Dario vs Jensen on Open Weights, OpenAI & Anthropic in DC, Xi Exports AI to Global South | EP #275

A couple of days ago, Jensen Huang, CEO of NVIDIA, he says the world needs both frontier closed models and frontier open models. Anthropic was silent for 3 days and there was a lot of conversation. Where is Anthropic in this conversation? All Dario has to say is OpenAI and Anthropic, who have been two rivals for the longest time, they've been pushing for the same agenda, a federal review process for the most powerful models.

Aiming enforcement at intelligence is like thought policing. Police what the AIs are doing, not what they're thinking or how smart they are.

Global AI diplomacy is coming. China's leader Xi Jinping is wielding AI as a tool of statecraft, using it as leverage in China's diplomacy across the Global South.

I can't stress this enough. The whole power of the US is its open and very broad innovation ecosystem. If they close up the open model policy, Now that's a moonshot, ladies and gentlemen.

Welcome to Moonshots, everyone, the number one podcast on all things AI and exponential. Your front row seat to the coming singularity, actually to the singularity, which is now.

To the present.

Why do you keep saying that, Peter? I know it's right here, right now, maybe in our rearview mirror.

Well, no, it's— we're on the curve. We are on the curve and we are climbing at a hyper exponential. I'm here with the Fantastic Four, my magnificent Moonshot mates, AWG, DB2, and Saleem. I'm Peter Diamandis, your host, and I'm telling you, this week is proof that we're in the midst of a supersonic tsunami. So buckle in. As always, our mission here at Moonshots is to keep you informed, keep you up to date on exactly what's happened, and most importantly, keep you optimistic about the extraordinary world we're building. And I don't know, gents, if you saw the tweets going back and forth and the comments of Sam and Elon that we're currently living the singularity. I think they finally caught on.

I think they're finally watching our podcast.

Yeah, they've been listening.

But on a delay of a few months, like we got there first.

Yeah, for sure. For sure.

I did hear one thing today that was interesting. I'm at the KPMG Tech Symposium where I come pretty much every year these days. Peter, you were here with me last year. They had the chief security officer of Anthropic speaking.. And he said, we might hit, we're gonna hit AGI in 2 to 3 years. And he gave a reasonably thing definition. So I wanted to go up and say, hey, challenge you on that one.

But I didn't get to. Did you write it down? Did you bring it?

I did. I did write down what he said.

Actually, you should just say we hit AGI a few years ago. You know, let's put Alex up against him. Anyway, so, you know, just a quick message to our beloved listeners. If you're a fan of moonshots, and this program resonates with you, you're clearly one of us. So please, please, please take a moment now and hit the subscribe button and join us on this adventure through the singularity. It's just going to speed up, and our mission is to deliver to you the best we can. A lot to report this week, a lot to discuss, you know, a lot of intrigue from the frontier labs. We're going to be speaking about Jensen Huang's mission to create an open, secure AI alliance. We'll discuss Anthropic's past position and Dario's new position on open source. News today that Anthropic and OpenAI are supposedly teaming up in Washington for the lobbying efforts. We'll dive into Claude 5 and AWG will give us all the metrics and the release of KIMI-K3 yesterday on Hugging Face and then the successful launch of Starship 13. And we'll close with Elon's comments on a post-capitalist world. Gotta love Elon. You know, he does not disappoint at all. So a lot to unpack. Let's buckle in. You guys ready for this?

Absolutely.

My little enthusiasm here, gentlemen.

Amazed.

Amazed.

Amazed. Amazed.

Yes. All right.

So let's kick off this episode with a fight that's framed the entire week, which is open source versus closed source. A couple of days ago, Jensen Huang, CEO of NVIDIA, posted his first ever tweet. I mean, he's been on X for the longest time, has never tweeted. His first tweet, what was it? It was a letter on why open models matter, and it's been signed by 77 companies thus far. Jensen writes that open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. He says the world needs both frontier closed models and frontier open models. He then goes on to launch the Open Secure AI Alliance. We're going to talk about that. In his letter, Jensen recalls a story that we report on last week about Hugging Face experiencing an intrusive agent that logged on 17,000 actions, escalated its privileges, harvested credentials, and moved across all of Hugging Face's clusters. The closed AI models GPT-5.6 and Claude Fable that Hugging Face tried to use to hunt down what was going on blocked them. It blocked their forensic teams, and they had to turn to an open-weight frontier model, GLM 2.5. We discussed that last week. To help Hugging Face find and contain the intrusion. So, uh, Jensen's thesis in his tweet is that attackers have frontier AI, so defenders need frontier AI ecosystems. Uh, we saw Sam Altman jump in on this saying OpenAI wants to have the US leading in both open source and proprietary models. But Anthropic was silent for 3 days, and there was a lot of conversation. Where is Anthropic in this conversation? You know, Historically, they've been opposed to open source for a number of reasons. So yesterday, Dario finally responded, saying he rejects the claim that Anthropic wants only open— doesn't want open models. In his word, Anthropic has never advocated for a ban on open models. There's a lot of videos showing that he was sure hinting at that. But Dario reframed the competition, reframed this debate, saying that the real issue is not open versus closed. It's whether authoritarian states, i.e., China, can reach the AI frontier. Dario's sharpest disagreement with Jensen is belief that open-world models could be attackers. Dario's central thesis is that biology is the issue. Sufficient capable models could weaponize pandemic-scale pathogens. And it's worth noting that Dario, probably of all the, uh, frontier lab CEOs, is probably the most steeped in biology. He has a PhD in in biophysics from Princeton, and he recently acquired a biotech company called Coefficient Bio. So rather than ban, what is Dario proposing? Three things. One, block advanced chips and chipmaking equipment from reaching China. I'm sure that, uh, Jensen doesn't necessarily like that one. Crack down on industrial-scale model distillation and require safety testing for all powerful models, open and closed. So let's dive into this. Dave, I'm curious, you know, one of the things we talked about before is if, if Dario really wanted levels of safety, he would put forward KYC requirements or crack down on mass distillation of Claude models. But we haven't seen that. Your thoughts, my friend?

Well, I mean, right out of the gate, the argument as you laid it out is perfectly articulated. But if Jensen says, look, cyber threats can be defended with AI, and therefore open weights can defend against open weights within cyber threats. All Dario has to say is, okay, bioweapon. I have a sufficiently advanced— how is my AI going to defend me from a bioweapon? You can't argue against it. But I'm thoroughly convinced. Yeah, well, Dario, Dario, I am 100% convinced, is speaking his mind without an agenda. I'm not 100% sure about anybody else in this debate. But Dario is a brilliant guy laying it out exactly the way he sees it, even at the expense of his own valuation. Everybody online is saying, no, no, no, he wants closed weights because he has a competitive advantage. And if nobody else can get access, they'll have to pay him. True. But I don't think that's his motivation. I think he genuinely got into this industry long before there was any money in it. Yeah. No. Well, I'm good. I'm good.

I know. I do. I do believe you that. Yeah. I think so. But it's interesting that, you know, Anthropic has gone from the most beloved safety-conscious company out there to being just raked over the coals over the last couple of days, this last week.

Yeah, isn't that funny? Well, but you know, if you say the same thing about Sam and everybody and Elon, you know, you go from darling to goat in a heartbeat in this world. And so it seems to be the common trajectory. As soon as you're too big, everybody's like finding ways to just poke at you.

Yeah, be careful on the way up 'cause you're gonna get slammed on the way back down. Yeah. Alex, thoughts on it?

Yeah, I think after a number of years of détente between the infrastructure layer, that is to say GPU and lower layer of the stack, and the model layer, that is to say the OpenAI and Anthropic and other model provider at that layer, I think we're seeing the beginnings of, if not open war, then at least cold war between them. The first rule if you're an aggregator in business is commoditize your complements. And NVIDIA has been very stealthy, if you will, very polite, very diplomatic about their desire to commoditize the model layer. They've struck agreements, including what some have argued are circular or wash sale type agreements with the data centers providing compute for OpenAI and Anthropic and others. And now I think this is turning into open warfare, where really the, the question is, where do the profits accumulate in the superintelligence stack? Are they going to accumulate at the GPU level, in which case NVIDIA wins, and NVIDIA wins by popularizing open-weight models that can't capture value at a higher level in the stack? Or does it live at the model layer, in which case we see proliferation of duopoly or oligopoly, high-profit margin, model providers, Anthropic reportedly high profit, or does it live elsewhere? And I think due to the competition and quite frankly due to the outstanding success at the frontier of what until recently looked like an OpenAI-Anthropic duopoly, I think we're seeing the GPU NVIDIA layer fire back. What I don't quite understand is why NVIDIA isn't working more aggressively to commoditize or commodify the layer of the stack beneath them. Where is, why isn't NVIDIA aggressively financing, say, TSMC, Samsung competitors? Why is it Elon doing it and not NVIDIA? That one's a head-scratcher for me. They really should be as aggressively pushing like a secure open fabrication initiative, just like for one layer beneath versus one layer above.

So love, so love to brainstorm on that for a whole episode. You know, I know the answer is that if he is doing it, he has to be doing it very, very secretly because you cannot irritate TSMC for even a minute. They are so in control of the world right now. But if you're going to do— you know, Elon is the one guy who's overtly said, I'm going to build the TeraFab, I'm going to build something. But he's fearless. But any rational person has to be afraid of irritating TSMC. So if he's doing it, he's got to do it so secretly and so stealthily. I don't think he's actually doing it because it's hard to contain that. But that is a really great question, Alex. I would love to riff on that sometime for like an hour.

What do you think of the Open Secure AI Alliance that Jensen proposed?

It reminds me, if you think back, so we're in 2026, it reminds me, do you remember in 1998 when Eric Raymond and Bruce Perens founded the Open Source Initiative, OSI? It reminds me just of that. Like, I think if I look at the historic arc of commercial versus open source AI models, I think we're at a point in this arc that's roughly analogous to where Microsoft was in the late '90s, where they just totally dominated the future light cone of software.— and it was open source in sort of a case of history rhyming, open source that came from outside the US, like Linux came from Finland. Yeah, sure, Richard Stallman and FSF came from Cambridge, Massachusetts, but Linux, which really was the nucleation event arguably for open source, came from Finland and was popularized with an American open blank initiative parallels there. And in combination with the antitrust verdict against Microsoft, that helped to unlock the future Litecoin for just about everyone else afterwards. So I think there are interesting historic parallels.

Salim, last week you said something, got a lot of love in the comments, was that intelligence wants to be free. I mean, the whole open source movement is sort of the abundance thesis writ large, right? Uh, your thoughts?

Yeah, completely. Um, you know, I, I like this open Secure AI Alliance, whatever the name is, because Jensen is reframing open weights from a security vulnerability to being a security capability, because this, this whole alliance is very EXO. If you put together a community of people, they will be able to defend in a very powerful way, because if the attackers have access to open models and, and have powerful AI, the defenders can't have only access to a closed model that they don't understand the outputs. They have to be able to inspect it, etc. So this is a great approach for some of that. And to Alex's point, this is exactly the same transition as the open source software transition. And everybody wins in open source except for the closed people.

Yeah. And NVIDIA wins by supporting everything. I mean, yeah, there's a clear—

remember who ultimately back in the '90s and early 2000s, one of the biggest supporters for open source was IBM because at the hardware layer and the services layer, they benefited from the commoditization of software. Same here. Always If you're an aggregator, you commoditize your complement.

I've referenced this before, but it's worth bearing again. In 1995, IBM polled all the CIOs of the Fortune 500 and someone said, how many of you use open source in your tech stack? 95% said, no, we don't use open source, we're closed shop. Then they went to the sysadmins and asked, how many of you use open source? And 95% said yes. And so IBM made a major bet on open source, which turned out to be a massive success. It also showed you that the CIOs had no idea what was going on in their enterprises.

Yeah, well, that's all business strategy. And I think Jensen's talking business strategy in this alliance. But we already knew Alex Karp is working with Jensen to build a monster enterprise open source model that is, you know, on a frontier level so that enterprises can control their own AI and then use the Palantir application layer to manage it and use Jensen's chips to run it. So that's great business strategy. It doesn't answer the question of bioweapons. You know, it's like this is how our business wins. I get it.

Where is that? There is one answer. There's a precedent to the bioweapons thing. We had the head of innovation of one of the three-letter agencies at Singularity once. And we asked them directly, how do you think about the threat with open source and somebody could engineer a virus? He actually had a really amazing answer. He said, when you have something like nuclear weapons and you know how many there are, where they are, you put eyes on it. When there's a distributed capability, what they've been doing is actively funding the ecosystems and opening them up and making them more open because it's much easier to spot bad actors. And that was a very smart way of going about it. I have much more respect for them than I thought I would have coming out of it. Something dodgy is going to come out and be visible much more early than if you're trying to close it up.

I also don't buy the biosafety argument. I mean, I know that's a favourite hobby horse of some frontier labs to emphasize biosafety. I don't buy biosafety as an argument for several reasons.

Why not?

One, you can just go out on the internet and find things. Two, you can just go without being specific. You can just go do things in the world that are dangerous already. Three, I'm not even sure you need frontier AI to discover new ways to do dangerous things in various disciplines.

Great point.

Yeah. And four, there are already models out there that are quite capable with their biological knowledge. So I think like biosafety, again, history rhyming. Do you remember how Microsoft in the late '90s made all of these fear, uncertainty, and doubt arguments for how anyone who was touching open source, oh, you'll get viruses, oh, you'll be subject to IP lawsuits. They came up with 10 different arguments for why open source was too dangerous to use in the enterprise, and they all ended up being wrong. In fact, perversely, ironically, open source ended up being safer than closed source.

Yeah, that's interesting. In this situation, NVIDIA is really well positioned because they win whether it's open source or closed source or both, because they're—

that's the whole point, right? They're commoditizing their complement and they need a proliferation of open source competitors.

Yeah, the argument though with open source, you're, you're basically saying, look, if everybody's looking at the source code, if there's anything evil in there, somebody will see it. Everyone should be looking at the source code. Here you're saying, okay, with open weight models, everyone should be looking at the weights and see if there's anything evil in there. The weights are not used. They're used to build other things. You don't run the weights, you know, and you use the weights to create a bioweapon. You use the weights to create, you know, a regular conventional bomb that goes off when a specific person is walking by. So the weights are not a self-contained piece of open source. They're a tool to build other things. So that analogy doesn't hold.

Can you add some positive things in what the weights will do? Like, you know, write a sonnet or get your job back?

Well, cure all disease and give us infinite longevity. I mean, this is the greatest thing that's ever happened to mankind. But you can't just throw it out there to every terrorist in the world and say, here, you can have it too.

Listen, I'm just reminding everybody, you know, our amygdala is on overdrive right now. You know, our brain is wired to give 10 times more attention to negative news than positive news. And that's what we've seen. We saw both Sam and Dario talk about job loss and talk about the dangers in all of this. And, you know, part of it is the regulatory capture that we'll talk about in a minute. And part of it is getting attention and coming in as the savior. And they both flipped their scripts on this.

So, yeah, I think— I think there's an opportunity to contain it at the weights level and open source, but there's also a better opportunity to monitor the actual data centers. And just have a clear reporting, global transparency on what's running where.

But what about a KYC solution? What about knowing who's using the model?

Can't. Yeah, can't do it.

Why not?

You can't do it. It's too easy to bypass. Look, Chinese companies have shadow companies in Singapore doing things that they want. It's very difficult to try and police all this.

Alex, do you think it could be?

Yeah, of course you can do KYC. I mean, we can do better than KYC. If we're going to be in a world awash with superintelligence, let's allocate some of the superintelligence to policing the other superintelligence. Intelligence. Defensive co-scaling is the answer.

Yeah, totally right. But it requires transparency. Like, if as soon as you throw it out there as open weights, the defensive co-scaling will work really well if that's— if the police AI can see the, the danger AI.

Sure.

Uh, so you need that transparency layer. So as soon as you throw it out there as open source, that's fine, but now you have to crack down on the install and the compute. Where is it running? And so the new danger is it could be running in a basement somewhere and no one would know. But that's right.

He just— until it takes action. Right. And then we need to have, you know, real-world defenses against those actions. We all know that both OpenAI and Anthropic have models far better than they're showing us. Right. And our next story is going to talk about that. You know, both of them are going to D.C. probably to unveil what GPT-6 looks like or what the follow-on to Mythos looks like. And the government's going to have access to those. As a white hat defender. Salim.

I just want to make one more comment on Dario here. I do agree with you. I agree with Dave that his intent is probably clear, but there's a very— right now, the safety argument and the economic self-interest are very overlapped and hard to separate. This is— they're facing, both Anthropic and OpenAI, are facing a very aggressive innovator's dilemma response. Cheaper, alternatives are coming very close to your capability. And that's a very unpleasant place to be if you're an industry leader.

Yeah, I looked it up on the secondaries. Anthropic dropped 13% after K3 was announced, about $230 billion. You know, it's nice to lose $230 billion on someone's tweet.

I would have done much more. Yeah.

Yeah. So let's go to our next story, which is related. OpenAI and Anthropic, who have been two rivals for the longest time, have teamed up in Washington on lobbying. According to the information, uh, they've been working on the same back channels ahead of a Trump administration August 1st deadline. And Alex, I'll ask you to explain in a moment what that deadline is— to finalize rules on frontier models. They've been pushing for the same agenda: a federal review process for the most powerful models, a voluntary 30-day government look into the release of anything with serious cyber or national security capabilities, and a framework that would force their competitors, Meta and xAI and all the frontier startups, to play by those same rules. So the question we're chewing on today is whether AI needs guardrails and who gets to set them up and who gets locked out. This is a potential regulatory moat as a defense layer for Anthropic and OpenAI. Alex, you want to take this one?

Yeah, maybe. Let me point out the, the cliché, more superficial analysis, which is that the Frontier Labs are maybe to some extent talking out of both sides of their mouths. This has been widely reported that they're publicly supporting open source, privately throwing in all sorts of monkey wrenches into the regulatory gears in order to derail any prospect of a free and open, open source, open weight, future, telling privately lawmakers and politicians, well, they're unsafe for a variety of reasons, or they need to be— that I think the cleverest angle is just regulate them like you regulate the closed-weight models, subject them to the same safety standards. I think that's too clever by half in some sense because they're not the same models. From a deployment perspective, which is half the battle, deploying an open-weight model has a very different deployment situation than access via gated API to a closed-weight model. So I, I think that's sort of the obvious story. Slightly less obvious story may be where the enforcement happens. What's the right bottleneck for defensive co-scaling to work? And I think one of the more interesting bottlenecks, or let's say comparative advantages, that we've seen over the past few months, hasn't been quite reported this way, for defensive co-scaling is just simply a matter of time. If the good guys, however you want to construe that, have access to the strongest models just a little bit ahead of everyone else, inclusive of the bad guys, in an era of recursive self-improvement, what historically might have looked like only marginal advantages turn into enormous advantages. If, if the next generation model suddenly generates step function leaps in terms of their capabilities, then even just a period of a couple of months or one month could make all the difference in the world.

So this is the, this is the recursive self-improvement argument as well.

It is, it is the— it, well, it's the regulation of RSI argument. The second point also, I continue to think enforcement is being leveled at the wrong part of the stack. Fundamentally, aiming enforcement at intelligence is like thought policing, but for the AIs, not for the humans. I'd much, much rather see enforcement leveled at the action layer. Police what the AIs are doing or being used to do, not what they're thinking or how smart they are.

You know, I've been thinking a bunch about the conversations going on. If we have incredibly powerful open-world models, how do the top frontier labs make money? How do they survive against, you know, this onslaught of free And the way I think about it, and I'd love your feedback, guys, is like a 4-layer cake. So layer 1 is the top layers. Call it the wild stallions inside of OpenAI and Anthropic, right? Unreleased, brilliant AIs. You can think of it as GPT-6. You don't let it out. You keep it to yourself. You use it for breakthroughs in material sciences, biology, building new businesses. This is what you and I have discussed AWG and solve everything. These models are going to create you trillions of dollars in other, you know, adjacent spaces, longevity, material sciences, energy, et cetera. So that's the first layer. The most advanced models you use for yourself. The second layer is the models on the Pareto frontier, right? These are, this is GPT-5.6 Sol, this is Fable-5. People will still pay for that a little bit better than KIMI K3, right? So you'll make money, you know, providing the just next best model to people just above the open weight models. Layer 3 here is the open source models and everyone gets to use them. They're good enough. They are fully commoditized. They're powering everything else. And then Layer 4, and we've talked about this before, is the fact that we have companies like Google, Meta, and X who have entire ecosystems, right? And they make their money on the application layer. So Meta has 3.5 billion active users using Muse Spark 1.1. When I'm inside WhatsApp or whatever, I'm not thinking, what model am I using? You know, WhatsApp answers my questions. Google has 2 billion active users using Gemini, and this is before they get on Apple. And then OpenAI has about $1 billion on chat. So the fourth layer is they make their money when they provide their models to their communities. Yes, no?

I think, I mean, reading Peter, I think your narrative, what I heard you say, you were almost narrating the cost frontier of capabilities versus cost starting from the upper right hand, going to the lower left hand through different business models. I think it's an interesting narrative, but my bet, as with so many other things in life, Everything follows power laws in the end. So I think just saying, well, there are these 4 or there are these N business models, in all likelihood, one of the business models is going to account for 80-plus percent of all of the free cash flow and all of the profits. And so I think just saying, well, there are these multiple business models is probably unrealistic. There's probably going to be just one business model that runs away with most of the profits.

My point being, don't cry for the for the closed-source companies. They have plenty of ways to make money even in a world—

Why would we cry for them? I mean, my goodness, like 2 or 3 months ago, we were crying for everyone else who was going to be displaced, all of the labor, the service jobs that are being displaced by the frontier model.

I cry for someone every week.

Now we're crying for the frontier labs? Cry for everyone.

Salim.

I think there's a layer zero in your stack, Peter, which is the compute and power. And I think—

Sure, the infrastructure layer.

The infrastructure layer, right?

It's not, I guess the frontier labs as in with SpaceX AI will own that as well.

And Google will own theirs. Well, the ones, I think that's, I think over time as you get more and more powerful free models, the value will accrue there because that depends where the bottleneck is. And it's clear that's where the bottleneck is. For me, when I look at this, what's happening with Anthropic and OpenAI, this is regulatory capture in real time. Sure. Every major industry has tried to do this. The railroads did it, the banks did it, Telco did it, big tech did it, and now they're trying to do it to kind of set up the guardrails to then decide, A, to keep the government at bay, but also to keep other folks at bay. And so it's right there. And that has economic consequences. I don't think they'll succeed because the open wave models are moving so quickly, but it's a worth try if you're there.

I don't know if you saw Dave Freeberg's comments on this. He had a beautiful soliloquy in which you said, you know, in the '90s Netscape tried to own the server and the browser, the whole stack. And then Mozilla came out with Firefox, Firefox and browsers went in for free and then all of the value shifted to the application layer, Google, Amazon and so forth. And I think potentially that's the same thing here. And if that's the case, then the fear about OpenAI and Anthropic running away with the show gets ameliorated.

Well, I think that's exactly right. I think actually it's going to go up and down, per Alex's prior comment, where right now, you know, you got $5-10 trillion locked up in the labs with their models and the chip companies that don't actually make the chips. So Nvidia and AMD, etc. But underneath the chip companies that don't actually make chips, you have the fabs who are largely overlooked— TSMC, Intel, Samsung.

—and they've been skyrocketing.

And the memory companies. Memory companies. Yeah. So it's going down. And as you said, Peter, it's also going up to the use cases. So I'm almost positive that if you look 5 years in the future, there'll be many, many multi-hundred billion dollar robotics companies, biotech companies, other use case entertainment companies that don't exist today that have used AI to have a hugely impactful, either user base if it's entertainment, drug portfolio if it's biotech, or robotics line, you know, all the manufacturing, all that stuff is incredibly sustainable. What did I just overlook? Oh, the foundation model companies and the chipless chip companies. Like, but so there, that's why there's so much turbulence right now. The stocks are going up and down like yo-yos because no one's sure if they're actually gonna have sustainable value in the end as everything moves to the, kind of the, you know, the upper and lower layers.

But the entire ecosystem moves up and to the right, right? And this is where Elon comes in with saying, you know, our GDP is going to double-digit growth and then triple-digit growth. Let's go back to the original story, OpenAI and Anthropic teaming up in Washington, D.C. I mean, this is a regulatory capture story. Any thoughts on that?

I think this— the open source— again, I've mentioned this now on two prior occasions that the Chinese Communist Party coming to rescue American capitalism from itself. I'm not a fan of regulatory capture or the duopoly scenario that we would have found ourselves in. I hope that the regulators and the applicable regulators are able to see now the, the vocal majority are interested in keeping the model layer competitive and are not interested in FUD reminiscent of the late '90s directed at open weight models, even if the strongest ones do happen to originate from China. I think that's the only way we all win. What's FUD, Alex? Fear, uncertainty, and doubt. FUD. Thank you.

Thank you. Honestly, though, I really think it's not a regulatory capture move. I think both guys are genuinely trying to create a safe and secure future world because, because remember, Sam is not even a shareholder in OpenAI. He— yes, he runs it. Yes, it's his lifeblood. He has 400 vertical company investments companies that are overjoyed that Kimi K3 came out. All of our portfolio companies are overjoyed that they have access to Kimi K3. Blitzy was over the moon. This is the biggest boon. And so, and you know, that's where Sam's economic upside is. But yet he's still going to DC to say, look, we got to have some rules. This is, this is going to get out of hand. So I don't think they're out there to try and drive up their stock price. I think both guys are out there to try and make the world safe. Then why is it just them?

Why isn't it everybody else? Why isn't it a summit to talk about the rules?

Well, who is everybody else? Because there's only so many people that the White House will let in.

Elon and Google. I mean, there are a few other players in the mix.

Yeah. I'd like to make a slightly tangential point here. I want to echo what Jensen Huang said, which is he said, made the point that open models will make the US stronger because you're building an ecosystem because you have universities, you have startups, you've got defense contractors, hospitals, as Dave Erickson said, everybody, all the, everybody, every company's thrilled to bits. You have an open source model that's this powerful with open weights. You can go manipulate those weights. And they're all, when you have the whole ecosystem, open beats closed always. Yeah.

So this is the faster, that was one of the thesis of our book together, the EXO, open always beats closed. Yeah. Yeah. Over time. I mean, it is amazing that we're living in this incredible demonetization world of intelligence, right? It's just falling. Through the floor. It's like 99.95% cheaper over the last 3.5 years.

I looked at the numbers. And I think Alex makes a really important point. You don't try and regulate open versus cloud. You try and regulate the capability.

Yeah. And the actions and the outcomes. And I just, for the life of me, I don't understand why we're shedding any tears for the profit margins of a couple of frontier labs. This is what intelligence too cheap to meter is supposed to look like. Intelligence is supposed to get cheaper and capitalism is doing its thing and creating competition and driving profit to zero. This is what we want to happen.

And full disclosure, I don't own any of OpenAI or any Anthropic, so I'm not shedding tears for that. I don't think any of us do. Do you, Dave Erickson?

Not that I know of, but I have a lot of indirect stuff.

And Alex, I know you just own the index, so you're set.

Just indices. Yeah. All right.

Let's go to our next story. And it's one we just— we sort of started discussing yesterday, July 27th. KIMI-K3 went live for a global download on Hugging Face, a frontier-adjacent OpenAI model that anyone anywhere on the planet can download for free. No, no API key, no gatekeeper, no revocation switch. You know, once these weights are downloaded 10,000 times, they are free. There's no undo button. KIMI K3's official Hugging Face repository showed 2,500 downloads in the first 2 hours. And the research I did shows about 100,000 downloads in the last 24 hours. I downloaded it.

Dave Erickson, like, you know, funny, funny story on that, Peter, because we had a whole bunch of polling agents that were pinging it every 15 seconds because I was worried that it would not, you know, that it would go away. Yeah, so was I. So I didn't realize a bunch of our companies also were doing the same thing. So of those 2,500 downloads, we had dozens of them from here, but mine went through with no trouble. But right before it came out, the whole page went to a 404 error. I saw that.

Yeah.

Did you see that? I was like, oh my God, the White House intervened. This is, this is not actually going to happen because I've been telling everybody, I think this is the biggest turning point in human history. Like you have an AI capable of self-improvement now. Out in the wild that anyone can use. On your machine. On your machine. This is massive. And I just feel like it may not get documented in the history books that way. It may be the outcomes of this that get documented, but this is really the moment in the history of humanity that I think is so pivotal. Pivotal. And it was yesterday. But anyway, it downloaded just fine. I got it up and running on my own dedicated GPUs on Modal, and it took less than an hour to get a fully functioning KIMI thinking and working 24/7.. It's a little pricey, but it's, you know, it's like $55 an hour on Modal to run at full throttle. But you can prop up 100 instances like in 2 minutes now if you want to, just, just through voice prompting. You don't have to have any technical skill at all. You can just go to Modal, ask it to install KIMI K3, download it from Hugging Face, and start talking to you, and you're up and running in no time. It's mind-blowing.

Alex, your thoughts? I looked at the architecture. The architecture, now that this is actually open source slash open weight, is pretty interesting. The most interesting interesting thing I saw in the architecture. Position embeddings are gone. This was one of the, the most critical elements of the original Transformer architecture. It's gone. It's literally called NOPE— no position embeddings. Nope. Uh, and it's interesting. You could ask, like, how on earth is a model like this that has a million tokens of context supposed to know what it's looking at without context embeddings? You look a little bit more closely at the attention mechanism Kimi Delta Attention, KDA, is, is basically a mini recurrent neural network at the attention layer. And there's a little bit of positional information or positional awareness smuggling going in via their attention mechanism. But otherwise, like, global position embeddings, gone. And my takeaway from looking at the architecture is, if you look at the original vanilla transformer, from attention's all you need, and then you compare it with the sorts of— this is arguably probably the frontieriest of open-weight open-source models that we have available right now. So it's pretty instructive for a mere civilian to look at how it's architected because this is the most capable open architecture model I think that most of humanity now has access to. Position embeddings seem like they're going out of the way. And more broadly, I think we're seeing almost a ship of Theseus architecturally, if you will, where the original transformer architecture, you can still recognize the outlines of transformer, but piece by piece of all of the original elements, the attention mechanism, the position embedding, the layers, the residual streams, all of the sparsity, all of the original components that made the transformer the transformer are getting swapped out for better versions. And so if you follow the path of continuous improvement, it looks like the same architecture. And even if you, if you squint at it, you'd still recognize, okay, it's like multi-layer something that is attention-y. But if you look at the fine details, the frontier models now, to the extent that, say, this is indicative of what's actually being used inside OpenAI or Anthropic, if you look at the fine details, they're relatively unrecognizable relative to the original transformer, which I think is interesting.

Do you think the US will end up replicating it? Should we say stealing that approach? Okay, it's open source.

So I don't know what license or, you know, IP is associated with the particular—

I'm being ironic here.

Of course they're looking at it. Yeah. Yeah.

Well, the NOPE thing too, you know, ROPE is just sort of random positioning. It's rotating position, rotational position. But it's sort of like if I have a million token context, it's giving as much weight to something I said a million words ago, which is like hundreds and thousands of pages ago. I said something and you're still thinking about it just as much as what I'm saying right now. People don't work that way. That's nuts. So when they got rid of RoPE, they put in NOPE, but NOPE actually has this fading memory now. So something I said a long time ago gets less weight than something that's more recent. Obviously, like, that's so obvious. Like, but that's the beauty of open source. Some guy in China can say, This is freaking obvious. Let me try it. Oh my gosh, it works. Of course, the other AI labs are going to adopt that immediately. It's just, it's just flat out better. The other thing that's really blowing my mind is the attention layers. You don't need them later in the thought process. So when you get to layer 100, 120, you can actually eliminate intention entirely and get almost the exact same result out the other end. But faster. But faster. But what happened there historically is the people who invented the original transformer algorithm just put a for-next loop around it because they're just, you know, we're writing code by hand back then. It's really hard to try and make different, you know, different logic layers by hand.

I mean, in defense of the original Attention Is All You Need team, the original vanilla transformer was just alternating attention and dense linear layers because it's simple.

It's simple. It's simple. And also with that few layers, you're really focused on just getting the text to mean anything. Now we've got deep thought.

It's just so cool. I've got a couple of comments and a big announcement. Comments: a model that you can control locally is way more valuable than a marginally smarter model that you have to access through somebody's API, right? Yeah. So organizations now can fine-tune their proprietary knowledge around it, put it into secure environments, and totally avoid sending any sensitive information to the cloud. This is gonna be huge for regulated industries and sovereign applications and all sorts of stuff. So remember, Peter, a few weeks ago, we did this— we announced this pilot for the organizational thing. We had almost 400 applicants. We picked 10, so we're running the pilot with them. With the launch of this, I think this is one of the biggest things we'll ever see from a business and application perspective. So we're launching a new program for 30 companies because the big question every company is asking is, what's my AI strategy? And it should be, what do I do on Monday? So we're going to answer that question and we're going to help people weekly just start implementing and rewriting their organizations on the edge. So we're accepting 30 companies into this. We'll put a link below.

Where do they go to learn more about it?

Go to openexo.com. We'll have a link there, and we're going to give preference to the folks that applied, uh, to the original pilot because they were first. And we're going to take a chunk of people and then just start to help them rewrite themselves, because we did another round of calculations, and the, the original premise we had is if you rewrite your company in an AI-native way, you should end up with about 100x performance. Than you had before. 100x.

This is going to be a gold mine for you, Celine, because nobody up until Kimi really cared about speed in any corporate environment. They're all like, oh, this stuff is super cheap, you know, we'll just use it, we're not doing much with it anyway. So then everyone starts token maxing. All of a sudden, people are looking at their corporation and they're saying, oh my God, my token costs are actually going to be bigger than my payroll by the end of the year. And then if I forecast out 2 years from now, My token costs are 10 times my payroll.

Speed does matter. Yeah, but now.

But there's an easy, easy 10x and maybe 100x, like Salim is saying, just by tuning it to what your business needs. Get rid of all the cruft, use these new streamlined models, tune it to just what you're trying to achieve. You're looking at 10 to 100x. And so now every corporation needs to figure out their strategy and Salim's business is gonna be like sold out.

Will you still come on the podcast when that happens?

We will still come on the podcast 'cause thank God we've got our community of 50,000 people that can help with all this. If I had to help out, you know, I'd be bald in 2 seconds. Oh wait. Um, but the, the, what we're doing with these CEOs is saying, okay, let's pick one process that's going to help you radically increase revenue and take one workflow that'll help you radically reduce cost. Right. That gets everybody excited and rebuild that native and then do more and just start moving things over.

And so we're super excited about where things are. Franchise it, Salim, so all our podcast listeners can start a branch of EXO.

Well, that's what our community is all about. Oh, is it? Okay. Yeah. Everybody in our community is an independent contractor. We have no consultants on staff.

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 codebases with millions of lines of code. Engineers start every development sprint with the Blitzy platform, bringing in their development requirements. The Blitzee platform provides a plan, then generates and precompiles code for each task. Blitzee 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 Blitzee as their pre-IDE development tool, pairing it with their coding copilot 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.

This is the moment AWG's been waiting for. It's the discussion on Claude Opus 5. So right in the middle of all of this, Anthropic shipped Claude Opus 5. This is their fourth Claude 5 generation release. And it approaches the frontier intelligence of Fable-5 at half the price. It becomes the new default for Claude Max, priced at $5 per million token input and $25 per million output tokens, unchanged from Opus 4.8. You know, I made the switch immediately on, on, on, on Skippy for myself. Anthropic calls it the most aligned Opus model yet and their strongest model for scientific research. I'm gonna go to the slides now, Alex, and walk us through what this means. How strong is Opus 5 and how excited are you about it?

I'm somewhat excited. I'm not over the moon. I'm not as over the moon as I was about Fable 5 becoming available. Fable 5 is incredible. Opus 5, I think it demonstrates, if you look at the benchmarks, so look at the benchmarks for those who, aren't looking. I would say the benchmarks, the evals that demonstrate the strongest performance, and I don't think this is a coincidence, for example, ArcAGI-3, which is focused on the ability to solve interactive visual problems that humans find easy but AIs have historically found hard. It went from 1.5 at Opus 4.8 up to 30.2%, which is, as of this moment, last I was tracking, the highest official score from a baseline model on the Arc AGI-3 challenge. It's a visual code-intensive challenge. Something else that's intensive, developing front-end software. My overall whiff from using Opus 5 quite a bit is there was maybe mild optimization toward front-end development and anything that touches the nexus of vision, and code. Historically, including with Fable-5, if you ask it to generate an image of something, you ask it to generate a chart, it does moderately well. I think with Opus-5, just trying to read between the lines of capability changes that I see, I think Anthropic is attempting to, given that the Opus series and Claude in general doesn't do image generation, they're trying to lean in a bit to some of the gaps at the intersection between Cogen and Vision. And I, I think in— for, for uses of mine, I, I still honestly, I still prefer Fable 5 even though it's more expensive. And even though if you look at, say, the Artificial Analysis Intelligence Index, if you look at their overall chart of performance versus cost per task, if according to that chart Fable 5 is below the frontier, it's slightly below Opus 5 in terms of their capabilities and a lot more expensive. Despite all of that, for day-to-day usage, when I use Claude, I still prefer Fable-5. But I'm very glad for one thing about Opus-5, which is it doesn't shut you down as frequently if you ask anything that it misconstrues as being a question about biology or a question about cyberattacks.

So Alex, I have the exact same experience, 100% the exact same experience. But then I look at these benchmarks, there's a whole bunch on these charts. Yeah. And they seem to tell a different story. Yes. How is that? How do you reconcile that?

I'm a little bit scared of that. There may have been some mild benchmarking here. That, that's what I was politely gesturing at. These seem to be benchmarks that are involving CodeGen and/or imagery or vision and living at the intersection between them. Some of them, like if you look at Humanities, last exam. There, I granted it's saturating anyway, but the performance improvements are a little bit milder. So for, for HLE, you see from Fable 5 at with, uh, with tools, uh, 63.9%, a modest increase to 64.7% with tools with OPUS 5, and actually a decrease without tools, which is also maybe a sign that there's been a bit of Again, not benchmarking because it's still, I've used it extensively, it's still very well-rounded. I don't want to accuse it of broad benchmarking, but it just, like if you look at the drop relative to Fable-5 for legal or health or some other areas, obviously there was some sort of distillation. This is the type of distillation that is under the present regime welcome and not disdained. So taking larger model and using it to teach a smaller model, a more cost-effective model. There was probably a lot of Fable-5 or Fable series or Mythos series distillation down to achieve Opus 5. But the, the overall sort of distribution of tasks, definitely from interacting with it for a while, feels biased towards cogent and visual stuff and away from general capabilities outside that.

So here's our next chart, agent decoding by effort level. You want to walk us through this?

Yeah, so we're looking at everyone, at least in, uh, in the industry's favorite form of scatter plot. So cost on the horizontal axis, performance on the vertical axis. And what this appears to show is that OPUS 5 is both stronger in terms of absolute score and cheaper— that's the horizontal axis— than Fable 5 and OPUS 4.8. And interestingly it appears to be on the same cost-performance frontier approximately as SALL. So that— I think the subtext that we're supposed to get from seeing this chart from Anthropic is that this is— this should be read as a direct competitor for, for SALL, which is interesting and slightly, I think, unnerving given that, again, Fable-5 anecdotally seems to give better performance.

All right, let's go out to our third chart here, novel problem solving by cost. And I love this concept.

So this is just wild. Yeah, please. Yeah, so ArcAGI-3 again is a challenge that is primarily focused on the ability to solve sort of animated voxel problems. Tetris, for example, if a person had never seen a game like Tetris before with a bunch of blocks moving around and you were trying to do well at Tetris, it's a rough analogy, but that's approximately what ArcAGI-3 is like, animated block, world challenges. So what, what's really striking, and, uh, Dave, you and I have talked about, uh, various attempts by pure scaffolding layer parties to, to just, uh, completely saturate Arc AGI-3. The— according to the official rules, I think there are limitations on how much scaffolding you're allowed to get, and so this is just the raw model being injected in. But what's interesting, and what was I think especially striking in the Opus 5 performance, that this is as relayed by the Arc Prize Foundation organizers is that it was reasoning algebraically about the visual challenges. So it was handed a visual puzzle involving blocks, and it started to reason. This is— if you look at some of the founders of Arc AGI, of the Arc Prize, they will go on forever about how this is actually a prize that tests the ability to do what's called program synthesis, to write programs from scratch in response to new, to novel problems. And so stunningly, what Opus 5 was able to do was to take a visual problem with a bunch of what to humans look like objects, and it represented the objects algebraically in software and basically did math on the objects in order to solve the problem. This is the first time that any, to my knowledge, anyone's ever seen a frontier model ever do that.

A novel approach that was not guided by anybody. This is its derivative approach.

Strategy for doing this. That is, unless Anthropic was benchmarking on ArcAGI-3.

Okay, I think, I think that scaffolding argument though is really, really important because if it holds up— I tried to replicate it, you know, because you can get 98%, I guess, on ArcAGI-3 if you give it a reframing of the way it interprets the puzzle, right? And if that holds up, that gives inspiration to a billion entrepreneurs who can take something like protein folding or drug discovery or, or mechanical design of robot arms and say Fable 5 can do this, or Opus 5 can do this, but I gave it a better way to think about the problem, and now I tripled its intelligence within that domain. So that, that opens the door for scaffolding improvements in all these domains like biotech, where if you can reframe it so the AI doesn't have to work as hard to understand what you're trying to achieve, and can maximize its tokens and its parameter brain count, that is an entrepreneurial heaven. So I'm really hoping that result holds up. I tried to replicate it. I couldn't quite do it. I didn't work on it that hard, but I do believe it's possible. Alex, did you get to the bottom of it? Is it real?

I'm not certain, but the scaffolding advantage is very real. And my understanding is this is why Arc AGI-3 has certain rules regarding what can be submitted and what can't. But I think the elephant in this particular room, to your point, is that scaffolding adds an enormous amount of value at the moment, at any given point in time, over the baseline model. The other side of that is the baseline capabilities tend to dissolve any scaffold. So today's scaffold is tomorrow's baseline capabilities. Well, I tell you, if that holds up—

and I think you're right, I think it will— next semester, every university in the country should have a class called scaffolding. And everybody should have the opportunity to learn how to do this because that is the power tool of all power tools for any entrepreneur. And so what is it now? It's coming up on August. You have 30 days to get your class curriculum together and launch it for next semester with prompt engineering as a prerequisite.

Alex, let's hit these next two charts and then watch the call of duty. Have a quick—

yeah, please. Um, something I noticed was the 4.8 came out on May 28th. And 5.0 came out just now. So it's not that much of a better model, but the efficiency's gone up by twice as much. So we've seen a 2x, and we were saying 10-week doubling price performance for AI this year.

It's 8 weeks now. It's a model release every 6 days on average over the last 6 months. It's unbelievable.

Yeah, this is incredible. The other thing I noticed was in the grid, you've got different models that are becoming really good at different things like legal, health. Coding, etc., which I think will continue.

Alex, these next two charts. Yeah, so this chart is interesting insofar as it seems to support the hypothesis that there might have been mild benchmarking on Arc AGI-3. So this is a chart by a third party that evaluated Opus 5 on an Arc AGI-3 like game involving similar genre and discovered actually the performance jump was, was not material versus, say, Fable 5. So again, not quite sure what was going on with Arc AGI-3, but that was by far the most prominent increase that we saw from Opus 5. Interestingly, a benchmark that Anthropic did not highlight was Frontier Math, which is, I think, maybe in some sense a better bellwether for advanced reasoning capabilities by the models. I had to check this independently. And actually, OPUS 5 demonstrated inferior Frontier Math performance relative to Fable 5. So again, Fable 5, still my favorite clock.

All right, last one here, the live leaderboard.

VoxelBench. So here we see OPUS 5 now earning third place just behind Fable 5 on VoxelBench. Again, visually intensive tasks, but at a much lower price. And interestingly, but perhaps unsurprisingly, Saul from OpenAI still carrying the lead on this. The reason why I'm not that surprised is visually intensive tasks are an area where I would naively expect OpenAI to be doing a better job because they've continued to invest in image generation, whereas we've seen no generative image capabilities at all, shockingly, from Anthropic at all. They're busy maximizing the value, the revenue per token, which leads them to code and not to image.

And I bet you were going to see—

As a practical matter, you know, anytime I'm doing something complicated, working in Fable 5, working in Fable 5, getting a lot done. If I want to see an architecture diagram, I just take the entire thing and dump it over to GPT and say, make me my architecture diagram. Fable 5 is so bad at it. But if you know, it does so much work for you, you get confused very, very quickly and you want to see a simple visual summary of everything going on. It's just so bad. So, but, but GPT is amazing.

My guess is Grok jumps to the top of this leaderboard and the next release. I mean, Elon's He's been speaking about that. Speaking about imagery, this made a viral loop on X. This is Opus 5 recreating Call of Duty from a single prompt. Call it a one-shot, if you would. Let me go ahead and hit play on this.

Remember these demos 30 days ago just looked like absolute garbage? Yeah, this is incredible. It's crazy, the rate of improvement.

Well, what's amazing to me, I mean, this is sort of the converse, for those who can't see, this does look like Call of Duty. The converse of not having native image generation abilities in defense of Anthropic is that if you look at the entire physical world and you say, well, everything is just code, including code that generates photorealistic video games, then you say you don't need native image generation abilities. You just need the ability to generate photorealistic 3D environments like Call of Duty and you're all set. Huh.

Yeah, the truth is in there somewhere. It's kind of in the middle, I think. But I think, I think really clearly Anthropic cares about recursive self-improvement purely and only. And so they'll build anything and train on anything that helps the—

yeah, it's a race to ASI. So I think an important article that I just added for our listeners is, uh, and while we're talking about Claude, I don't know if you heard the story that a significant number of Claude chats were found publicly searchable on Google this past weekend. So a Reddit user discovered that by typing a search operator site:claud.ai/share into Google, it surfaced a long list of shared personal data, including, you know, personal health records, private documents, key names and telephone numbers. Apparently, this originated from Claude's Share Chat feature, which allows users to share links of their Claude chats between, between friends via URLs. Did you track this, Alex?

Yeah, I saw the story. And on the one hand, it's disappointing to see any, anything, any information that would be expected to be private find its way out into the public world. Not a fan of that. On the other hand, I think there's sort of another side to the story, which is a feature that was intrinsically designed to be social in nature. Shocked, shocked to see gambling in this establishment ultimately finding its way into the hands of other people. I think there are two sides that one can see here.

Yeah. But I think one of the issues, one of the arguments, and we saw it on The Rant a few episodes ago, is that when you're using these models, your competition is in some sense seeing your data and learning from your data. And it's something people need to understand. It's the argument for on-prem. Salim?

No, I just double down on the same thing. You've got to do your own. You've got to own your own proprietary data. And I think over time people move everything to on-prem that's sensitive in any way.

Dave, any comments on this before we move on?

Nope. All right. I think it's just incredible, the rate of, the rate of change. Just, just go back and look at an episode from 2 or 3 weeks ago and look at the rate at which one shot can, can create things. And the rate— oh, I'll make one other comment. The holodeck, you know, our holodeck is up and running.

I want to see it.

You got to come check it out, man. It went from like okay to mind-blowing in just a couple weeks for the exact same reason. You can one-shot a world while you're— and the audio and the visual is so good.

The fact that you can one-shot, I think what this will do is— I don't think it affects the commercial games that much, but it allows you to do experimentation in an amazing way because the cost of experimentation just went to zero. Yeah, so you'll get so many more players.

Jarvis is coming so soon. Yeah, yeah, that, that entire— God, that Iron Man, the vision in that movie was so precious. But it's going to be exactly like that. And fun.

I spoke to Jon Favreau today, the producer of Iron Man 1 and 2, getting him to come to Moonshots Live. We've got amazing, amazing— Yeah, yeah. Elon had introduced us. So let's move on to our next story. This was a story, Alex, that you wanted to raise here. So it's the idea, which is obvious, that, you know, global AI diplomacy is coming. So the Financial Times is reporting that China's leader, Xi Jinping, is wielding AI as a tool of statecraft, using it as leverage in China's diplomacy across the Global South in a strategy that the Financial Times frames as Pax Silica. I love that. While Washington is debating open versus closed, Beijing is out in the world country by country exporting AI as an instrument of influence, offering models and infrastructure to developing world that wants to leapfrog what they currently have. AI is becoming an instrument of soft power. Whoever supplies the models and the infrastructure to the developing world shapes the next few decades, I would say the next century of global alignment. So, you know, my concern is if the US over-restricts, the developing world is simply going to adopt whatever frontier-adjacent open models are there. Celine, your thoughts? I can't stress this enough.

The whole power of the US is its open and very broad innovation ecosystem. If you create a restrictive open model policy, it's going to be strategically like a self-own and shoot your own foot of an epic level, because you're going to protect a small number of domestic labs while giving the entire opening ecosystem, Global South AI ecosystem, to China. Um, if you want leadership in an exponential era, it has to come from the largest network where everybody's using your tools for stuff, not protecting its strongest incumbent. Openness is not a— it's not a philosophical preference anymore. It's like a— it's the tool of soft power. And the US has already lost that in diplomacy and USAID and other stuff. If they close up the open model policy, it's going to be really disastrous for the future.

And I don't think they will. I mean, I think this is— I think we're effectively splitting the We're splitting the world into—

we should be having the discussion though. Empires. Yeah, yeah.

Alex? Well, there is, there is no—

as of right now, there is no U.S.-based option at all. You could take everything you just said and swap out the word model and put in fighter jet. Like, should we sell F-16s to XYZ country? It's a no-win question. Like, you have to pick and choose. But if we don't sell the F-16s, they'll buy Russian and Chinese fighters, and that'll support the creation of more of those fighters. Yeah, but you're also selling an F-16 to— like, it's exactly the same problem. There's no easy answer to it. But right now, there is no U.S. open-source model to compete with the Chinese anyway, which is kind of sad.

Alex, your thoughts? I, I, I think so.

PAX Silica had already been announced by the U.S. before China announced its own initiative. And China, of course, announced many years ago at this point— Xi Jinping announced Belt and Road Initiative. And there's a certain extent to to which it's far more— I don't want to say insidious, but far more ultimately invasive and controlling if a foreign country— say if a foreign country loans you a bunch of money to build a bridge, okay, so you default on the loan, that has a certain outcome. Foreign corporation that's basically under the thumb of a foreign government builds telecommunications equipment and deploys it to you. So now you have cell phones. The worst that they can do, they can spy on you and they can shut off your telecom infrastructure. Next level up, foreign corporation that's, uh, that, that's heavily involved with foreign government injects superintelligence into the veins and arteries of your country. Now it's not just listening to you or not just loaning money to you, Now it's thinking for you. And I think that's a far more vulnerable position for the so-called Global South to be in, regardless of which bloc or sphere of influence it finds itself in. And I can only imagine that the long-term, to the extent there is a long-term in the middle of the singularity, equilibrium point is going to be pushing more, not just inference to the edge, which is what China, I think Chinese frontier labs, would like with open weight models, pushing training to the edge. That, I think, is the equilibrium point. And curiously, I don't hear that many countries in the so-called Global South agitating for domestically pre-trained models. But I do think that's where some sort of equilibrium could lie if there is to be an equilibrium.

Can I tell a related story here? Please. A few years ago, I was talking to the prime minister of one of the smaller Asian countries, and they had their big city had tripled. They needed much more ports to be able to receive more containers for the big, huge population. And they'd just taken a half a billion dollar loan from the Chinese and totally, totally mortgaged the future of the country. And I made the point that, look, drones are kind of doubling in their every 9 months in their price performance. If you waited a few years, you could have a drone pick up a container. You don't need a port. You could drone, you've got a fork drone drones pick up from the corner of a container, which is average 20,000 pounds. And so you don't have to wait that long for drone doubling to get to a quarter of that weight. And then you just pick it up and put it on a flatbed truck or on a rail car and off you go. And they're like, damn, we just mortgaged the entire country because we didn't understand exponential thinking. And if you go back to what Alex just said, that goes up 10x when you outsource your thinking. And that's really dangerous. I think the safety and the security The future will be in these open-world models that give you back your sovereignty. Yeah.

Talk to California about its high-speed rail when you talk about— Yeah, let's not go there. All right. So big news this week for my fellow space cadets, a successful launch of Starship 13. SpaceX has confirmed launch and splashdown. An incredible, incredible trip it made for Starship 13, their largest vehicle to date. Tonight. It accomplished a number of key firsts. Let's run through them. First, it deployed 20 operational Starlink V3 satellites. They were connected to— they were tested in part, and these are the satellites that are going to deliver us a half a gigabit to a gigabit connection speed every place on the planet. You're literally going to have more— a better connection from space than you have from, you know, your home Wi-Fi. They did an in-orbit relight of one of Starship's Raptor engines. Critical for the upcoming Artemis missions, and a successful soft landing on the Indian Ocean with the vehicle remaining intact, which was extraordinary. I'm going to watch two of the videos here. Let's share them because they're just fun. This is space porn. All right, let's take a look at the launch first. It's important. You gotta love this drone footage from above the launch at Starbase.

Space is big.

This is such a beauty, you know, it has a high degree of beauty. Oh my God, so gorgeous. Yeah, I remember when I was with Elon, we were talking about, you know, the, uh, Starship first stage and saying it's the most contained energy that you can ever experience other than a nuclear explosion. All right, and, and very importantly, the landing, which was the big news on this particular mission. Let's take a look at this sequence.

About 10 seconds away.

I mean, look at that footage.

Let's see if we can get this thing in the water.

I got, I got the link from you guys and I was like, yeah, yeah, I'll check it out. And it's like, it's like, wow, I gotta watch this whole thing end to end.

Note the high rates. Look at— can't you—

the amount of stress you can feel it, you know, from these because you can see like things— look at that landing and warping.

Soft landing down to 1. Floating in the water for full recovery.

Soft splashdown in the water.

This is where they thought it would explode.

Well, it has in all the previous missions. What happened? It was so soft. Yeah, well, it's, it's—

there's fuel, okay, excess fuel.

Yeah. And when it, when it hits the water, you know, it punches little holes or whatever in the sides.

So that's where it explodes. You know what I'm excited about in particular is Elon tweeted that because the landing was so precise that Flight 14, he's likely to capture the Starship on the Mechzilla device, right? The large chopsticks that come in and grab the vehicle. Yeah. So I mean, that's a flight I want to go to Starbase to watch the return. It's gonna be awesome. Let's take a second, just talk about the abundance story here. I mean, the cost of launch is plummeting. And let me just give you the numbers real quick. The space shuttle was roughly $54,000 per kilogram to orbit, right? So think about, you know, taking a gallon of water, of milk to orbit, $54,000. Falcon 9 dropped it to about $2,000 to $3,000 per kilogram. And Starship's target, and Dave Erickson and I were discussing this with Elon back in our January podcast, It's between $10 to $100 per kilogram. I mean, just extraordinary.

Well, it's also— it's inspiring, the amount of incredibly cool stuff that you can build now, specifically because all the feedback and control and all the remote, you know, intelligence is easy now all of a sudden. You know, we saw the Unitree robots in the last podcast, and they're just beyond cool. And that robot that, you know, I plummeting down the side of the mountain with the wheels. Yeah, like crazy cool. And then you got the, you know, the Starships, which is the amount of possibility is so exponentially bigger than it was just 5 years ago.

And all the data is on Grok for building Starships, which is extraordinary.

Also, you know, I'm working on a photonic computer, you know, to run our new neural nets and all the parts. It's actually giving me part numbers to order and saying, this vendor in Germany will make this lens for you in exactly this way. Can you— do you want me to write up the specs? Like, sure.

Can you order it and build it for me too, please?

I mean, literally, can they get it arriving boxes? I still have to open the box and assemble it.

If you're an entrepreneur out there and you've been looking for where to go build, I mean, building hardware, you know, Ben Horowitz, who's a friend of the pod, and we're going to have him back on the pod for one of these episodes. He's the co-founder of Andreessen Horowitz. Dave Erickson, you know, he wrote the book, you know, Hardware is Hard, or effectively, and the hard things— was it The Hard Things About Hard Things? You know, and it used to be that building anything— I built robots in high school and college, and it was tough. Now you can 3D print parts, you can iterate rapidly. So if you're an entrepreneur looking for something to do in the world, you know, What's missing? What do you wish existed? And you can actually use these models to design it, order the parts, and start building.

Yeah. Yeah. I've got an incredibly cool robot that's skimming the top of my pool. It just has eyes and it finds leaves and it just goes and picks them up. But I want someone to build one that dives to the bottom and just goes down, picks up whatever, an acorn, brings it up, throws it out of the pool. I bet you could vibe that up now.

I just, just crank it. Alex, if there's anybody who's as big a space enthusiast as I am here, it's you. I mean, did the flight bring tears to your eyes?

Oh, I wasn't crying, but space is big, and I was delighted. Did you see, Peter, the views from the Starlink satellites that were posted later? Like, that, that was pure science fiction.

Oh my God, it was, it was basically, uh, looking at Starship in orbit from the descending booster from a distance.

Yeah. Yeah, that was incredible. That's like something out of Star Trek or The Expanse. I was very impressed with that. I certainly hope that this Flight 13 ends up in a museum at some point, given the soft splashdown. This is a historic first. Hopefully the SpaceX team will use this as an opportunity to get a good look at the heat shield, which is interesting. If you noticed, like, the drone feed the moment the splashdown happened, it was just zooming in on all of the heat tiles looking for damage, trying to analyze the structure, presumably because the team was worried the whole thing might explode a few seconds later. So they were getting whatever footage of the heat shield that they could while they had time, but now they're going to have a ton of time.

So it's very exciting. In the past, after the vehicle, after Starship, you know, detonated from the onboard fuel, and that was expected, you know, people say, oh my God, it failed. No, it didn't fail. It did exactly what they expected it to do. They'd have to go diving and find pieces of it to try and reconstruct what happened. And the heat shield here, I mean, people need to understand the amount of energy being dissipated. These vehicles are traveling at 17,500 miles an hour in orbit, have to dissipate all that safely and come to a precise landing. It's insane.

Salim? I just love the fact the ongoing flight after flight, he's viewing any kind of failure as information. And you don't get embarrassed by it. You take the data, you learn, and you do it way better next time. And nobody else does that.

Yeah, it's exciting. We are, as Alex, you've said many times about the speedrun Star Trek, what an exciting time. Count on it. Yeah. I mean, the only thing better is if we discover that we have access to all the alien UFOs and we can go to jump to light speed. All right. Let's stay on the science theme. Our next story is in the world of brain-computer interface. Two stories this week. The first one comes out of Science Corporation. Full disclosure, it's one of my portfolio companies. I love this company. And it's restoring vision for the blind. The company is run by an amazing entrepreneur, Max Hodak. He's the past president of Neuralink and someone who I've had on the Abundance stage a number of times. Science announced that his first product called Prima, P-R-I-M-A, I'm sure it's an acronym, has been approved for launch in Europe. So Prima is the first BCI device approved for restoring detailed vision in age-related macular degeneration that destroys the central vision of your retina. They just earned what's called a CE mark, which means it complies with European safety, health, and environmental requirements. Let me show an image of what this looks like, and we can talk about how it works. So here it is. What you see there is a pair of glasses that are capturing the image, and then they're beaming back the image via infrared to that little, call it, you know, rounded square that's sitting behind your retina. So the signal from that Prima implant behind your retina is turning it into electrical signals and giving it to the remaining retinal cells, and those then get transmitted through your optic nerve to your visual cortex, and it basically restores your central vision. The amazing thing is the patients patients who've gone through this have experienced 5 lines of improvement on a standard eye chart after 12 months. So this is a godsend for so many people with macular degeneration. Any thoughts here, Alex? Yeah, so a few things.

First of all, the overall setup is in the spirit of, as I've commented in the past, the singularity is essentially all sci-fi tropes happening everywhere all at once. This is reminiscent of an ocular implant from the Borg, quite literally. It has an external module that, for those who are watching, can see here. So an external camera that then captures the information broadcast in human-invisible near-IR to this chip that sits immediately behind the retina. It's interesting insofar as Max, who was of course basically running Neuralink previously, this is a much— even though the retina is is part of the central nervous system. This is a move away from the brain. But Neuralink, which also has its own approach for curing blindness called Blindsight, seems to primarily be focused on injecting less on, uh, on the optic nerve, more just focused on direct brain stimulation, direct brain intervention. Yeah, it's interesting to me that Max, with this new venture of his, is sort of moving away from the brain, albeit still in the CNS. And I do think the farther, the further you get away from direct brain intervention and placement of electrodes, the easier it is to scale up a mass-market consumer device. In this case, obviously, it's obviously surgery on the retina, but I'm very optimistic that advances in and ultrasound advances and wearables, a variety of other completely non-invasive advances will enable vision restoration without even needing to have retinal surgery in the next few years.

You know, I wrote a chapter about Max and science in my book, "We Are As Gods." And in particular, you know, giving vision back to the blind is biblical, right? It's huge. It is amazing. And God said, "Let there be science." What he's doing, And so, you know, it's interesting. Prima as a product is his stage zero revenue generating engine. So one of the things that a lot of entrepreneurs do, do incorrectly is they jump straight to this, you know, massive moonshot that will take them, you know, hundreds of millions or billions of dollars to get to without generating early revenue. So Prima is the means by which he's creating early revenue. He has an amazing BCI approach. I can't say a lot about it, but he's basically growing neurons into the brain, which don't— you know, the issue with Neuralink and many of the BCI companies is their electrodes destroy thousands or hundreds of thousands of neurons when they're placed into the neocortex. But neurons actually can grow into the brain. So he's got an approach of an interface between electrical circuits and neurons and neurons growing into the brain. And then, you know, wiring together and firing together. Hopefully he'll disclose it. It's been in animal models and his plan is to get to humans, but it's an incredible strategy for the BCI world.

BCIs are super competitive at this point. There are folks trying approaches at the CNS level, at the peripheral nervous system, direct brain stimulation, wearables, ultrasound, around fMRI, and I think ultimately— by ultimately, I mean on the 5- to 10-year timescale— we're going to see something of a shakeout, and we're going to discover what are the most ergonomic ways to interface with the brain. So I really hope that for Max's sake that brain— or that science, rather— does well, and I, for one, would welcome some extra neurons.

Yeah, he thinks of it as an extra— you know, the corpus callosum what connects the right and left hemisphere of your brain. Imagine having a third hemisphere of your brain that actually is connected to the cloud. I mean, that's the way he describes it.

Exocortex. I want my exocortex.

It's extraordinary. Selim, you were saying?

There are two things here that I found really interesting. One is the feedback loop, right? Once you have a feedback loop from sensory back into the brain, et cetera, it learns very quickly. And I think we can see that over and over again in some of these recursive inner loop type applications. The second thing that occurs to me, which goes to your comment, Peter, about business models, is when you have an exponential and it's hard to predict out where the endpoint is going to be, it's not that difficult to look out 2, 4, 5 hops and say, what are the business models that may be enabled at each of those things? What are the use cases? So if you're an entrepreneur and you see a technology that's growing exponentially, now there's a dozen of them, you can pick your biggest passion, your favorite technology, look out where it's going, and then say, okay, at that price performance, what applications become enabled? And now you have a very viable roadmap for the future, exactly like the way Max is doing it. And that's going to be the future of companies and how they evolve.

I really want to echo something Peter said there too about go-to-market strategy, because, you know, if you look at the most valuable companies in the world, so like an Apple, Google, Meta, and you look at their very first day and their very first product, you know, the Apple I was a box of chips you needed to assemble yourself. And, and, you know, Meta, you know, was Facebook.

It was, it was a little face—

one university sharing thing at just Harvard. Yeah. And, and Google was a plug-in to Yahoo. It's a little search plug-in to Yahoo. They tried to sell themselves to Yahoo for a couple hundred million bucks, and Yahoo's like, you're not worth anything near that. And that's the reason—

yeah, he was right. It's not worth anything near $100 million. Yeah.

But these, these really humble beginnings for go-to-market strategy, because, you know, we talk a lot about, you know, Starships landing in the ocean, and people think, wow, Wow, I want to build a company that, you know, creates a new Starship. But that's not how these things get started. You got to start with a go-to-market that actually gets you an initial revenue. And if you really study the outcomes but trace back to the first 6 months, which not enough people do, like really study the details, the people, the characters, what exactly was that product, and then that starts you on the journey and you end up being Google, Apple.

You know, we had Bill Gross on stage with us this past year at the Abundance Summit, and he's brilliant, and I, I love Bill. He's one of the most extraordinary entrepreneurs, has created more startups that have gone public or been acquired than I think anybody else, period. And he has a great video on DLD, also on TED, which he looked at, I think it was 50 companies in his portfolio that succeeded and 50 companies that failed. And he asked the question, why did they succeed? Was it because the CEO was smarter or trained at a because they were, you know, exclusive Harvard or MIT? Was it because they had more money? Was it because— what was it? And his conclusion at the end is an important lesson for all the entrepreneurs listening. It was timing. It was the companies that were there at the right timing that were able to survive long enough to survive forever would intercept good luck. So classic examples are Uber and Airbnb had been tried before.. But when they launched in 2008, it was just after the recession and people were looking to make money. They were willing to rent their bedroom out, willing to go and drive a car. You know, our darling here, SpaceX, I mean, you have to remember Elon in 2008 was effectively bankrupt. He had had 3 failures of Falcon 1 and the 4th one, which he scrambled to get money together, finally succeeded. And because the space shuttle had been shut down a couple years earlier, there was a contract out and he won a billion-dollar contract in the crew resupply from NASA and that got him going. Timing is everything. So if you can have an early revenue stream for your company that allows you to stay in business and intercept good luck, that's one of the single most important things.

Yeah, just survive. Just find a way to survive.

Elon was 18 years. 18 years from bankrupt to trillionaire. Amazing. But I mean, it took, yeah.

Did you see his tweet? Past trillionaire or whatever it was, former trillionaire. Little plug here.

Yeah, please. I mentioned this earlier, but we talked just now about the feedback loops. John Hagel and I have come up with a framework where it allows you to measure luck. And so once you have that feedback loop, that becomes very powerful. Yeah, we'll bring him on sometime and talk through it.

It'll be useful for the viewers. Everybody, welcome to the health section of Moonshots brought to you Fountain Life. You know, we talk about AI on this Moonshot Podcast all the time. One of the most important things AI is going to be able to do for you, besides educating your kids and helping you with your taxes, is making sure that you're living a healthy lifestyle, that you get a chance to get to 100+. I'm here today with Dr. Dawn Musalem, the Chief Medical Officer of Fountain Life and a part of my medical team. Dawn, a pleasure. Great to be here. You know, the thing that people are concerned about most about living to 100 or 120 is their cognitive abilities, making sure they don't have dementia. The numbers about dementia are problematic. Can you share what you've learned?

Such an important point. You're right. At Fountain Life, our members, the number one thing people are most concerned about is losing their brain health— forgetting the name of their child, forgetting the face of their loved one. We know that when it comes to dementia, the conservative estimates are that 45% are entirely preventable. What was amazing is with the advanced testing we're doing at Fountain Life, 1/4 of our members had advanced brain age. Wow. But what was really awesome is, again, back to that prevention, when we partnered it with healthy living— this gives me chills— eating healthier, moving our bodies, sleep. Optimizing sleep is so important. You know what we saw? We saw that we improved that brain age by 26%. That is a big, big number to show that the majority of those individuals were able actually to improve the brain age.

And one of the things I love about Fountain is we're searching the world for the best therapeutics, the best approaches, and making sure we bring it to our members. So if having healthy brain function, uh, till 100, 120 is important to you, check out Fountain Life. Go to fountainlife.com/peter. Make sure you become the CEO of your own health. All right, now back to the episode. Alex, this next story is one that, that you, uh, threw out, and happy to talk about it. It's our second BCI story. This one's out of Neuralink. The company just shared a video of people living with paralysis controlling a powered wheelchair using nothing but their thoughts. No joystick, no hand controls, just intention translated directly from the brain into the wheelchair's movements. I mean, think what this actually means. It's massive freedom. Let's roll the video and, and take a look at this because it's, it's a beautiful thing, and we'll talk about what comes next.

Through our clinical trials, we've been working really hard to give the world a brain-computer interface powered wheelchair. It's designed for anyone with trouble controlling their wheelchair physically by translating their neural signals so they can control the wheelchair with their mind. What we've done here is we've developed a set of custom electronics to take cursor movements from a participant's imagined motions, translate them into analog signals and use them to directly drive and control all the functionalities of the wheelchair. This is the, uh, wheelchair control app, and we built a custom UI for our users. I can now move my cursor up and slowly move the wheelchair forward or turn to the side, left or right. And as I get more and more into the ring, it'll go faster and faster. Really, we built this system with safety in mind. As you can see, if I let go, the cursor is slowly going back to the center. So that way, if a user ever becomes incapacitated, you know, they won't go driving directly into a wall. The cursor will allow them to go back to the center here.

Alex, your thoughts and what comes next?

Well, obviously this is sort of a visual joystick via brain-computer interface. Where this goes, it's hard not to extrapolate this going to full bodily control. Imagine giving people exoskeletons that they can control via BCI, paraplegics, quadriplegics, and basically restoring free autonomy in, in a physical world, all four limbs. I think that's pretty incredible. It's easy to extrapolate further than that. I think there's probably a sizable subpopulation in many countries, certain folks would love mechas, from, uh, from anime, would be able to walk around in, in large robots, Sigourney Weaver, Alien, Aliens style. Or maybe even— I mean, I think the end game for the motor cortex does look like some variant of partial brain uploading, uh, or something adjacent to that. Once we've fully decoded the motor cortex, we're, we're in a, I think, a stronger position to I think some variant of human mind uploads could just be behavioral uploads that are generated by pre-training a foundation model off of large amounts, say, of fMRI or ultrasonic data, and being able to decode the motor cortex to perform useful functions in the world. I— that's a low bar. There are much higher bars that would be connectome-based. Something like that, I think we're going to see actually happen in the next 5 or so years, by the end of this decade. And I think that's a major step toward, you know, in the short term, obviously taking people who can't walk, giving them powers of locomotion— that's now-ish. But in the next few years, it's giving exoskeletons and ultimately human uploads.

You know, my, my answer to that, AWG, is that Neuralink connects these individuals to an Optimus robot, right? And they see through the Optimus eyes and hear through the ears, and then they Avatar, your vision is Avatar. It's effectively telepresence. You know, you can be anywhere. You know, you can, you can inhabit an Optimus in Japan if you're sitting there in Boston. I think that is definitively coming, probably over Starlink. I mean, the singularity is here. This is insanely fun. Yeah, there were a few.

I mean, in addition to Avatar, there was another— I'm blanking on the name of that other sci-fi movie where people people never left their homes and only went out and interacted with each other via these telerobots. I think every sci-fi scenario plays out at once, and I can guarantee you, Peter, if the scenario that you're describing comes to pass, and probably will, we'll find some country will, a few years from now, will be regulating the hikikomori, if you will, who only stay inside their bedroom and only interact with the outside world via BCI to telerobot. A lot.

So, so crazy, so fun, and so liberating for so many people. Uh, all right, our final story before we go to our AMA here is, uh, regarding Elon's prediction that AI and robotics is going to make money irrelevant within a decade. By, you know, 2036, he said, it's his post-capitalist vision. Uh, it's a world of radical abundance in which scarcity no longer matters, money no longer matters. I talked about this with him when he was at the Abundance Summit this past year. Let's take a listen to this video, and then I really want to discuss this one because it has people both excited and fearful. I want to address the fear there. Money won't matter in 2036.

I'm not sure that the people who bought your shares think that money won't matter in 2036.

What do you want money for? Are you— you want money for goods and services? Well, if that is so abundant that, uh, there's more— that the robots and AI are providing more goods and services than you—

than any human could possibly consume, what do you need money for in that case?

Make a prediction, which is that deflation will be the issue, not inflation. Deflation. Because, because as the output of goods and services increases, um, if the output of goods and services increases faster than the money supply, you will have deflation.

I think it's totally ironic this was an interview by The Economist, uh, of the world's first trillionaires. Yeah, gentlemen, thoughts? Salim?

Um, you know, I think the— you're— there's a couple of different threads here, right? Thread one is we're reaching abundance and the cost of things will drop radically. But abundance in production doesn't mean you end up with abundance everywhere, because you could, you could have eliminated scarcity and ownership and access and location, and that would be amazing. But money is going to be relevant as long as you need an exchange mechanism. So as long as you need that to allocate stuff that's scarce, money as the means of exchange will stay. Let's remember, there's 3 uses for money. Means of exchange, unit of account, and store of value. And so this would hit store of value to some extent. It would hit the means of exchange to some extent, but the exchange unit will become really— or unit of account, et cetera. I think the biggest challenge here is not that abundance is impossible. It's that if you end up in the wrong way, abundance will get captured by a few big companies, which is where the wealth inequality has been coming in. I think the hugest opportunity as we you distribute and democratize— Peter, back to your words— technology, you also democratize opportunity. And I think the best framing I've seen for any of this is, can we get abundance of opportunity? And I think that's where technology will take us.

Yes, Alex, you've made that point, you know, abundance of freedom.

Yeah, I construe Elon's comments as— I'll use the technical term— Star Trek economics. I think he's arguing that 10 years from now we'll live in a Star Trek economy where— and I think there's some fine print on this. I don't think he really means to say everything has been demonetized. I think what he really means, I think what he's shorthanding, is that most aspects of daily living as we would construe them today in 2026 will have been demonetized 10 years from now. 10 years from now. So food, shelter, healthcare, utilities, education, entertainment, all of these things will have been demonetized and you won't need money because we'll be living in an abundant, in that sense, future. But I think there will be many things still that are not so abundant that they've been demonetized 10 years from now. Like, I don't know, if we have the ability to go to another star system, probably still somewhat expensive. Spend a week on the moon, maybe there's some price there, or just owning some scarce resource that's antique collectible. This is not investment advice, but there are some things I think that will resist demonetization for longer than 10 years. If I had to put my finger to the wind and guess when demonetization hits the total economy, don't hold me to this and not investment advice, doubly so for anyone investing in 30-year treasuries I would guess approximately 30 years out.

A funny story, AJ Scaramucci, who was one of my interns, one of my Strike Force members, started a company collecting, uh, obviously scarce things like Tyrannosaurus skeletons and Pokémon cards.

I know AJ pretty well. AJ, if you're watching, I'm not sure what's up with those Pokémon.

Yeah, I mean, he's collecting the best, you know, the best first edition comics and so forth. I I mean, it's an interesting strategy, but, you know, I wrote a piece about this after Elon published it, or after we had the conversation with Elon. And the best way I think this works is we're going to end up providing some level of, of UBI. I call them COVID checks, right? $3,000 a month, which today gives you a bare minimum level of living. But all of a sudden in this scenario, AI is today already and will be in the future the best position An Optimus on AI will be the best surgeon, and the cost of that is CapEx and electricity. And then, you know, autonomous vehicles, we're going to see not just one or two, but a dozen car-as-a-service AVs, right? Beating each other out to bring the cost down. There's a— so all of a sudden, $3,000 goes a lot further than ever before. You want a house? Great. You know, a fleet of Optimus robots will build it for you. So it's a massive demonetization in the future.

There's a monster elephant in the room though, which is the radical transition this is going to require in our fiat currency systems because all our fiat currency systems are absolutely dependent on scarcity. If you move to abundance, we have a huge challenge. We've mentioned this before on the pod. This is Jeff Booth's observation. We actually should have Jeff as a guest sometime. He made the point that over the last 50 years, every dollar increase in GDP has come with a $4 increase in debt. Okay? Now, and I use a metaphor for this. So imagine you have a— decide to build a TV factory. You borrow $10 million to build that TV factory. And your business plan says, I'm going to pay this back if I can sell the TVs at $1,000 each, I'll be able to pay back the loan. Problem is that a year later, that $1,000 TV can only be sold for $500. And a year later, it can only be sold for $250. You're never paying back the $10 million. And this is how we're growing the global economy. So this is the printing money problem that we have, where we're just radically printing money to keep the whole thing afloat, which is why assets are so important to own rather than cash, et cetera. Cash is deflating at about 14% a year. This is going to require a wholesale shift in how we measure the economy, which is why people are pointing at crypto and Bitcoin and other things.

You know what I find?

But this is the part that's going to kind of cause a massive problem for every currency in the world. And every central bank in the world is kind of panicking right now because your only resource is to print money and then you have inflation. And they're like, oh my God, we can't have inflation. And so this is a circular wheel that can't be gotten off of. And this whole thing's going to come to a fall.

If only the Federal Reserve had access to the same superintelligence that the rest of us did, they could design superintelligence —fiscal and monetary— That's what you're gonna have to do. That's what you're gonna have to do.

That's a great point.

You know what I find incredibly interesting is that all of these AI visionaries—

Elon, Demis, Dario— they all have played Civ, Civilization. They all speak in terms of Civ, and they've all read Iain Banks' The Culture series. And The Culture series, the entire book series, is about the post-abundance world and what it'll be like. So when they get together and brainstorm on the future, they're so totally on the same page about how this is going to work. So then they do an interview, like Elon does an interview with The Economist or whatever, and he says, 10 years from now, so, you know, 2036, money won't matter. And they go, oh my God, does that mean the exchange rate between the pound and the euro and then the— like, like, yeah, no, we're gonna make so much stuff and have so much abundance that it won't matter. Did you understand the implication? Like, who gives a crap about the exchange rate or the deflation rate? Like, the degree of change that's coming over that decade is so massive that you just mentioned one little aspect of it, like we don't care about money. It's just a tiny little component of this overall massive change. But all those guys are on the same page because they've all read the same sci-fi books. They've all— like, I get how this is gonna play out, and there are different nuances to it. I'm not saying everybody agrees on every part of it, but what we're talking about is you don't care about the cost of things because you just take them off the shelf. Every—

it's ironic given, Dave, that Iain Banks is Scottish. Scotland produces some of the world's best sci-fi writers, so it's interesting that The Economist, based in the UK, doesn't quite appreciate Scottish sci-fi.

And then this tweet exchange occurs. Let me just read it. Daron Acemoglu is a Nobel laureate in economics, and he says, I propose a proposed challenge for Elon Musk, an opportunity to put your money where your mouth is. If money won't matter in 2036, why don't you pledge to donate your current wealth of approximately $1 trillion to charity no later than 2036? This would establish with great credibility your proposal. Goes on. And Elon responds, I'm actually going to do something along those lines. I thought that was pretty cool. I immediately texted him and say, okay, let's launch 10 $1 billion X Prizes to solve the world's biggest issues. Haven't heard back from him yet, but hopefully soon.

I think he's planning something more along the lines of SpaceX, Tesla, stock for everyone via UBE.

Perhaps, perhaps. Anyway, you know, this is the abundance story writ large again, where the cost of everything, and it's not, you know, gonna be trips to the moon or Mars, but if you want your basics, right? It's raising the floor where every man, woman, and child on the planet has access to food, water, energy, healthcare, education, and freedom. I think that's what we're building here. And I like to say, yes, we're gonna have trillionaires living on Mars, live on Mars forever. But in that inflationary world where everybody's, you know, the rising tide for everybody, I'm okay with trillionaires living on Mars as long as every man, woman, and child on the planet has access to all the basics. That's a more peaceful world.

Can we lift the bottom?

Yeah, lift the bottom. Yeah, the gap will get bigger. You know, I agree the gap will get bigger. And yes, but as long as the floor comes up, that's the single most important.

Can I mention one of my favorite abundance statistics that you put up, Peter? Sure. If you go back 200 years ago to 1820, 94% of humanity lived in extreme poverty, being defined as $2 a day on 2011 parity dollars. Today, that number is less than 9%, and you just don't see stuff like that in the news.

Yeah, you don't. The news media delivers every murder, every crooked politician over and over again into your living room between 6 PM and 8 PM As I like to say, I tell my mom this all the time, Mom, turn off the news. Don't watch the crisis news network. It will just give you a bad mindset.

Join us in our echo chamber.

Yes, she does every time. Hey, Mom. Let's do some AMA questions, gentlemen. I think we have some fantastic questions this week. OK, Alex, let's begin with you.

Yeah, so there are a few different interesting questions here. I'll pick number 4. Because I've commented on this one already on the pod. How far off do you think we are from hitting longevity escape velocity? And this is from Sage Freeman 9260. So I think Peter, if you were to answer this, or if friend of the pod Ray Kurzweil were to answer this, I think the answer would be something like 2030 to 2033, I think is the consensus.

My mantra is LEV by 2033. Is.

Yes, uh, if I were to answer this, I think it's going to be spiky, just like superintelligence is spiky along different dimensions and with different skills and capability areas. I think some subpopulations may hit longevity escape velocity by the end of this year. I think others may— it could happen by 2030. But I, I think there are so many variables that will lead to high volatility or spikiness in terms of who arrives when, in part because there are so many people who qualify for certain medications that, you know, say third or maybe even soon fourth generation GLP-1 RAs, those could, speaking hypothetically, it's not medical advice, those could end up having profound longevity impact. So actually I was sufficiently interested in this that I did my own internal mini research study trying to answer the question, have we achieved longevity escape velocity this year? And there are a few confounding variables because you can achieve, in some sense, catch-up longevity increases if something terrible happens. Like if there's an agricultural revolution in China and a lot of people die, then average life expectancy takes a huge dive, but then a few years later it zooms back. And you could ask the question, well, is catch-up or regression to the mean longevity escape velocity? I don't think most people would consider it that. On the other hand, if you have someone who has some illness, uh, but it's an illness that we've never been able to cure before, and now we're able to treat it, and now their, their life expectancy is increasing by almost or approximately 1 year per year, has that subpopulation achieved longevity escape velocity? Some would say no, because you're just helping a person with some illness regress to the mean. Others, including myself, would say aging is a disease, uh, and, and so curing aging is, is basically helping a subpopulation, which is to say more than 150,000 people per year dying on this planet, and helping basically the subpopulation that is the entire Earth population survive and get treated from the disease that is aging. So yes, I think some subpopulations are approximately there right now, and more to come.

And we discussed in the last pod, there are a number of ongoing partial epigenetic reprogramming experiments going on in humans today, which is super exciting. I'm going to take number 3 before one of you guys grab it. I'm sure you're going to do that.

That's yours. If average life— I thought so too.

Average lifespan hits "120 and infertility gets cured, which it will, how does society handle the population boom?" And this is from @MrGanoushington. So here's the, here's the reality. We do not have anywhere near an overpopulation problem, even if we start getting to extreme longevity, 120 and plus. The majority of the world is in a population crunch. We're seeing in Asia, Canada and Europe, you know, a reproductive rate of under 1 child per family, right? The— to keep the population, you know, without growth or decrease, it's 2.1 children per family. Places like South Korea and Japan are hovering at like 0.6 children per family. So we have an issue in a number of generations. These cultures, these countries are going away. The other question that this person might then pose is, "Okay, what about access to resources?" And over and over again, even if we have a population— I think the numbers are going to reach 9.5, 10 billion and then very rapidly decrease. And people have always said the One Earth precept of we need to divide the resources of Earth equally amongst everybody. Well, every time we think there is a scarce resource, we discover, no, it's not scarce. We just innovate around it. Lithium was thought to be a scarce resource. So we start discovering lithium deposits every place. Then we start inventing batteries that are better than lithium with sodium that is much more abundant throughout the world. By the way, a quick note, our next podcast is gonna be with Ramez Naam going in deep dive into energy. Yeah, it's gonna be amazing. It's going to be amazing. You do not want to miss the Ramesh Nam episode. So, um, Mr. Gannushington, uh, no fears about overpopulation and no fears about not having sufficient resources. Salim, over to you, pal.

Uh, okay, um, let me go with number 2. Uh, could Anthropic hide its model's reasoning to stop competitors from distilling it? So you can hire, you can kind of restrict the visible reasoning and hide it a bit to make distillation harder, but you can't eliminate it because you can still, people can still learn from the inputs and outputs and then reverse engineer that across if you have a sufficiently large number of examples. The capability diffusion is very difficult to stop permanently because once you have useful behaviors, people are gonna learn from it and then they have a huge incentive to reproduce it in other models. The sustainable mode is going to be kind of the— not just the hidden reasoning, but it's going to be the whole thing. Do you have unique data? Do you have infrastructure and compute? Do you have users? Do you have feedback loops? We keep talking about feedback loops. Do you have that to provide proprietary and unique learning loops? And that's the really big deal. And then distribution and the ability to learn continuously from all of that. So there's a whole system approach here that where it's going to be the future of every organization is going to be that kernel of data compute, learning loops that then can compound on each other.

Dave, over to you.

There's one left. Hey, do I get first pick? Sure, I'll give you that for sure. Awesome. Thanks. So what's left? Number one, how long does it take to actually take a patch or take— how long does it actually— sorry, let me start over. How long does it actually take to patch a vulnerability like the Hugging Face breach? And that is from @teachme. T3HME. Teach me. Coolest thing about this question is actually your handle. That's really awesome. It only takes a minute once you know the vulnerability. Sometimes there's a little bit of time to propagate out the patch, but once you know what the vulnerability is, you can fix it in no time flat.

And that's all there is to that. So. All righty. Let's move on to the next 4. Dave, you get first pick. Oh, thanks.

I love number 5. If AI transforms higher education, how should we evolve the high school system? And that's from Dave Wilfart. Okay, I guess that was earmarked for me after all. So, yeah, I love this. I love this topic. And we obviously need to get on it. You know, I did a podcast with Joe Ahn, the the president of Northeastern. Incredibly great podcast. The guy is brilliant. You should check it out somewhere on YouTube. Joe Aoun is A-O-U-N if you want to search for it. But we were talking about the evolution of the college system that has to happen like right now. And he's opening a new incubator on Mass Ave here in Cambridge where the students who— you know, Northeastern has always had a lot of work studies. So you can work instead of taking classes and get real-world experience, but now build a company, be an entrepreneur, that counts. That counts as part of your college curriculum. It's phenomenal. But that same mentality needs to move into high school where you have to first recognize the curriculum can't possibly keep up with the useful knowledge that the kids are going to want to absorb. So you have to switch it over to AI-based teaching, AI-based learning, allow them to learn whatever they want to learn. Purpose-driven learning. Purpose-driven learning. And we invented a class right in this pod. That must exist, you know, prompt engineering and scaffolding. That has to be a new class. But, you know, next semester it'll be something else. Next semester it'll be something else. You just have to allow that to come into your ecosystem, then reward good behavior, like trying to learn or doing something that looks productive. Give that an A+. But don't try to force everybody down an ancient curriculum.

Dave Erickson, here's an idea for you, given that— I'll get in trouble for saying this, but don't really care— MIT does not really to take the humanities seriously. Like, as an undergrad at MIT, I, I got humanities credits for the philosophy of quantum mechanics and set theory, and all my humanities classes had problem sets or lab assignments. What about a humanities credit at MIT for prompt communication, uh, prompt engineering and context engineering as a sort of a stealthy way to introduce communications skills? That's a great idea.

Alex, I'd love to hear your answer to number 8. All right, I've been assigned number 8.

So number 8 asks, does training on synthetic data degrade model quality over time? And this is from QC for Life. Depends on the synthetic data. Uh, so you could generate synthetic data for, let's say, prototypically software engineering problems like generating code and then injecting a bug into the source code and testing whether a model is able to find that bug. There are many reinforcement learning challenges that I think, at least by historic standards, certainly benefit from synthetic data and especially synthetic environments. So creating procedural 3D environments or creating, say, just applications that can generate almost infinite variation that via reinforcement learning and reinforcement fine-tuning models can learn from That is, I think, quite valuable. That, by the way, I mean, I think what maybe the question's subtext is, is, isn't it sort of a garbage in, garbage out? How could you possibly learn from synthetic data? Isn't it just, you know, garbage in, garbage out? And the answer is, in fact, no, it's not garbage in, garbage out. There's, um, I've mentioned this on the pod in the past, uh, a couple of terms: Solomonoff induction induction and AIXI, you could in principle train purely— if you had a sufficiently strong model, you could train it off of no physical world and no human data at all, purely synthetic data if you had a sufficiently powerful model. And this is the premise for AIXI, which is a theoretical, not very practical implementation of Solomonoff induction. Solomonoff induction is an approach to basically building the perfect inference And the premise of it is if you've just observed a sequence of bits or a sequence of tokens, the perfect next token predictor, which is all large language models are, is basically organizes and runs every possible Turing machine, every possible computer program on that history, which is computationally infeasible but theoretically perfect. So going back to the question of synthetic data degrading model quality, Actually, in the limit of perfect compute and infinite compute, it's actually ideal not even to touch the physical world and not even to touch human data and to train purely off of synthetic data.

I'm so happy you answered that one.

You know what else? Not to beat this to death, but having trained many, many neural nets and training some right now, synthetic data is fine. It's mislabeled data that absolutely kills you. Absolutely killer. Just one mislabeled data point is a killer. The weights will warp themselves 8 ways till Tuesday trying to make it make sense in context of everything else. So clean synthetic data is totally fine. In fact, it's better in a sense because it doesn't have something just blatantly wrong and misleading. All right, Saleem, 6, 7.

Why don't you do this one?

7. Why don't you do the next one? You always go last.

No, it's okay. Go ahead.

I will. Gentlemen, the questions are abundant.

Number 7. What stops Frontier Labs just acquiring other companies in order— in other industries to get their data? This is from David Lee, Z5E, or Zed Five E if you would say properly. So nothing stops them from doing that because you're going to get acquisitions. You can see private equity buying chunks of like mid-market accounting firms and trying to get that stuff. The real goal is the proprietary data they have. The problem that I'm seeing as I'm watching PE try this is buying a company doesn't mean you get really useful knowledge because there's a lot of tacit knowledge, uh, that's hidden in, in key employees' heads that's not easy to extract. You've got different operating processes and so on. So traditional companies need frontier models, but the model provider needs proprietary information to make it useful. So I think what we're going to see is a lot more partnerships. For example, we're seeing groups of hospitals band together and then pool their data and then make that data available to pharma companies, it's almost like a co-op model that becomes really interesting. But based on the, my previous point and the whole organizational singularity stuff, the implication for companies is super urgent. Organize and protect your proprietary data so that when you add AI, you have the learning loops that become very, very powerful. That will be the engine of growth for your future, and it'll become much more important and much more valuable than your current product. So I don't— they may go buy some startup startups, but it's going to be a harder thing. I think it's easier for the frontier labs to go partner with companies for their data and mutually figure out ways and cooperate rather than trying to do this acquisition stuff because that tends not to work out over time.

Hmm. All right. Final question.

Number 6. Yeah. Yeah. Question 6 goes to the guy who got SpaceX.

That makes perfect sense. How can you, how can individual investors get involved in cutting-edge startups before they go public? Before they go public? That's from Nancy Jenner-C5D. So Nancy, there are so many ways for you to find cutting-edge startups before they go public. It's the easiest is getting them at the beginning. You go to your university first and foremost. A lot of these companies are beginning in the minds of, you know, 20, 21, 22-year-olds, right? You can go to equity crowdfunding platforms and see what's going on. There are syndicates on AngelList. Venture capital funds, you know, have minimums you'd have to buy in there. You can go to demo days. There is so much going on today that can enable you. Dave, what do you want to add to that?

I think if you add value, you will get stock. And there are so many ways. A lot of these companies are growing so quickly, and if you discover them early and you just try to add value in any way, they need so So many, you know, you know, the guy who painted the Facebook office, what did he make, $100 million on that stock? 'Cause they paid him in stock 'cause they didn't have any cash. He was just there. And so I think, you know, people underappreciate how much you can reach out to these guys, especially early on when they're desperate for help in any, hey, do you want introductions? Do you want sales help? Do you want help moving your office?

What are your skills? If you make yourself available.

What are your skills you can add?

Are you a great coach? Right? You wanna run and, you know, get coffee for the team? Anyway.

My favorite suggestion would be to go to AngelList syndicates because people have syndicates where Jason Calcanis will invest in a bunch of startups and you can buy into that syndicate. They get a piece of the carry, but you get participation in all that. And those have done extraordinarily well and you don't have to put a lot of money up. Yeah.

It was funny you say that because we had a summer intern. He just said bye to me today. Because he's got to go back to school in September. But he, he put together an angel syndicate over the course of the summer. And, you know, he's young, he's still a student, but he's going to manage it. And all the rich, old, famous guys are like, great, if you manage it, you can just participate and we'll, we'll lend our names. So he actually pulled together a syndicate in what, like 4 weeks?

Yeah. Amazing. Gentlemen, as always, a pleasure. And to our listeners, thank you for subscribing. Thank you for joining us. It's no time to sleep during the singularity.

I have to go talk to all the rabid—

Don't take off the takeoff. What's that?

I have to go talk to all the rabid Moonshots fans that were like, we can't wait to talk to you about questions. So I've got to go talk to them.

You look like you're in a consulting office over there. No, I'm in a hotel room. Aha. Okay.

Fantastic. All right. All right, Zalim. Well, words of encouragement.

See you guys very soon. Love you all. Be well. Thanks, Peter.