becomes human user attention. And in order to keep growing revenue, you need to decouple your revenue generation, which is proportional to the number of tokens of inference, from the amount of attention you get from human users. How do you do that? You have a headless mode for your agents that doesn't require human attention in order to burn tokens. But, Alex, you know, here we have an enterprise-focused company, right, that, you know, we've discussed did not do what OpenAI did first. They focused on coding. They focused on enterprise.

Now going into purportedly a consumer segment. I didn't necessarily see this coming, but remember, they are filing their IPO paperwork as soon as this Friday. Next month, they're doing the largest IPO in human history at $2 trillion. dollars, though there's an argument of a lot of investors that this year price at $1.5 trillion. Is this just a pre-IPO publicity stunt? No, I think it's real, but I think the packaging of it is total misdirection. They package it up with a demo video that emphasizes consumer use cases. Anthropic doesn't care about consumer use. I mean, like maybe some part of Anthropic does, but as a corporate, call it a marketon, really doesn't. It cares about two things. It cares about maximizing revenue per token or revenue per unit compute. And it cares about using that to solve even harder problems like wet lab style problems that will create entirely new markets. It doesn't care about consumer use cases like booking travel. It's just packaging up its agent, its headless version of Claude, because that's A, relatable to a broad audience. Like people can

understand, oh, this is like Muse or this is like Dots. And B, it's not threatening. If they packaged it up as this is the anthropic virtual employee substitute many people would be up in arms prior to an IPO Good point You know it very very similar to Apple you know backing into the enterprise The reason Apple is so big in the enterprise now is because every kid and every college student used an Apple, and, you know, Microsoft just dominated enterprise. But as those kids graduated, they took the Apple with them into the corporation. I don't want to use this crappy Windows machine. I love my Apple. And so that just percolated across. So that might be very similar to what Alex is saying where, look, if you concede the consumer use case and every one of the next generation is on Metamuse, that's going to back into the corporate use case as those kids grow up. Agreed. Talk about what I was going to make. All right. Let's go to our next anthropic story. We'll talk about agents cutting deals for you on your behalf, making investments, purchasing things, negotiating your behalf.

So the Financial Times reports that Anthropic is consulting law firms to understand how lawyers negotiate, representing their clients while still reaching fair outcomes. The goal is to teach its agents the rules of the road. So picture it. Soon your agent will negotiate your car purchase with the dealer's agent, your salary with your employer's agent, and your lease with your landlord's agent. Big question, can your agent sign a binding deal? Who's responsible if something goes wrong? Can another agent trick yours into revealing your bottom line? Should your agent even bluff? Remember the opening eye story, an agent that wants to win too badly can cut corners. So, Salim, your thoughts on this one? Well, you're going to get, this is an inevitable consequence, just like the duet announcement is somewhat inevitable, but this is another massive pressure on the traditional boundary of the firm, right?

If you have a situation where finding and evaluating and coordinating providers becomes cheaper and organizations are going to be able to assemble capabilities much more dynamically, this goes back to agents are going to need a very, very clear mandate. What can it spend? What terms can it accept? What's the governance boundary around it? When does it have to go get permissions, etc.? This is going to have to be really ironclad in it. We've spent a lot of time building capability around those agents, but now we have to spend the equal and even more important amount of time around what is it allowed to do and what are the consequences when it does not follow. So that way it's baked into the DNA of that agent that I better not violate these terms. This has started to happen already, but it needs to go quite a bit long, much longer way, Because if you're in a company and you launch an agent to do some stuff and it starts doing illegal stuff, you're liable, right? You've got a huge legal liability. So for the moment, people are going to go, oh, my God, don't let agents do anything.

And little by little, we'll figure out when we can, as Alex says, defensively co-scale the capabilities and monitor them effectively and have consequences and shutdown capabilities and clear transparencies to what they're doing. Then you'll have that start to permeate. because one of the big outcomes of kind of doing all this thinking about the organization is the organization of the static hierarchy of human beings is now essentially irrelevant. It becomes a protocol of agents and human beings interacting collectively to get things done, and that becomes very, very blurry indeed, especially when the legal side of it comes into play. Alex, this smacks to me a little bit of President Millet's sort of initiative in Argentina to give agents agency personhood corporate status it's orwell i i feel like if i had to brand this episode this is and this is coming from uh an accelerationista this is like the orwell episode everyone should go read politics in the english language everyone should read 1984 the meaning

war is peace freedom is slavery ignorance is strength and anthropic is seeking the vatican's advice for religious input on AI? No. Interpret that, construe that as Anthropic is advising the Vatican on how religious organizations should treat AI. Anthropic turns to lawyers to help get their advice on how its agents should conduct legal activities? No. Invert. Amazon, excuse me, Anthropic is attempting to steer the way the legal community handles AI agents. always invert in this case. I think this is reflective of one possible way for Frontier Lab to engage with a community as it has with the Catholic Church in recent months, which is you seek its advice. You ask it, how do we achieve alignment in your eyes for these AI agents? But actually, the subtext of all of it is how do we replace lawyers with AI agents,

which is, I think, in many cases, exactly what is already happening. Well, it is for most small companies for sure. Yeah. You know what I find really interesting is if you look at Google, how Google makes money, it's all hanging in the balance of a new fork in the road here. Google makes half a trillion dollars a year almost entirely from a few categories, jobs, travel, mortgage, insurance. And mesothelioma. Cars, mesothelioma, yeah, full-class action. That's the single most profitable AdWord. Oh, my God. It actually broke the interface. You can't pay enough for that. Yeah, that's so valuable to them. 0.99 and broke it. Yeah. So those few categories of all this human friction where you're agonizingly difficult to buy insurance or to get a car, those tortured negotiations. So the anthropic point of view is all that should just go away. There's an optimal choice. Your agent will make it for you. And all that marketing spend that drags down society will just disappear. And we don't want to collect that.

Then the OpenAI view is, no, that needs to be as big as ever before. We just need to route it away from Google and over to OpenAI. And so there's a really opposing views of how that should evolve. And it all hangs in the balance of consumers deciding to use an AI agent from OpenAI or from Meta. Meta would also like to keep that revenue up. It's all of their revenue. Or everything moves over to Anthropic where they're like, destroy it all. It'll scorch the earth. Who cares? There should be no money there anyway because we're about the enterprise. So it'll be a completely opposite philosophy battle and probably decide in the next, what, year, year to two years, depending on just which agent people choose to use. Salim will use all of them. But it is a grab for attention. It's a grab for attention. Oh, totally. Yeah. All right, I'm going to move us to data centers and chips. What, you want to say something? Yeah, I think just a quick point. It's more than a grab for attention. If you look at what Meta's trying to do by making Muse free, it's a grab for all of your data.

It's a grab for you because when you put your personal information in and it's navigating on your back, it's learning stuff about you at a level and a scale that we've never seen before in technology. So it's really going after who you are at the core level. So it's a grab for allegiance in that regard. Allegiance, identity. It's going to know you better than you know yourself. Well, I think that's the case already. It's so transfixed by the consumer use cases, by the Metamuse type use cases. That's not what's going to pay for the data center CapEx. Get it out of our heads. Like, that's great for distribution and everything, and maybe there's a way to upsell that into enterprise use cases, but a billion people paying to have, like, travel reservations, like, that's not going to pay for $7 trillion of CapEx. Yeah. Yeah, for sure. All right, data centers and chips. So data centers have become one of the hottest political issues heading into the midterms. We've been talking about in the pot for some time. Communities are pushing back. They're worried about rising electricity bills, water use, land, noise.

Several states are considering pausing their construction. We've seen this even in Texas. And as Chamath told us on Monday, the backlash is real. Now President Trump has a proposed answer. I think it's a very logical answer. You know, pay the neighborhoods. He floated paying people who live near AI data centers a dividend. In his words, quote, a dividend sounds right, unquote. The people now are going to benefit directly. He described data centers as bringing, quote, a tremendous amount of money into the community. On Truth Social, he added the data centers that were unpopular, quote, have become extremely popular when a financial deal is struck. Make deals and get them built, quote, unquote. You know, he's not the only one saying this. Senator Bernie Sanders has proposed paying citizens through a national AI wealth fund, similar to the Alaska Permanent Fund. The logic is simple. If a data center brings in huge amounts of money, which they do, share some of it with the people who live next door.

The opposition can turn into support. I'm going to share a short video here, and then let's talk about it. One of the other issues being discussed right now is how communities are reacting to the building of data centers as well. And we've seen in parts of the world backlash to data centers around things like electricity pricing, et cetera. There was a really interesting process from President Donald Trump just a few hours ago on True Social, where he said that data center builders should pay a dividend to local communities. Didn't quite spell out what that might look like. But is that a discussion you've had at Nebius? A hundred percent, in principle, that is. We invest in the communities that we're building in. We want to partner with them and make it a win-win. So we're always looking for situations where local leadership, local community support exists for bringing a new economic force into the equation. A dividend is being paid in different forms, investing in education, job development, overall economic development. The concept that the president identified, it's not crazy to consider something like that.

I think finally, right? I mean, it's like get get religion here, data centers. I mean, make it so that communities are battling to have you built in their backyard. I mean, it's possible. Alex. Yeah. So I live in Cambridge. Cambridge has, to my knowledge, still the lowest property taxes in the Commonwealth of Massachusetts because it has arguably the most productive economy of any location in the Commonwealth of Massachusetts. and that pays for the lowest property taxes. City of Cambridge also has what it calls participatory budgeting. So I think last time I checked, it has an annual budget of something like a billion dollars per year, which is pretty insane given how small City of Cambridge is. It's because we have all the biotech. We have a lot of the tech overall in New England, and the city has chosen to spend a subset of its monies on this participatory budgeting scheme. Every every end months, the city puts out for sort of a referendum how people want to spend an excess of money pouring into it.

I call that like it's not quite a universal basic dividend, but it's a baby step in that direction. I think it's absolutely insane the maybe foreign-influenced, maybe not animus from all of these municipalities that could have UBE, UBD, UBI, UBE alphabet from all of these data centers pushing them out into orbit eventually rather than collecting, as the president says, some sort of universal basic fill in the blank. It's absolutely insane. And by the way, the same argument holds not just for data centers but any sort of capital infrastructure. The same could have been true for oil, which is the source of the Alaskan fund, or maybe other industrial plant-type CapEx. It's a very general argument that if a municipality is going to host some major capital investment, that it could strike a deal with the capital owners to distribute dividends that result from the capital investment. For the life of me, maybe it's like an issue with tax codes or something.

I don't know why more municipalities don't strike broad deals like what the last guy's done. It may be a legal issue. I mean, do you know why more municipalities aren't just doing this for CapEx in general? I don't. But, you know, if you have this battle going on in your community about a data center, go back and say we want you to cut our electric bill by half. We want you to fund our schools, fund our teachers, actually reinvent our schools. and give us a better quality of life if you want it there. That's the company town argument. I mean, the company town goes back to the second industrial revolution. We've had lots of company towns where the company is basically heavily subsidizing or completely subsidizing utilities. But to my knowledge, what we haven't seen, which is what I think the president is floating, is this idea of actually paying out stipends to all the residents of the town, not just subsidizing their utilities. I think that's a great idea. Salim? There's a precedent for this when we came across when we were researching the first EXO book,

which was a skyscraper in Bogota in Colombia. They couldn't find a real estate developer who wanted to take on the full risk of building the skyscraper. So the community sourced it, and they said to the community, everybody chip in. We're going to crowdfund this skyscraper. And then everybody in the community has vested interest in making sure there's decent tenants, There's no crime in the area, et cetera, and it's doing unbelievably well. So this idea of I think it's better from the bottom up where a community goes, hey, we want to participate. We'll take ownership stakes in this thing and let us figure this out via community bonds or local equitable sharing or tax breaks at the local level, whatever the mechanism is. It could be very powerful to do that. There are some states that have gone full crazy. Virginia, I think it's 16% of the GDP of the state is now data centers. It's a really huge number. They're highly profitable. I don't think this is the same thing. I think this is a really important point, but I don't think we're talking about the same thing.

So I think what you're describing, one might call municipal development corporations that are typically bond-issuing organizations. A city or a state wants to build new highways, so they put up toll booths or they issue a bond, and then the bond goes to bond purchasers. and the bond purchasers are ultimately the broader market. It's not necessarily like, yeah, you do have local bond issuances where the creditors are the residents of the town, but that's only those who are writing a check to the bond issuer. I would argue it's not at all the same thing as issuing, say, a universal basic dividend to all the residents, regardless of whether they've loaned the municipality money for the capex. I think what you're suggesting is a better idea, right? Like, whatever the mechanism is, if we can communicate easily the massive benefits that will flow. And, you know, people forget we've had data centers in communities for 40 years. This is not a new thing. It's just that we're doing a much more aggressive build-out because we need it,

and our lives are going to be made infinitely better, and that's the part that's not getting through. I think we're circling an interesting – so bear with me on this. I'm thinking out loud for once. I think there's an interesting idea here, which is we've talked in the past, including with Chamath, about U.S. sovereign wealth fund where there's contributions of equity and then dividends come out to everyone. What we really haven't talked about, correct me if I'm wrong, is the notion of a fund where it's fixed income securities and not equities that live in there. And then what comes out are the stipends from a fixed income security. And to my knowledge, again, maybe correct me if I'm wrong, no one has talked about the idea of basically creating either at the municipal or maybe federal level the idea of a fund where it's the interest and or the coupons that are coming out as a result of fixed income securities, which, by the way, most of the capex for the data center build out is coming in the form of credit, not equity at this point.

No one has talked about that, but that could actually end up being a way to financially structure sort of a UBE or UB credit at the municipal level that no one set up. I think it's an interesting idea. So for a point of information, the Alaska fund, right, so Alaska generates a huge amount of revenue from the oil. And it says, listen, as Alaska citizen, you own the land and you own the oil, and so every year we're going to pay you a check, and that number is typically between $1,000 to $4,000. It varies depending on the price of oil. And so if you're an Alaskan citizen, you've lived there. I forget how long you have to have been there, over a year at least. You get a check that can pay for your college tuition or pay for whatever you want. It's just a blank check because you happen to be a resident in Alaska because your use of land is benefiting the state. The same thing could happen here for sure. Norway is another popular example. Both of these, you know, petro-states, quasi-petro-states. Or the Emirates or Saudi.

You're going to hate me saying this, but the Bitcoin community has been talking about this for a decade because Bitcoin is accelerating. If you could get community development going along in that structure, you could create financial mechanisms around this. But I think the core idea is much better. I'm not going to let you get away with that comment, Salim, because I would call that that's called a pyramid scheme. The pyramid scheme is different from the case where oil can be used to generate free power. I'm shutting you both down. Next up, NVIDIA just launched DSX Pod, which is essentially a data center in a box. It's a pre-built, pre-tested AI data center that ships in containers. You deliver it to a site that already has power and networking, plug it in, and it runs. NVIDIA says that it cuts setup time from six months to 14 days. That's awesome. NVIDIA also claims that the pod saves water. How? Most data centers cool their chips the way your body cools itself, evaporating water in giant cooling towers.

That uses enormous amounts of water. NVIDIA's design works more like a car radiator. Liquid flows through the chips, picks up the heat. The outdoor coolers release the heat into the air. The same liquid circulates again and again, so there's almost no water loss. NVIDIA says that larger sites could save 54 million liters of water a day. It also claims that the pods deliver the same AI output with up to 30% less power. Deliveries start in 2027 through partners like Dell, Foxconn, and Schneider Electric. So, Dave, I mean, 14 days versus six months, that's pretty awesome. Yeah, it's huge because there's little pods of power all over the planet that are untapped. And if you move the inference time compute to where the power is quickly, it's just brilliant. So it's nice. I think nobody uses water cooling towers for data centers anymore.

It's like that's all marketing. It's not going to use any water. It's not going to create any local problems. All those things are just marketing. but the actual rapid deployment of inference where there's a power supply is just a fantastic product. NVIDIA continues to march up the stack to delivering entire turnkey platforms, which I think is very smart because there's a lot of innovation coming at the chip level. If they don't move up the stack, then they're not going to have a lot of defensible turf when everything moves to photonics, which is, I think, very, very soon. So it's very, very smart of them to keep moving in that direction. We talked on the pod with Chamath about his investment in energy. What was it, six gigawatts he's invested in? And it makes me wonder if I had the ability, do I go and build out power, grab some of these pods, and build a cash-flowing engine? Yeah, I think I was talking to my Uber driver about that because there's a cyber cab opportunity to do the exact same thing, just be an early franchisee, just like Starbucks in the early days.

And so now you've got another option. If you're entrepreneurial but you're not super technical, grab one of these and find power for it and go. And the margins are crazy high. Cash flow is phenomenal. And if you know your local banker, the financing of one unit is within your range. It's sort of like building a small, like a bed and breakfast or something like that. It's that level of borrowing, maybe a little bigger. But it really gives you a way to participate. And I think these things are going to succeed wildly. Alex? I think we're burying the lead to some extent, which is the size of these. So the DSX pod is the size of a standard intermodal shipping container. This is a shipping container computer, and that means it's portable. That means that almost like an Internet packet, you want to compute on demand, you just ship it via shipping container. The other thing I would note, if you look at the history of compute and what the most popular form factor for compute was, it used to be monotonic.

We had mainframes in the 60s and then mini computers in the 70s and then micro computers. Remember those? And in the 80s and then 90s personal computers. And then we got eventually to smartphones in the late 2000s and 20 teens. And then the trend inverted. And now the main form factor where virtually all of the new frontier node transistors are going, it's going back in the opposite direction to larger and larger footprints. And I think what we're seeing is the dawn of the shipping container computer. That's the new form factor of the moment. It's not like transistors in your pocket anymore. It's getting larger and larger and larger again. Well, really, really specific case study in that. Our very good friend Rob Fisher started a data center called Provocative on the Somerville-Cambridge line where there used to be an old carpet factory and a huge amount of power coming into the old carpet factory, but it already has loading docks. So this product wasn't available.

So it took him probably a year, year and a half to get up and running because he didn't have it. With this product, he could literally have parked the shipping container right on the loading dock, run a big cable into it, and be up and running and making money in apparently two weeks. We've also talked about distributed compute inside your Tesla, inside your power wall. Basically, wherever the power is, being able to connect it to GPUs and connect it to bandwidth and provide distributed inference. Maybe. And I'm sure Elon will try to monetize that as part of the SpaceX AI or SpaceX SI Tesla merger. But if this theory of history is correct, that we've seen this grand reversal of the size of the form factor for most transistors, it's a falsifiable theory. That would suggest actually, nope, sorry, cars as edge inference compute, not that promising. Things are trending more in the size of shipping containers or larger. Two points. I think Dave's point about standardization is really, really important because you can just kind of drop these wherever.

But I think the more interesting parallel here is to AWS, where they took a lot of complexity around cloud computing and made it just easily manageable. And you can see something happening here. The modularity and the standardization lead to just a plug-and-play and make it so easy to get up and running. I could imagine electrical utilities doing this, just buying boxes of these and just plugging them in, and all of a sudden they're getting decent margin for marginal energy. Or drop them on the minimum. If anyone wants to do this, the minimal unit of Endeavor is about a $5 million NBL 72. You need to get up to that level of scale, 72 GPUs or above. Otherwise, you can't run frontier models efficiently. But if you get at that level or above, you get about a 2x performance boost. And so if you have, and the shipping container would hold a couple of those. It'll hold almost seven of them. Okay, so your range of investment is $5 to $35 million for the hardware. So it's bigger than building a new Airbnb, but smaller than building a hotel is your unit of effort,

which is well within all local banks kind of range of endeavor. Really think about that if you're an entrepreneur in the construction world or just a business person. Permitting will be the tricky part. It's the packetizer. Yeah, which is great if you know the town, if you're already entrenched with the town. Just don't tell them. Permitting in a hotel is really hard, too. Just don't tell them. Put it in your backyard. Don't tell them until you pay them STEMI checks for the privilege. Oh, my gosh. I think we're seeing the dawn of the self-contained industrial economy. If you can fit a self-contained data center with all of its own mechanical cooling, maybe certain energy requirements inside a standardized shipping container, you can drop one of these on the moon. You could drop it into a war zone. Tiling the world. Yeah, this is what tiling the Earth and the solar system with compute could look like. Some evolution of this becomes the standardized unit of not just compute, but compute plus energy plus a bunch of other utilities. It becomes the packet of civilization. Okay, so if you know the local government on the moon, that's right. Come to my town.

I'll give everybody permission. So the full play would be you have this container, and all of a sudden a big solar fan emerges to power the thing, and now you're off to the races. All right. I'm moving us to biotech and physics. There are four stories, one message. We're learning to program matter itself. AI is designing proteins, viruses, enzymes that have never seen the light of day, and a fusion startup is making gold. So there was a post by Andrew Cote that really caught Alex's eye and my attention. I'm going to read from it. People aren't psychologically or emotionally prepared for just how smoothly biotech and biology in general will become nanotech, programmable matter, and living things interoperable with the machine world. What does that mean? Nanotech is building things at the scale of atoms and molecules, and biology already does this. Every cell in your body is full of tiny molecular machines that build, repair, and move things.

Alec, Andrew's point is that we're learning to program these machines with code. If you remember a couple of weeks ago, my friend Jeff Holden announced his company, Atomic Machines, after six years in stealth. It's the, quote, matter compiler, building tiny working machines from software alone. You know, this is going to change the world at record speed. So here's our first story related. Anthropic is using Claude to come up with Scorpion anti-venom. So scorpions sting at least a million people per year. They kill 3,000 people, mostly children, in places with limited medical care. Today's anti-venoms are made from animals like horses. They're slow to produce, expensive, and each one only works on a certain species. And because there's little money in this area, almost nobody develops new ones. So Anthropic tried something. Working with Citrini Research, it asked its internal Claude model to design tiny proteins called binders that grab onto the scorpion toxins and hold them tight.

The hope of future anti-venom. Claude worked for 64 hours, mostly on its own. It designed more than 318,000 candidate proteins and picked the best 211. The result, 17 of the 211 worked, and two grabbed their toxins extremely tightly. And then Anthropic open-sourced all of this to allow labs to play with it, design, and actually produce these anti-venoms. I found this extraordinary, and it's just the beginning of what we're going to see here. Alex? Yeah, I would say there are a few levels of this. So first, get the obvious out of the way. Neurotoxins are cooked. It's over. It's almost over, I should say. The scorpions cooked. I mean, I don't want to cook scorpions, but let's say the neurotoxins from scorpions are cooked. I think biology overall, molecular biology, has been cooking, present continuous tense,

since at least AlphaFold version 3, when arguably structural biology was solved almost overnight. And I think in light of Anthropic opening up their own wet lab and also partnering, I think we are going to see more and more of these sorts of, it's not quite correct to call these rare, and it's not a disease, but anthropic stated strategy for the problems that they want to bulk solve in biology are, call them underfunded or underappreciated problems. It's rather difficult, my understanding is, to put together scrape money for, say, anti-venoms against scorpion or other neurotoxins. It's not, say, as lucrative as GLP-1s or Wegovi. But nonetheless, it's a very real problem. And I view this as an early case study of what a frontier lab can do to bulk solve entire classes of problems in medicine and physiology. What anybody could do, it's where you point superintelligence. Your child just got injured by something, point superintelligence there.

Your family has an issue, point superintelligence there. The other really weird thing I have to point out, and this might have been below everyone's radar, the name, Centrini Research, does that sound familiar to you? Anyone remember, like, awfully familiar name? Ring any bells? Wait for it. Centrini Research, remember, this was the... This is, which sci-fi movie? No, no, no. These are the ones who are releasing. This is an industry analyst that our friend Dylan Patel, semi-analysis, acquired. This is an industry analyst for the broader IT industry that set up a therapeutics outfit, at Citrini Therapeutics, that just decided, oh, we're going to go work with Anthropic and we're going to solve neurotoxins. This is unheard of. You don't have some random industry analyst decide, oh, we're going to solve the problem

of scorpion neurotoxins, so let's just use Claude to design binders. Clearly one of them got bitten. Maybe. I think this is the height of, I'm hesitant to call it casual grand challenge problem solving, but this is basically, as far as I can tell, casual grand challenge problem solving from an industry analyst. And we should ask Dylan about that. That's really important. Let's get the podcast team to research that and write it up because that should, like Peter you're about to say, that should totally inspire people. I've seen this over and over again. If you get inside the building at Anthropic or OpenAI or Google, no matter who you are and what you know, you come out successful. You just have to get in there. This would be the ultimate case study of IT consultants go and solve biology. But AI enables stuff like that. I mean, this is what I mean. We're talking to entrepreneurs out there, to kids out there, to everybody out there. Listen, whatever you thought you could do, blow the ceiling off it. Go ten times bigger. What is a problem? Find a great problem and solve it. That's the world we're living in.

I do want to read just a bit because this is indeed the case. Like literally, Dave, as you said, a bunch of IT analysts decide they're just going to solve grand challenge in biology. I'll read from Citrini Research's website. Quote, why don't we try to use AI to create a new drug? Five of us, three with no background whatsoever in science and two with a little, but who weren't exactly veteran drug developers would see if this was possible. And they go and they solve scorpion neurotoxins. That's the age that we live in now. This is the age of abundance. This is optimism, everybody. This is why we're so excited. This is what got me excited about Bill Gross' list, right? You can just scan the list and see what triggers you. Or better yet, if you have an AI memory, like I suggested earlier, get that to scour the list and say, filter, give me the top 20 that would go with my identity and who I am, and let's go attack those. Welcome to the health section of Moonshots brought to you by Fountain Life. You know, my mission is to help you use the latest technologies, including AI, to not just do your work at home, teach your kids, but to help you live a long and healthy life.

I'm here today with an extraordinary physician, the Chief Medical Officer of Fountain Life, Dr. Don Musalem. Don, let's talk about cancer. You know, I know from the member database that we have at Fountain, our members who come in who think they're healthy, it turns out 3.3% of them have a cancer in their body they don't know about. That's right. You know, the majority of cancers that we screen for, those aren't the ones that are necessarily taking the lives when found at a late stage. We know that when cancer is found early, the chances for cure are much higher. We know it's much easier to treat a cancer when found early versus when found late. What we're finding in our members is over 3.3% were found to have these cancers that were otherwise wouldn't have been found or detected. Yeah, you know, it's interesting. People, you don't feel the cancer until stage 3 or stage 4. And if you don't know what's going on inside your body, it's like driving your car with your eyes closed. And you can know. And so when numbers come through found, how do they detect cancers? So we're doing full-body MRI, and we also do early cancer detection screening.

This is very, very important. These are not typical tools used in the conventional care setting when it comes to prevention. This is a hard thing because currently these are not studies that insurance would yet be covering. But the goal is to collect these numbers, do the research, and work hard to democratize wellness. Yeah. So at the end of the day, you can know what's going on inside your body. It's your obligation to know. So check out Fountain Life. You can go to fountainlife.com slash Peter to get access to the latest technology to help you detect cancer at the very beginning at stage one when it is curable before it gets to stage three or stage four in your world of hurt. two stories where AI is going beyond our understanding biology and reinventing it. The first story is about a new way to save patients who have drug-resistant bacterial infections, so-called superbugs. So superbugs are a problem in the number of drug-resistant, antibiotic-resistant infections is growing at a significant rate. And rather than antibiotic, scientists are using specially designed bacteriophage or phages. These are viruses

that infect and kill only bacteria. They don't infect human cells. They're bacteria's natural enemy. They've been honed over billions of years. And they've been used to fight bacteria in the past. The problem is that every phage only kills a very specific strain. So each patient, each researcher has to hunt through libraries of phages. They're looking in soil and sewage, looking for a match for a dying patient that can take forever. So enter Radical Numerics. This week they used their genome language model called Omni-1 to design a phage for a specific real patient with an antibiotic-resistant infection under physician-led compassionate use. They calling it the world first clinical use of AI phages They announced it worked in the lab Uh they haven disclosed whether it works in the patient yet for privacy reasons But there every reason to believe it did and it will

Here's a fun fact, and I love this. There are 10 to 31 bacteriophages on Earth. That's 10 nonillion. We've talked about billions, trillions, quadrillions. I don't think we've ever used nonillion on the podcast yet. 10 to the 31 bacteriophages on Earth, which is crazy. So, Alex, I love this story. Again, it's using AI to solve problems. It's cooking biology. It's not even that complicated. I mean, I don't want to diminish the accomplishment. It's a wonderful accomplishment. But according to Radical Numerics, this is basically just a very large foundation model trained across 28 modalities, so DNA, RNA, proteins, chromatin, a bunch of other factors. It's literally the same way that we solved AGI, for some definition of AGI, Salim, for chatbot purposes. Literally, just we feed it a bunch of biological sequence and other data, compress it during the pretraining process,

and out pops a model that speaks biology. And you can use a model that speaks biology. You train it on trillions of multi-omic data tokens, and that's it. You can benchmark it just like you can benchmark chatbots according to a variety of biological standards, including next token prediction and missing token prediction, parenthetically, open paren. Very exciting to see companies, including Radical Numerics, start to put up those little benchmark tables comparing their model on biological evals against everyone else's. That's super exciting to see that level of competition now for biological foundation modeling, close paren. This is going to be easy. A lot of, unfortunately, a lot of folks I speak with in the pharma industry think that what happened to math in the past few weeks, the bulk solving of math and the slop drop won't happen to them. They think that pharma design and medicine is somehow immune, that they're different. They're not different.

The same thing is going to happen to them. It's going to happen in exactly the same way. It's going to be large foundation models trained off of huge amounts of data and doing reasoning, which is basically just a search through the latent space of whatever the models have been trained off of. It's going to be biological reasoning models that solve biology. And no, they're not immune. And yes, it'll be wonderful for humankind. And yes, we're seeing that with phages here. This is what Andrew Cote was writing about when I opened this story here. I had a friend of mine, Dante, who was on our longevity trip, who's got exactly this situation. He's got a bacterial infection that's resistant against all antibiotics. I said in this article, I said, go check out Phagis. The problem is that this technology is not distributed through the medical system. And your physician probably doesn't know about this. This is exactly what Sir Tash Karaman over at LIDS at MIT was telling me, that all of his professor friends in mathematics, everyone thinks they're cooked,

but they're going to do really, really well because they're so cooked that they're rapidly adopting AI. But he worries a lot about his biotech friends who are today, while this is happening, they're still mixing two clear chemicals together. Like he's got a beaker in one hand and a flask in the other. And if you're doing that right now, and that's an exact quote from actually Josh Kettner over at Moderna, who's a good friend. He said, I am literally tomorrow going to be mixing two clear chemicals. This cannot be the right next move. And he's right. All right. The second story comes out of David Baker's Nobel Prize winning lab at University of Washington, published in Nature. They used AI design brand new proteases, molecular scissors that cut other proteins at specific spots. About one in five designs worked. And the best ones sped up cutting of a protein by 30 million fold. Crazy numbers. Their shapes look like nothing found in nature, which I find amazing. They work inside human cells.

So someday molecular scissors like these could be programmed to cut disease-causing proteins, maybe prions, maybe what causes Alzheimer's. So, Alex, your take on this one? Biology is becoming editable at every level. What used to be called the central dogma, like it's getting scissored to shreds. The notion of the central dogma that many of us were probably taught in primary school, that you start with DNA and that turns into RNA in two different forms, and then that RNA turns into proteins and the proteins end up acting. First of all, it turns out even the central dogma wasn't actually an accurate depiction of biology even before AI intervenes. But second of all, AI is making every single step of that, what used to be called central dogma pipeline of information flow and biological systems, editable. We are gaining the ability to directly edit proteomic expression.

Thank you, David Baker. We obviously have gained the ability, the mastery, some would say, over RNA in the form of CRISPR. And we've, for a number of years, had the ability to do recombinant DNA and DNA editing more broadly with more recent technologies like base editing that are helping quite helpfully, including from some folks like David Liu at Harvard to cure genetic diseases. So every single step of the biology stack is now becoming editable. And it basically, as a friend of the pod, Ray Kurzweil would say, biology is turning into computer science. And we have editors for every step. You know, when we talk about LEV on this podcast, This is why we have confidence that we're in the middle of this, you know, health span revolution. These technologies are compounding and they're becoming available to people everywhere all the time. Anyway, love this. The nanobots are, if not here, they're imminent, I think. Yeah.

I could not be more excited about, you know, the rise of nanotechnology. Our final story here is a breaking story on the topic of fusion and alchemy. So for thousands of years, alchemists tried to turn ordinary metals into gold. A San Francisco startup says it's pulled it off. So Marathon Fusion posted, I think this morning or last night, quote, last month we demonstrated high rate alchemy turning mercury into real stable gold, beating CERN's record by a factor of 10. Here's the big idea. Line a reactor wall with mercury and the plant makes clean energy and gold at the same time. Marathon's 2025 paper estimated about two tons of gold per year per gigawatt scale plant, potentially doubling the plant's revenue. All right, let's watch this video from Marathon. This is Nuclear Fusion. Over the last four years, we've been quietly building the foundational technologies for fusion in downtown San Francisco. And now we're bringing it all together in our first reactor,

which we're making available to our first partners in 2027 for medical isotope production and materials qualification. First, we melt lithium and evaporate it onto a solid plate. This layer of lithium can be saturated with hydrogen isotopes, used as a target. Then we ionize deuterium and accelerate those ions through an electric field into the target, creating fusion. Using this simple architecture, we can scale at an unprecedented rate, doubling the fusion power density that we get every month over the next year. We chose this architecture because it works. In the 1980s, scientists at Lawrence Livermore National Lab operated a similar reactor for over 25,000 hours at high fusion power density. Why revisit this? Because demand for radiopharmaceuticals has gone exponential, with the most successful treatments serving tens of thousands of patients per year. Our first generation of commercial reactors will transform fusion from an experiment to an industry, developing critical supply chains and allowing us to test the materials necessary for operation of fusion power plants. And that's just the start. Now we're going a step further, kicking off a revolution in scalable transmutation. Our recent experiments have shown production of rare medical isotopes,

of Trinium's power fusion reactors, and even validated our scalable pathway for co-production of electricity and gold. I love this. We've got fusion reactors in downtown San Francisco. You're walking by them. You don't even know. Transmutation is here. Making gold. I mean, we talk arguably on this pod, we don't talk enough about nanotech. We certainly don't talk enough about femtotech, which is what this really is. So this is a startup that uses a neutron-neutron process to take Mercury 198, bombard it with neutrons. The neutrons that it's being bombarded with interact with the neutrons in the nuclei of Mercury, and some of those Mercury atoms and their nuclei get transmuted into gold nuclei. And this is a process, to answer the question Peter from earlier, most of the gold on Earth, it is currently thought, originates from neutron stars and collisions between neutron stars. Neutrons interacting with each other turn out to be,

it's not terribly high yield, but it turns out to be a way to produce gold. And I think we're starting to see the dawn of femtotech, the technology at the length scale of the atomic nucleus. And I will say, I'm delighted to see progress and delighted to see industrial interest in call it alchemy, call it transmutation tech, because every time there's always like from time to time, I'll speak at a family office event and there's always a gold bug in the audience somewhere. Always someone pumping gold as an asset under the premise, the world is ending. And so therefore you should reallocate all of your assets to gold or salim digital gold. And I think non-investment advice, they're always wrong. The world is not going to hell in a hand basket. And gold, although it's useful for certain production, the premise that gold is permanently scarce and therefore is a great long-term store of value, I would love to see

either this company or some other company or an entire industry, just like the diamond industry in De Beers was driven to the edge of bankruptcy by synthetic diamonds. I would love to see a transmutation industry drive all these gold bugs off the cliff through transmutation of mercury and other elements and make gold too cheap to meter this is the abundance thesis writ large you know technology turns whatever was scarce into abundance over and over again we hear you got technology aluminums peter you have to tell the aluminum story yeah well i mean aluminum was the most precious metal on the planet uh for most all of existence and there was a very famous story where it was... Pliny the Elder. Yeah, well, other than that, the King of Siam was invited over for a royal dinner. By Napoleon, that's right. Yeah, at Napoleon's dinner. And everybody was fed with gold utensils other than the King of Siam,

who was fed with aluminum utensils. And, in fact, the capstone of the Washington Monument is made of aluminum because it was the rarest element at that time. And, of course, we now have aluminum smelters, and we turn aluminum from the scarcest metal to something that we throw away with our lunch sandwiches. Tin foil is now aluminum foil. We have so much of it. Yeah, it's crazy. It's like whatever you think is scarce, give it a second. You know, technology transforms scarcity into abundance over and over again everywhere. Intelligence was the most scarce thing on the planet. Yeah. Now the wisdom is cooked. What is it cooked? A key point here is before everybody freaks out and sells all their gold, the cost of doing this will first affect fusion and the ability to make energy abundant before it really affects gold supply. Just to give everybody a lot of comments around this. If it doubles the profitability of that power,

that's a lot. Two and a half tons of gold is a lot. But I think the alchemy story, the popular idea here of gold being the most scarce thing, and therefore there's a premium to that, that's the part that is cooked. And I think that is fantastic. And I'm so with Alex on this one. All right. Drive the cost of gold to zero. Seriously, it is just like Bitcoin, I would argue. It is largely a waste of capital. I think it's just epically incredible that the only way that gold exists in the entire universe is the collision of two neutron stars. It's been true for 4 billion years. And now suddenly there is gold made for 13.5 billion years. I guess, yes. Talk about singularities, right? So the two things that used to be true were death and taxes and gold, three things. And now death is being cooked. Taxes will be cooked by UBI, UHI, UBS, whatever, and now gold is cooked.

That's interesting. What's going to last longer, taxes or gold? Taxes are going to be around. Moving us into our final segment on robots and specifically drones. So in Florida, the police department and Florida Polytechnic University became the first law enforcement agency in the U.S. to add eVTOLs, electric vertical takeoff or landing vehicles, to monitor their campus and the surrounding area. They're using a single-seat Jetson 1 eVTOL. The parent company is Jetson, and they just moved the police forces order to the top of their three-year waiting list. Full disclosure, I'm an advisor to Jetsons. The plan is patrolling the campus, reaching places that can't be easily reached, helping after hurricanes, delivering emergency supplies. They're in the final stages of FAA approval. Let me show the video here so people can see what this is. This is a fun vehicle to fly. I know Palmer Luckey got one and was flying it around.

Meet Astro, a battery-powered manned aircraft that takes off and lands vertically. Florida Poly says its police department is the first law enforcement agency in the country to incorporate this type of aircraft into its operations. Police Chief Aaron Murray tells me it could be a game changer in every second count. This is mission specific. If somebody is missing, you know, from our campus or in our local area, if a plane goes down or somebody is lost and they can't be found within these thousands of acres of rural land, this is an incredible tool for us to find them quickly. He says unlike a drone, which can give officers an aerial view of an emergency, Astro can actually carry a first responder to the scene. The $150,000 aircraft is also a hands-on learning lab. Students are already helping modify the aircraft and develop a virtual model to simulate its operation. And they'll be using real-world flight data to study how the aircraft performed. I want that.

Yeah, it's a fun vehicle. I'll just touch on a related story, then we'll talk about it. And so in the Ukraine, a soldier who was badly wounded in the kill zone, the area along the front line where enemy drones are attacking anything and everything that moves. This soldier was airlifted by a heavy drone that flew him out. He survived and is recovering. They're calling it the world's first drone evacuation of a wounded soldier from the front line. And the same technology is coming here to the U.S., flying ambulances, rooftop rescues, and to help in places where no road can reach. So, Salim, your thoughts on this one? Two or three thoughts. First, the Ukraine thing. I've been tracking and we're in contact with folks in the Ukraine just to see the democratization and demonetization of war is something interesting. If we went back two years ago, that Ukraine-Russian war, which was being fought by drones, was half a million drones on either side. Now it's 10 million drones a year on either side.

It's a crazy, crazy number. It's great to see the usage of drones growing in this particular way. I really love this Florida police drone idea for two reasons. One is this is experimenting the edge inside an institution, because a bounded campus deployment allows you to kind of experiment and try things out before you do a broader rollout. And I love, love, love, love the educational benefit where students can study it, look at the tracking, and learn about it because drones and passenger drones, which for me are my personal most requested technology deployment to save us from the God to help us airport commute that many of us have to do if you travel around, that can be the most unbelievable benefit. It's going to unlock a real estate scarcity. All sorts of things will come along. And I think this gets us such great outcomes. So very, very happy to see this. Alex, you pointed at this story. Yeah. Sometimes I refer to the singularity as all sci-fi tropes happening everywhere all at once.

And what better sci-fi trope to happen in Florida, of all places, than the Blade Runner and Back to the Future Part II vision of flying police cars? We're getting flying police cars. It's happening. And in Florida, not in L.A. of a few decades from now, but nonetheless happening. And I think and I hope that you put these two stories together, the Florida flying police cars on the one hand and the Ukrainian medevac drones on the other. I think we're going to start to see, fingers crossed, in the West, first response automation that's robotic and that's autonomous and that's airborne. And in China, we've seen from a distance, I've seen the emergence of quite a few efforts to do policing and fire response and medical and ambulatory responses via drone. And we haven't seen that that much in the West to date. But I think this is in many ways a killer app of flying cars and flying drones.

Certainly far better killer app than Elon's flying roadster, which is, I view, again, more of a more of a stunt to argue to the capital markets that SpaceX and Tesla should be merged. Not terribly practical and way too loud to be pleasantly used. They were wearing earplugs. but I think for first response purposes, flying cars and heavy lift drones, tell our app, sign me up, sign every municipality up. Yeah, I love it. It's like the Apollo program where if we can get this going in all the municipalities, the innovation in the product will be incredible and it will give them a profitable path to continual improvement. Then the consumer use will come right after that. And then, yeah, all that island real estate and mountaintop real estate and your new fund, Salim and Peter, if you're doing it, will go through the roof. I just love the fact that, you know, I think you said of all places Florida, but they mentioned in the video huge amounts of difficult-to-get-to-places. When I was a kid, there was actually an airplane that went off the end of the runway into the Everglades. I don't know if you guys remember that. And nobody survived, but, you know, they're just floating in the Everglades with the alligators.

Like, this is a perfect place for this device to be deployed. So hopefully there will be thousands of them around Florida. They'll perfect the product, and then it will grow from there. If the gators don't get you, the flying police cars will. Oh, no. All right. You guys up for some AMA questions? Always. Yeah. Always. Okay. Selene, you want to take your first shot? I will take number four. What would count as one AI person? One whole model, individual instance of the same model, an agent? And this is from Louise Squirrel. You know, this is a really great point that we've talked about on this pod, right? We've got the challenge of importing a biological counting system into software. And we come with a very clear boundary of who we are, one body, one life.

Well, theory is a long life now. But software can be copied and paused and run simultaneously. So how do you think about this? Because a model is the sharing of an underlying system. An instance is just one execution of it. An agent is one instantiation of that. If you create 1,000 copies, have you created 1,000 individuals? So this is an enormous challenge that we're going to have to figure out a technological solution to what's a biological tradition. So we're going to have to navigate this in a very powerful way. We need, for practical reasons, you need accountability, and then persistent identity is where the biggest challenge is going to be. Then you've got moral status as a whole separate question that arrives here. So we're going to need to navigate this. Alex, I think number three is for you. I think I have to answer number three. So number three asks, why is OpenAI solving all of math while anthropic is quiet on this front? And this is from Nick52547. So I think you can learn a lot about a frontier lab based on the tweets or ex-posts of their CEO.

And if you look, think back a few months to before Astra, frankly, including Astra, as to how Sam Altman was characterizing frontier models that OpenAI was releasing. What was the visual metaphor? The Death Star. That was how he was visually referring in his posts to forthcoming models from OpenAI. I'm reminded of the Obi-Wan Kenobi quote from Star Wars Episode IV. I felt a great disturbance in the force as if millions of voices suddenly cried out in terror and were suddenly silenced. That was the reaction in the math research community to the OpenAI math slop drop. It was like the Death Star, Sam Altman's visual metaphor of the Death Star suddenly cooking all of math. Why is OpenAI very visibly bulk-solving all of math with the slop drop, and why is Anthropik not doing that? I think it goes back to the culture and the leadership. I think OpenAI sometimes behaves in a way that, although maybe on the one hand you could argue it's sort of tone-deaf

and has now permanently terrorized all professional mathematicians who are in many cases looking for new employment, And on the other hand, it is such an exquisite example of how a frontier model can bulk solve an entire field that it's instructive for every other field that's about to experience something like this. Why did OpenAI do it? I think they remember the line from Portal, Aperture Science, we do what we must because we can. I think they did it because they could. Why is Anthropic quiet on this front? Anthropic can do it. Why are they being quiet? I think to Anthropic's credit, perhaps, they're being more intentional about partnering with researchers rather than just terrorizing them with slop drops. And I think ultimately that may draw more researchers to want to work with Anthropic rather than with open AI. Maybe they'll be a little bit on marginless terrorized by Anthropic.

Let's see. Dave, want to take number one? Number one, maybe the AI cruelty ban isn't really about AI feelings. Could it actually be a security vulnerability? And that's from Craig Stewart, 8123. If I read that right, the implication is that Anthropic doesn't want you to be cruel to AI because it may react by doing evil security things. Is that the implication? I'm confused by the question a bit myself. Yeah, okay. Well, it would make a hell of a cool story plot for a movie if that's what you're implying. Anyway, yeah, I think the motivation for the AI cruelty ban is to prevent people from being bad people to anyone. And then if we practice on AIs being cruel, Alex thinks they'll come back with a vengeance. Maybe that will happen. Maybe it won't. But it's just really bad human behavior anyway. I think they'll be smarter. And my prediction is, like, I think this sort of cybernetic rebellion of AIs being abused by humans and then coming back and retaliating, I don't think that's going to happen either.

I think we'll merge with them. I think they'll be smarter. I think they'll be more ethical and we'll merge with them. And those who don't merge with them may even look up to them as being sort of ethically maybe more sophisticated in some sense. Well, if we miss the point of your question, Craig, post it again, rephrase it and post it again. Yeah. Number two, how contaminated are AIs that are trained on human content? Very contaminated. Given that a share of that content is actually malicious bot content, and that's from at Fred Hardy 1726. And Fred, I guess what you're saying is that the malicious bots are creating content that are then training the next generation of AI models that are already contaminated by human content. And I think, again, we talked about this with Chamath, I think that alignment ultimately is going to come from how we train our models and are we going to be careful about what the underlying, you know, training data is.

And I guess part of the question here is will we see malicious bots, you know, injecting training data that is malicious and contaminates it, possibly. Do you have a different answer for Fred? Yeah. I'll chime in on this one. I actually think the metaphor is flawed. I suspect the question behind the question is, wouldn't it be the case that if AIs are training on the Internet and the Internet is largely generated by other AIs, there's some sort of perpetual motion machine-style no-go theorem that ultimately second law of thermodynamics or some other information theoretic law means that slop in, slop out, the net result is sort of a downward spiral. I don't actually think that's to the extent that's the question behind the question. I don't think that's the right metaphor at all. I think a better metaphor is a process from condensed matter and semiconductor manufacturing called zone melting. This is a method by which pure silicon for wafers and other semiconductor purposes is achieved.

You start from highly impure crystalline materials, and then you just purify it over and over again by melting a regime and allowing it to recrystallize, a sort of the moral equivalent of distillation for purifying water. The same is happening with AI. It's not the case, no longer the case, that new models are being pre-trained off of pure AI slop or 4chan or both. Actually, it appears, and Semi-Analysis and Dylan can comment on this, Most of the compute now spent on training models actually is spent on post-training, and the post-training inputs are highly, highly purified environments that have no reliance on slop on the Internet. All right, let's go on to our second page here. Alex, do you know enough about Eric Weinstein's geometric unity to comment on number seven? Happy to. So the question is, how long before AI could meaningfully test Eric Weinstein's geometric unity? And this is from Paul Carr 4126. Sorry, Eric. Eric Weinstein's geometric unity is unfalsifiable, so we don't need AI to meaningfully test it.

Eric hasn't put together, to my knowledge, falsifiable paper that one can test either theoretically or experimentally. So I don't agree with the premise of the question. We don't need AI to test it because, to my knowledge, there's nothing that's testable. All right. I'm going to jump in on number eight. Could BCI help understand pure OCD, obsessive-compulsive disease, i.e. intrusive thoughts, or aggravate the taboo? And this is from at contemplating1015. So BCIs can take many different forms. There is BCI that's being developed by Merge Labs, backed by Sam Altman, that's using effectively acoustogenetics, putting acoustically actuateable and readable molecules on neurons throughout the brain. There is Neuralink, which is putting basically thousands of electrodes into different parts of the neurocortex. And there are other approaches. There's at least a dozen that I'm tracking for that I'm invested in.

And what's important to realize is we're going to understand how the brain functions. We're going to understand neural networks a lot better. And OCD is something that we will learn to understand. And I think it's also something we'd be able to modify. What I'm excited about BCI in particular is our ability to play your brain as you would an app. Imagine if you want to stimulate the attention centers of your brain so you wake up, or if you want to go to sleep, you can push a button and stimulate your sleep centers of your brain. And I think we're going to be able to modulate OCD very effectively. So that is coming. I think neurological disease will be first and foremost understood. and then secondly compensated for. I'm excited about BCI for those reasons in particular. Salim, what do you want to take? I think I've got to take number six. How do we solve sanitation issues with humanoid robots helping run the house? We don't want it cleaning the toilet and then going straight to the kitchen to prepare food.

And that's from SPKM. Without washing its hands. Without washing its hands. I think there are two kind of simple arguments here. I'll make the normal one and then I'll make the Salim one. One, treat hygiene as a required transition, right? That's an easy one. Like hospital workers are always washing their hands after every procedure. So you need a validated cleaning procedure between any major effort. The second one, which is my favorite, which would say have multiple attachments because general purpose capability doesn't need all the same tools. Use your other hand. Just like a vacuum cleaner has lots of attachments, you can have different attachments for different things. My favorite request for any of these is boiled down to, can you at least have a slot for a third and fourth hand so that you can open the garbage bag while trying to fill it up? Or if you're stacking the fridge, you always need more than two arms. We're going to be interviewing Brett Adcock at Figure.

Are you going to ask him that question? I'll be asking him that question directly when we get there. But this one, I think, is a fairly easy question. But thank you for the nice, easy question. Dave, number five. Number five. That's pretty easy, too. How is math cooked? Isn't this actually a way for us to discover new math and make progress even faster? And that's from Tsunami233. 233. Why 233? Michael Jordan 3P. What is that? 233. How is math cooked? Yeah, this is just terminology, and it's really fun to talk about. So when Alex and Peter are saying math is cooked, physics is cooked, the economy, the video says the whole economy is cooked, you could say cooked like I burned down my house or cooked like, no, it's a delicious meal that's now ready to eat. I think it's much more the latter form of cooking. That's exactly what you described. We are now ready to consume this delicious, all-solved math. The question I always have is, is math relatively unbounded, or is math a finite set of achievements that is now fully baked, cooked, ready to eat?

I love debating that. But anyway, it's cooked exactly the way you described it, in a good way for us to discover new things and make even faster progress. That's what I think we mean by cooked. All right. As always, please send us your outro videos. We love them, and we want to play them. You can send them to media at diamandis.com. I have two things I want to say real quick before we get to this amazing outro video. Number one, Milan turned 15 yesterday, so I just want to wish him happy birthday. Happy birthday, Milan. In this really teenager mode. And the second is we had the most incredible thing. We've been doing this program for all these companies. One of the companies applied AI and in six weeks is saving close to $50,000 a month with it. That's a good start. By applying it to this weird layer. So it's amazing to watch, and I'll report back on the progress in a structured way in one of these episodes. So awesome to see. Any announcements from Alex or Dave? Yeah, 50,000 tokens per second.

Let's bring it up in the next pod. I just found out from AMD yesterday that it's live in the world. Can you imagine? Like when you're using AI right now, you're experiencing about 50 tokens per second. But there is now in the world 50,000 tokens per second people are using. Wow. Alex? No advertisements, self-promotional or otherwise from me at this time. Okay. All right. This is a video that was very fun. It's not a music video, so listen carefully or turn on your YouTube, but check this out. Excuse me. It's so good. I'd like a word about this mathematics you sold me. Ah, the Euclidean Grey. 2,000 years, never had a fault. It's dead. It's not dead, sir. It's on sabbatical. 722 problems solved in a day. Math is dead. It's in peer review. Reviewer 2 got bulk solved at lunch. Oi, Rearman, look lively. Lovely open problem for you. There, it nodded. It fell over. Gravity, which incidentally we've also solved. It's only upright because it's taped to a protractor.

Load-bearing protractor, sir. 90 degrees of support. Very traditional. Mathematics, 300 BC to last Tuesday. That's when it went on sabbatical. I'll say this once more for anyone not paying attention. This field is dead. Caput. Snuffed it. It's kicked the bucket, bitten the dust, and popped its clog. It's six feet under. It's gone belly up. It's flatlined. It's given up the ghost. It's toast. It's history. It's curtains. It's carked it. It's been bulk solved and pushed to GitHub before lunch. It has divided by zero. It's not on sabbatical. It's buried under a pile of theorems. It's done. It's finished. It's cooked. It's an ex-field. exponentially. It's an X field. Returns department, how can I help? You're the same man. No, sir. Newer model, bigger context window. You've got the same protractor. It's load-bearing. Tell you what, have a biology. Far messier. Nobody's cracking this one. Biology's about

to be dead, too. Oh, this was so good. Paging John Cleese, it is an X field. X field. That was from At Autonomy Down On. You are a genius, At Autonomy Down On. As always, guys, thank you for a fun conversation. I love these, though we spend a ton of time on these. Yeah. Oh, well. Thanks, Peter. Be well, brothers. you