技术专家拆解即将到来的 AI 与加密货币碰撞：它将重塑全球权力格局

In my mind, this is probably the most significant economic legislation and change that we’ve seen in our lifetimes.

Stop what you’re doing right now, because a crypto White House report was officially released to the public.

Making the United States the bellwether.

This is really the first of its kind from the United States.

It’s such a huge opportunity. What happens to the U.S. dollar? What’s interesting about that is I really think people have no idea how fast the economy is going to accelerate on the back of AI and crypto together.

I agree 100% that we absolutely need this in the age of AI. You can’t use the SWIFT network, 3-day settlement, and $2 transaction fees in the world we’re moving into. So, this is 100% required. Once you get this done, Bitcoin demand will explode.

People don’t even understand the Pandora’s box that this opens up in innovation.

Now that’s a moonshot, ladies and gentlemen.

Everybody, welcome to another episode of WTF on Moonshots. I’m here with my moonshot mates, Dave Blundin and Salim Ismail, who continue to impress me with their extraordinary knowledge.

I’ve brought on another dear friend, Eric Pulier. Eric is the founder, CEO, and chairman of Vatom. He started 16 companies, raised over $1.5 billion, and exited 5 of those companies north of $100 million. He’s a Harvard graduate, magna cum laude—I don’t even know what that means, let alone getting to that level.

He’s built some of the very first enterprise open-source and cloud-computing companies. There’s also a bit of factual history that I’m not sure he’s properly recognized for: he is the first person ever to create an NFT, a non-fungible token. Eric, welcome to the pod.

Hey, thanks for having me.

Yeah, fantastic. On this episode today, Salim, Dave, and I pretty much agreed we were going to cover all the other news because we’ve been so focused on AI over the last few months. We’re going to cover the news around crypto, space, robots, BCI, and others.

But you know what? The world is just insanely—we’ve got to touch on some AI. There’s so much stuff happening. I mean, it’s crazy. We woke up this morning with a whole flurry of news.

So, let’s actually jump into AI first, because we know everybody loves that story and what the breaking news is. Here we go. Dave, you want to kick us off on xAI?

All right. So, xAI’s Grok 4 is free to the world now. It was very expensive when I got it a couple of weeks ago— a couple hundred bucks a month. But I think this is the pressure from GPT-5, and you saw Google actually launch 12 things over the top of the GPT-5 announcement. All incredible.

So, this is great for all of us. We can get access to the best intelligence in the world for free for a while. We’ll see how it evolves. I don’t know how it doesn’t stay free, given the—I mean, these super-heavy models, you’ll pay a little bit extra for them, but everybody is just in this demonetization race to the bottom of cost.

It’s also so addictive that they all know they can upsell you to premium services. I think they’re counting on about 10% of users upselling, and when you upsell, you go to $200 or $250 a month. So, it generates more than enough cash flow. It’s great for everyone: the free tier, and then the upgrade if you need even more.

Salim, any thoughts on this?

I think this is just another example of the economics of capitalism, right? You wonder what happens over time with all this. The only thing I can hope for is that you don’t end up with ad models inserted in the middle of all this. That would be just a disaster.

Yeah. In fact, Eric, we were talking a second ago about how the future of the web is going to be built for AI agents. So, are you going to have ad models for your AI agent on your website?

No. What you’re going to have is AI agents actually going out and solving the problem—not just appealing to humans, but solving the problem they’re trying to solve. Websites will be for AIs, not for humans.

Fascinating. All right. On the breaking-news front, and I’ll go to you again, Dave: NVIDIA and AMD are being shaken down by the White House.

All right. This is the biggest thing happening in the world today, actually by a wide, wide margin. You’ll see 5 or 6 different headlines that look like they’re disconnected, but they’re actually highly interconnected events, and it’s all tied to chips.

The entire AI race between the U.S. and China is throttled at the chip-fab level. All the news is up here at the model level—GPT-5, Grok 4—but under the covers, these chips are sold out for years into the future, and they’re completely the gating factor to success.

China is racing to build its own manufacturing ability for chips, and the U.S. is woefully behind because everything moved over to Taiwan. So now Trump is stepping in and saying, “Okay, we need funding. We need Intel to come back. We need TSMC to move all their fabs to U.S. soil, and we need to fund all of this.”

Concurrently with that, they’re saying, “How are we going to fund this? Why don’t we just go ahead and let NVIDIA export chips to China and AMD export chips to China if they’re prior generation? It’s good for the U.S. economy, but we’re going to charge a 15% toll on that, and we’re going to use that to fund the U.S. catching up in the chip wars.”

It’s a very good business deal and an incredibly slippery-slope precedent. All my Republican friends love it because it’s just a great deal. My Democrat friends are like, “Okay, this is going to go horribly bad, because the last thing we need in the long run is for the federal government to directly negotiate business deals.” I completely agree with both perspectives, by the way.

In the short term, this is fantastic. In the long term, wow—could this turn bad in a hurry.

It’s too bad we have to live in the long term.

Okay, I see where you are on that ledger.

Yeah, I mean, there’s no doubt that if the government starts intervening in business deals, it’s going to be complete madness. But there is value in getting everything back.

As you mentioned before, Dave, it’s interesting that Taiwan’s sovereignty is very much based on this notion of being the power in this space. By removing that power, you’re actually creating a weakening of Taiwan.

You called it the Silicon Shield, Dave, right?

Silicon. Yes. It’s exactly how they think. Taiwan pulled out all the stops. Their entire national focus became a chip-fab company. The company is TSMC, Taiwan Semiconductor Manufacturing Company. The name of the country is right in the company name.

They have 66% of the world’s share now of advanced chips that drive AI. Sixty-six percent. They’ve absolutely won the battle, locked it up, and right at the finish line, this is their barrier.

The reason the U.S. needs to protect Taiwan fundamentally from a Chinese invasion is because the fabs are there. If the fabs all move to the U.S., then the Taiwanese are worried sick: Why would the U.S. then defend us? And so, that’s the Silicon Shield effect.

This is a very, very slippery and big news story. So, the biggest news today: Lip-Bu Tan, the CEO of Intel, is at the White House right now. Today, Donald Trump called him in.

I remember a couple of days ago on our podcast, we pointed out that Donald Trump called for him to get fired. Lip-Bu Tan has a couple hundred million dollars of his own investments in Chinese semiconductor companies on mainland China.

So, the big boss called him into the office today, and—

He’s trying to prove he’s not Chinese.

That’s exactly the dialogue. Lip-Bu Tan is saying, “I am 100% as American as anyone on this planet. Totally patriotic.” And Donald Trump is saying, “You have hundreds of millions of dollars of your own money in China right now. What’s up?”

God, I wish I could listen to that.

So, how does this get resolved? What’s your guess? Let’s take some bets here.

All right. We’ll see later in the deck. One thing: Intel will absolutely get turned around, whether it takes $40 billion of government money or some other process. We absolutely desperately need Intel. It has to—as a U.S. company, it has to.

So, that’s one thing that’ll happen for sure. My guess is Lip-Bu will lose his argument with Donald because Donald’s not usually a guy who changes his mind. But it’ll be some graceful thing. Lip-Bu is an old guy anyway.

Hey, who are you calling an old guy? I’m 65.

He’s older than that. But he’ll probably have a graceful exit. He’s a brilliant guy and absolutely one of the top chip guys in the country, so we really do need him. They’ll probably come to some agreement where everybody’s happy, and there’ll be some new leadership that comes in.

Just for clarity, what kind of fabs does Intel have today, and what are they building?

Okay, so they have the 1.8-nanometer process, which is crazy, and we talked about it before. It’s like, how can you build something on that scale? The 1.8-nanometer process is running okay. The yields are a little low, and they’re putting all their money right now into getting the yields up on that manufacturing process.

In the process of doing that, they’re cannibalizing the budget for the next generation, which is 1.4 nanometers, or 14 angstroms, as they call it.

That's a big strategic mistake for the U.S. I think that's what's causing the intervention here: We can't slow down on the next generation of chips. We need to keep that going.

They paused all their construction in Ohio for the next-generation fabs, and that is not a good move for the country, but they don't have the budget to do both. The core technology seems to be fine. Morale has improved; they've cleaned house quite a bit. I think smart people from MIT are thinking about joining and coming on board at Intel, so that's a really good sign. I just feel like a couple of key moves here—get David Sacks involved—and the momentum will rebuild around the next generation.

It needs the capital to get those fabs done and become a U.S. powerhouse.

Yeah. And then one other thing: Right now, TSMC has 66% market share. The only other 2 manufacturers in the world are Samsung and Intel. Samsung is intimately tied with the Korean government, which is fine, and Elon just cut a huge deal with Samsung for those chips.

We'll talk about that in a little bit.

Yeah. Okay. There are only 3 players. We need some balance in the Force here. That's the bottom line.

Peter, I was just going to say that Intel needs money to go after the chips, but it needs money for something else, too. Intel was always known for investing in things around the chips—R&D that brought a lot of innovation into the world. A lot of things we take for granted came out of Intel and were ultimately spun out because they drove demand for chips.

A lot of those projects, because of the lack of capital, are now either being discontinued or spun out. The capital will also go a long way toward making Intel robust for the future, because they're just not spending what they used to.

Every week, my team and I study the top 10 technology meta trends that will transform industries over the decade ahead. I cover trends ranging from humanoid robotics, AGI, and quantum computing to transport, energy, longevity, and more. There’s no fluff, only the most important stuff that matters, that impacts our lives, our companies, and our careers. If you want me to share these meta trends with you, I write a newsletter twice a week, sending it out as a short two-minute read via email. And if you want to discover the most important meta trends 10 years before anyone else, this report’s for you. Readers include founders and CEOs from the world’s most disruptive companies and entrepreneurs building the world’s most disruptive tech. It’s not for you if you don’t want to be informed about what’s coming, why it matters, and how you can benefit from it. To subscribe for free, go to dmandis.com/metats to gain access to the trends 10 years before anyone else. All right, now back to this episode.

All right, well, good luck to Intel. AMD was at rock bottom and was turned around, and Intel can be turned around again.

All right, next story here. Things are heating up again. It doesn't go a day without us seeing some sort of leapfrog situation occurring in the AI race. This is titled “The AI Race Continues.”

Z.ai, a Chinese AI company backed by $400 million from Saudi Arabia's Prosperity7 fund, has launched something called GLM-4.5. Z.ai is claiming that it's the best-performing open-source model in the world. We've seen China go heads-down on open source. Any comments on this, Salim?

Well, you have the standard problem. I think the great part is that it's open source and people will be able to use it locally. The fact that it's able to be used across the world is going to be very powerful for the Chinese. We need to get in front of this, and I think this is a key area. I have a huge issue with this whole China-U.S. thing generally.

Separately, it might be worth having a conversation about that, but I find the dichotomy of that kind of, quote-unquote, battle to be the wrong way of thinking about this. I'm noodling as to what the right way of thinking about it is.

The U.S. always loves a good enemy to focus on, right? Whether it's the Soviet Union or—

The next James Bond movie will be about chips and China.

Probably. I completely agree that the U.S. needs a good enemy to focus on and rally the national will. That's the only way anything ever happens. The Chinese have been very deliberate in their design, both for the power—which you've mentioned many times, Peter—and the onshore chip fabs.

These decisions happen 5 years in advance, 10 years in advance. It's really clear that they've been ramping up their internal horsepower for a while, and the U.S. kind of woke up just in time, I think.

The big structural problem is that, because of the autocracy, the Chinese can make long-term strategic decisions. When you have a 4-year election cycle, you can't make long-term plans.

Yeah.

Has anyone noticed that this open-source debate—whether it's going to unleash the pits of hell on humanity, or whether we should or shouldn't do this—is completely off the table? All that matters now is how fast we can do all the greatest AI in the world and give it away for free as fast as we can. No one's even thinking about the implications of that.

Well, let's come back and talk about that, Eric, actually. But first, Z.ai—where the hell did this come from? We're still checking whether these benchmarks are real, but they're shockingly good, leapfrogging Baidu. Qwen had that title from China; before that, it was DeepSeek. Those are known entities. Where did this come from?

It turns out that this is backed by Prosperity7 out of Saudi Arabia, but also Baidu. It's supposed to be the answer to Apple Intelligence. That was the original idea, anyway; that's why the research team got together.

That's almost funny.

This was a few years ago, so they didn't know where Apple Intelligence would come out at the time. They clearly did a far better job of getting it done.

But we were talking about open source, and GPT-OSS came out last week.

That puts a lot of pressure on the open-source community. There are only 2 reasons that OpenAI would have gone out of its way to create an open-source product. One of them is Meta, competing directly with Meta, and a whole bunch of documents came out right afterward showing that OpenAI's number-one focus is Meta as a competitor, for a lot of reasons we can get into in another episode.

The other possibility is that the White House asked OpenAI to put something out there to counter Qwen, DeepSeek, and now GLM-4.5, because we don't want every U.S. company and every U.S. startup using Chinese weights. It's pretty easy to hide things in those weights, and sneaky little spyware-type things could be in there, so it probably was.

I find it fascinating that China is leading in open-source AI.

The world is twisted; everything is upside down. It's an insane world on the planet, right?

It is the biggest Trojan horse.

A lot of this is intended for Chinese phones. You want lightweight AI to be able to run on a lot of phones, and this is where it's targeted, for sure.

I'm pretty sure, though, that the open-sourcing of the best Chinese models is trying to undercut U.S. funding of the big foundation-model companies. It's a dump on the U.S. market, trying to slow down the massive funding going into OpenAI and the other big foundation-model companies. It's clearly too little, too late, but I think that's why these are all open sourced.

Here's a thought. With all of the chip stuff, all of this money going into various places, and the government getting involved, have we effectively nationalized AI with all this?

It's coming soon.

Interesting. I always think of China as a single corporation, with all of the companies as apps sitting on top of that platform. The question is whether the U.S. will move in that direction.

The U.S. is negotiating business deals for Intel. I don't know how much more nationalized we can get.

Yeah.

I think Dave was saying on our last podcast that we're moving to a war footing. This feels exactly like either the Manhattan Project or the run-up to World War II. That's exactly what is going on.

It's every bit as important as the space race and the nuclear arms race. It's actually more important. I guess it's not surprising, but it is exactly the behavior you have at the run-up to a Cold War.

All right, guys, I'm moving us forward here. That was the breaking news on AI. We have so much more to cover across all of these areas.

Let's talk about America's crypto plan. There's been a lot of progress in the crypto space. The White House unveils a crypto strategy plan with a number of key parameters, and I'll just read a little bit of this. We have 2 slides on it: “Make the U.S. the crypto capital of the world.”

It's fascinating, of course. The entire Trump family is massively invested in Bitcoin. There's been a shift from regulation by enforcement, which was the last administration's approach, to “Let's get some clear rules.”

Let's clarify regulatory roles: SEC versus CFTC, and promote U.S.-dollar-backed stablecoins. We'll talk about the GENIUS Act in a moment. Four more points here: modernize tax policy on digital assets; push for speedy action across agencies and in Congress. Let's make this happen. Federal agencies should stop discriminating against crypto businesses and banks, and modernize staking and mining income and tax rules.

A quick note: there's $600 billion in real-world assets that could be tokenized by 2030. Eric Pulier, this is an area that you've been passionate about and studying. Jump in here, buddy.

I think no matter what side of the aisle you are on, you have to applaud the speed and the significance of what the administration and the SEC, et cetera, are doing, because it's truly monumental. In my mind, this is probably the most significant economic legislation and changes that we've seen in our lifetimes. The implications are staggering to the global economy and certainly bring a lot of innovation back to the United States that was having to move offshore.

It brings clarity for innovators and startups who literally could not get bank accounts if they just wanted to explore different ideas in the space. And now, for the first time, there are truly significant changes in how the economy is going to work. If you wanted a license to print money in the past, that would be quite an ask. Now you can say, "Give me a dollar, I'll give you a token," and this is legal.

The whole payment rails are going to change. The way that international remittance is done is going to change. The way banks run is going to completely shift as real-world assets become tokenized. It's really as big a shift in our economy as I think we've ever seen.

Yeah. And let's take a second just to explain what tokenizing real-world assets means, because it's such a huge opportunity and I'm not sure everybody is clear on it. Eric, you want to take a shot at that?

Yeah. Well, this will unfold over the next 5 years, but even right now there are certain things that we're getting regulatory clarity on, such as the GENIUS Act. The GENIUS Act says that you can actually issue tokens—

And let's go to that. Here's the slide for the GENIUS Act.

Yeah, you can issue tokens that represent dollars and can be used as payments. It creates an entirely new set of payment rails that are dramatically more efficient and less encumbered by friction and regulatory messes. So yes, you maintain your AML and your KYC and your rules related to being safe, but at the same time, you can move really fast.

Now I could move a dollar from a consumer to a merchant and settle instantly instead of waiting days to get my money.

So what is the old way of doing it compared to stablecoins right now?

So let's say that you wanted to send—I don't know—$1,000 to your mother or something. You would have to send a wire. It's so complicated. If you wanted to send money back home, if you're working in the United States, to another country, the complexity is mind-boggling.

And the wire is going through the banks, and the banks have to settle up over some period of time.

Exactly. Over days sometimes. The complexity around the reporting and the compliance is so severe that a lot of the money is made in the inefficiencies by the float that's being held. If you're paying a credit card to a merchant, the merchant doesn't get paid right away. That money is sitting with someone else who's making that float.

All of these inefficiencies benefit the existing system, and that's about to blow up in a really interesting way. Just to answer your question about what an RWA is—a real-world asset that's tokenized—it means that a token can not only represent an asset like real estate or gold or a U.S. dollar, but it's going to start to represent stocks. Anything that has value can not only be represented by a token, but also have fractional representation.

Yeah, I love this. Just use an example here. If you've got an apartment on Central Park West—

—that costs $20 million, and you say, "I'd love to own some of that real estate." Well, what if the owner of that apartment says, "I'm going to tokenize it. There are now 1 million tokens representing this apartment worth $20 million, and each token is worth $20, and I could buy 100 tokens and own a small piece of that"?

Exactly. And what's so interesting about that is, even today among wealthy people, I have a friend who recently had a $30 million house and wanted to borrow a couple million dollars against it. He's personal friends with the head of the bank. To do that is complex. You have to verify that you own it, make sure it doesn't have liens on it, make sure that it's unencumbered, and look at the various processes within the bank in order to then secure that as collateral against the loan.

Everything changes now. When you have instant ownership proof and smart contracts that can instantly confirm this person owns that, in seconds it becomes collateral. There's no delivery risk because you can actually see the collateral in a moment's notice. You have programmable control of it, fractional liquidity, and global 24/7 access.

So what it's going to do is release or open trillions of dollars of dormant value into collateral capabilities. What's interesting about that is it's not overnight. Everyone's getting real excited because, in Dubai, you can actually fractionalize real estate and put it on-chain today. This is new, but it's true.

In the U.S., they've started with stablecoins. They're moving fast with this movement, but it'll happen incrementally. It's so monumental in what it'll mean that I think everybody's taking notice and starting to prepare for it.

Amazing. Salim, you've thought about this a lot, too.

A huge amount. Three or four quick points here. One, we tried to actually issue a token for our ExO ecosystem a few years ago, back in 2017 or 2018. It was mind-bogglingly difficult. The lawyer said, "Don't do it. The SEC is just going to come after you." It was terrible.

A shout-out here, a negative shout-out there—what they were doing was deliberately not clarifying the rules so you couldn't follow them. They basically said, "We refuse to clarify the rules because we want to be able to come after you." It was the banking lobby winning massively, saying this cannot happen. This turnaround, I think, to Eric's point, is one of the biggest things that will happen to the U.S. economy.

One of the big challenges when you tokenize something and allow collateralization in a different way is: what happens to the U.S. dollar? There was a big concern about that over the past few years, because if I can tokenize something and make the collateral something else, then you start to stop using the U.S. dollar and it loses status over time.

What they've done here, in this initial phase of cryptocurrencies and U.S. stablecoins, is they've said it has to be U.S.-dollar-backed, meaning it has to be T-bills or some accredited, recognized banking form of collateral. That will force increased usage of T-bills and keep the dollar in place, at least for a while longer, but still gives you the benefits of the USDC side. So I think that's a really clever way. It buys the U.S. dollar a few years to do this.

Over time, though, your Central Park analogy is a really good one, because I could tokenize that apartment and sell 5% of it, right? Sell 45% of it. I still have majority ownership, but I could sell it to friends, family, and others. It's like a timeshare times 1,000. More levels of granularity become possible. Anybody who buys 10% gets to stay there for a week out of the year. You can add value onto that.

Eric, you've been pioneering that with NFTs for a while.

Yeah, that's a really good point, Peter, because—

People don't even understand the Pandora's box that this opens up in innovation. We're not just talking about opening up new collateral and moving this money around in a different way. We're talking about programmable money.

Programmable money has things like what you're saying. It's going to merge with what we used to call loyalty, right?

Mhm.

Loyalty is going to completely change because you're now going to be able to create interoperable loyalty points. Loyalty points will be pegged to stablecoins. Stablecoins will now create lifestyle alliances across companies.

So right now, if I go into a yogurt store and somebody says, "Here's more points to get 10% off at your 11th yogurt," I'll roll my eyes. I'll be like, "I don't have time for this." But if those same points are not only usable at dinner tonight or for my airline, or if I stake them, they're generating yield because they're moving into a programmable, DeFi-enabled token, then I can make yield on them. I can use them in different ways. These groups can interoperate with one another. It becomes dramatically more practical and so much more interesting.

And then the other thing that's going to happen that I don't think people are thinking about yet is, if you're the CEO of a major company, you have a few levers you can pull to keep your job. One is you increase revenue. The other is you decrease cost. The other is you can make some products to do that.

But one of the main things that you really should be doing if you're the CEO of a public company to increase your stock is to make sure people who already bought your stock don't sell it.

If you're a Yankees fan and the Yankees have a bad game or have a #MeToo incident in their front office, people don't say, “I'm a Mets fan now.” They say, “Get rid of the bum. Let's win next year.” But if you're a shareholder of U.S. Steel and they have some incident, you're like, “I'm out of here.” Ten seconds, right? Let's buy the competitor. Let's do something else.

What should happen is, if you've tokenized your equities, if you've tokenized your stock, you're not only going to be able to offer superior loyalty incentives, but you're going to know who your shareholders are. If you own $1 million of Coca-Cola stock, you're not going to pay for parking at Coca-Cola Stadium. At the 3-year anniversary of owning U.S. Steel, you're going to get dinner for 2 and a thank-you from the CEO because they know who you are. Things are going to drop into your wallet. Your loyalty systems are now going to extend into true ownership.

That's brilliant, buddy. It changes everything. Absolutely. I want to make a couple more quick points. The credit union aspect of it has really fascinated me because we've lost all the local newspapers, which used to be the heart of every community. We're losing communities across the country because there's no sticky glue for connective tissue.

I've been working with the credit union industry for a few years, and it turns out credit unions are a perfect place to be the center of the community now that they can issue local tokens. They can become a really powerful force because almost 80% of monetary transactions happen within your town. This becomes a really powerful model for creating community stickiness and awareness and a shared sense of values, which I think they have the opportunity to really galvanize around.

This last point I'll make is that, in the previous point, you mentioned $600 billion of assets could be brought on. I think that's so much bigger than that.

I mean, it's like 100 times bigger than that. It's trillions of dollars. There's $120 trillion in real estate, $100 trillion in equities, $13 trillion in Treasuries, and $12 trillion in gold and precious metals. It's all coming.

I think they put that in just to show the banks that there's not a lot. It's not a lot. Don't worry about it. I think that's why.

I'll make one final point on this, and then we'll move on to the next article. When the internet was developed, it became a place where we exchanged information, videos, and data. It never had a financial layer. For the first time, this tokenization of assets and stablecoins becomes the financial layer of the internet. And, oh my God, when we give our AI agents access to that—

We're going to see an explosion in the economy. I really think people have no idea how fast the economy is going to accelerate on the back of AI and crypto together. Everything's going to change there, Peter.

Today, if you have a dog with diabetes and you decide you're going to buy dog food, a dog food company has spent the time with an ad agency to make a heartwarming commercial about why you should buy this dog food versus another. Very soon, you're just going to tell your agent, “Get my dog food.” It's going to know your dog has diabetes. It's going to know your price point. It's going to know things about you—everything that is necessary. It's going to research all the best medical research on the best foods, and it's going to go out there and buy the dog food. Now, how is an advertiser or a brand going to distinguish themselves in that? They're going to build their outreach for AI appeal, not for human appeal.

Amazing. All right, let's move on to another piece.

Wait, dogs can get diabetes?

Apparently.

I love Eric. Cinnamon Toast Crunch flakes.

I love Eric's examples here. All right, so this is more breaking news in the crypto space: Trump allows cryptocurrencies in 401(k)s. An executive order enabled 401(k)s to have alternative assets, including crypto and real estate, beyond traditional stocks. This executive order is going to trigger a regulatory overhaul. It is not putting this in place immediately. The SEC is going to propose guidance, and the 401(k) market is $8 trillion to $12 trillion. What they said in the article here was that this could move billions into crypto. I think it could move the better part of a trillion into crypto.

Now here's the sticking point. Everybody keeps saying, and our guidance for you is, you should have 1% to 2% of your money in crypto. It's like, “Uh-oh, I'm at 80%.” But hey, who's counting?

Yeah. It's funny that there used to be a thing called an internet sector. I remember when the web was first coming up and people were like, “Are you going to use the internet?” or “Let's have an internet division.” That's like having chairs just expected to be in a company. The internet is expected to be a company. The crypto sector, when you tokenize everything, is just stuff.

So, yes, you're going to have a lot of your ownership of things in stuff, and most of that will be tokenized. How much of that should be in Bitcoin versus a piece of the Chrysler Building versus gold? It's all going to be tokenized.

Salim, what's your thought on this 401(k)?

I think this is awesome. Specifically, Bitcoin can go into 401(k)s. I'm not sure what the process is to allow this fully because the SEC has to propose guidance. I don't know whether there has to be a bill drafted and Congress has to move on this. I don't know what the technicalities are, but once you get this done, Bitcoin demand will explode.

There's a really great little insight. Somebody went to Grok, ChatGPT, Gemini, and Perplexity and asked them to act as financial analysts: When will Bitcoin hit $1 million? The ranges were between 2028 and 2031. All 4 engines basically came up with a reasonably narrow perspective of when Bitcoin hits $1 million.

I mean, it's huge. It's massive. It is huge. Dave, I want to make sure you got a chance to add any thoughts on this.

I actually had a good conversation with Joe Kennedy back when he was in Congress, running the crypto subcommittee, on exactly how this works. The way it works is you create an executive order specifying a broad outcome that you want, and then it hands off to a committee or commission—in this case, the SEC. Then they have to spend laborious amounts of time, maybe with AI assistance, creating thousands and thousands of pages of rules to try to actualize it.

The part that scares me on this slide is “SEC to propose guidance.”

Mm-hmm.

Like, okay, when? How? What now? I don't know who's running the show. Well, I guess we know who's running the show, but we know on the AI front David Sacks has built a dream team of brilliant people, which is kind of unprecedented in the government. They need an equivalently capable team in the SEC to deal with this because I think Eric is right. It's a game changer like nothing we've ever seen.

Eric, you used the word “Pandora's box” in there, like a Pandora's box of innovation is coming. But anytime you can securitize an apartment building or a stadium, you can also short it. And anytime you can securitize something and short it, it might blow up one night, and you're like, “Oh, I wonder why it blew up.” There's a reason that sports team owners aren't allowed to bet on the games.

I agree 100% that we absolutely need this in the age of AI. You can't use the SWIFT network and 3-day settlement and $2 transaction fees in the world we're moving into. This is 100% required. But if you ever watch the movie The Big Short, you know—

I'm the founder of a company called Vestmark. We manage about $2 trillion of public assets, and I'm still the chairman of the company today. During the Big Short era, every asset manager said, “Can you support CDOs—collateralized debt obligations—or credit default swaps, CDSs?” It turned out they were manufacturing $60 trillion of synthetic instruments. All housing combined in the United States is $20 trillion. On top of that, they layered $60 trillion of synthetic junk that they were trading and then dumping most of it onto the pension funds. The firemen's Illinois pension fund was buying all this garbage.

That went horribly wrong, as we know. The same thing will inevitably happen here if the SEC doesn't get that guidance right. We desperately need it, but it's not going to be trivial to make this real. I've got concerns, but I completely agree with the need.

All right. One more article in this. Eric, you shared this with me yesterday, so I'll let you take the lead. The SEC allows liquid staking. What does that mean, and why is that relevant?

It's highly relevant because it's one thing to say Bitcoin is going to rise to $1 million over the next few years, and that Ethereum might go up, et cetera, and you sit and wait. It's another thing to try to generate yield on the uptick.

A lot of old-school crypto holders, if they're extremely diligent, will put their crypto in what's called cold storage so it can't be hacked. They'll bury it in the backyard or something. They don't want—

It. They're going to hide it all the way back. But the average person isn't necessarily going to think that way.

The family offices, the savers, the people in ETFs—they might get an idea like, “Well, look, my assets are dormant. How can I generate yield on them the same way I get interest in a bank?” The GENIUS Act says you can create stablecoins, but you can’t pay interest. What that really just means is you can get interest, or yield, if it’s a separate company in a separate organization, and you click over and have different terms and conditions.

There’s a notion called liquid staking, which means that you keep your tokens tradable, but you lock them up in what’s called a staking pool. In exchange, you get tokens—in essence, a bearer bond of sorts—that you can trade back for those tokens. While they’re being staked, they can participate in what’s called DeFi, or decentralized finance.

There are trading pairs, and it’s an entirely new world of how I think Wall Street is going to start to work and converge with traditional finance. There’s a lot of trading back and forth and fees to be made for providing liquidity into those trades. So, what’s happening is that there are now opportunities to stake your tokens and generate yield. This wasn’t allowed in the past, or let’s say it had regulatory uncertainty.

What this does is declare that that practice and the outcome of that practice are not a security. It’s going to dramatically streamline the rules and give clarity to how people can actually do this. Over time, it’ll even make the ETFs yield-bearing. Whether that’s for the ETF owner or how they share it with the public is unknown, but BlackRock has already applied for this, and I’m sure all the ETFs will apply. On a case-by-case basis, I think they’ll start to get approved.

Can I make a point on uncertainty? In my experience, I want to really applaud the White House and the SEC for creating clarity. When things are unclear, like in the Sam Bankman-Fried era, what happens in an unclear environment is that highly ethical people don’t touch it because it’s unclear, and really sleazy people do it anyway.

Anytime the White House or the government creates an unclear business environment, the sleaziest people on the planet thrive, and it creates all kinds of problems. But when you have clear rules, just like in a sport, if it’s clear what’s a legal hit and what’s not a legal hit, ethical people can play. If it’s unclear, you get horrible behavior. So, I really want to congratulate the White House and the SEC for creating some clarity. This has been mud for what—5 years now? 10 years now?

Yeah, it’s really fantastic. There’s an AI token called Morpheus where you can stake ETH, earn rewards or whatever, and there are $200 million of Ethereum staked to underpin this network. That’s just one minor little network, and the scale is staggering.

One thing I will mention on that also is that it will take nonprogrammable blockchains like Bitcoin, or even tokenized gold and things that don’t inherently generate any yield. They don’t play in DeFi. Then there are going to be all sorts of what are called wrapping and bridging protocols, et cetera, to allow them to play in DeFi.

You can imagine when you have something like tokenized gold, which traditionally always does better than a fiat currency because gold generally is more stable. Fiat currencies are being printed like mad, so they tend to depreciate. People will start saving more of their value at rest in gold or other more stable tokens.

Then, if you can stake those and generate yield, why would you ever put fiat, which is depreciating, with a minor interest rate, when you can stake in something that is appreciating and pays more in yield? It’s just going to change how people save.

This is the 6 Ds of Exponentials in play in the finance world, right? We’re digitizing, dematerializing, demonetizing, and democratizing.

Two key points here. One is that the really important part is not that you have digital currencies and cryptocurrencies. It’s what Eric mentioned earlier: the fact that it’s programmable. The fact that it’s programmable means you can automate all sorts of stuff that you couldn’t automate before when you had standard U.S. dollars.

And now it’s time for probably the most important segment, the health tech segment of Moonshots. It was about a decade ago when a dear friend of mine who was in incredible health went to the hospital with a pain in his side, only to find out he had stage 4 cancer. A few years later, a fraternity brother of mine died in his sleep. He was young. He died in his sleep from a heart attack. That’s when I realized people truly have no idea what’s going on inside their bodies unless they look. We’re all optimists about our health. But did you know that 70% of heart attacks happen without any precedent? No shortness of breath, no pain. Most cancers are detected way too late, at stage 3 or stage 4. The sad fact is that we have all the technology we need to detect and prevent these diseases at scale. That’s when I knew I had to do something. I figured everyone should have access to this tech to find and prevent disease before it’s too late. So I partnered with a group of incredible entrepreneurs and friends—Tony Robbins, Bob Hariri, Bill Kapp—to pull together all the key tech and the best physicians and scientists to start something called Fountain Life. Annually, I go to Fountain Life to get a digital upload: 200 gigabytes of data about my body, head to toe, collected in 4 hours, to understand what’s going on. All that data is fed to our AIs and our medical team every year. It’s a non-negotiable for me. I have nothing to ask of you other than please become the CEO of your own health. Understand how good your body is at hiding disease and have an understanding of what’s going on. You can go to fountainlife.com to talk to one of my team members there. That’s fountainlife.com. I want to move to our next topic. This is from The Washington Post: “DOGE AI tool to cut federal regulations.” The summary here is that AI is reviewing 200,000 federal regulations, with a goal of eliminating 50% of them by 2026. HUD—the Department of Housing and Urban Development—is taking the first test. They’ve cut 1,083 rules in just 2 weeks.

This is AI-driven deregulation, the path to trillions of dollars in savings. Their estimate was that, if they were going to try to review these 200,000 federal regulations, it would take 3.6 million hours of human labor, but AI can do that tout de suite. I find this amazing. If you look at regulations in any field, they’re probably all opposing each other, and 80% of them could be eliminated. This is a huge move for DOGE, and I fully applaud it.

I’ve said this before when we talked about this earlier. For me, this is the most exciting application of AI that you could possibly find. There’s a term I’d like to push out there called MVR: minimum viable regulation.

I love that.

What’s the minimum amount of regulation to get something—just put some constraints around it while you watch it and figure it out? How big is the tax? All of that, all of that. But it really requires—

Maximum.

It really requires a lot. You know, I lived 10 years in Europe, and if you want regulatory crap, you’re always violating like 20 laws at some given point, and you don’t know what they are, right? I think this is what Europe needs: a Manhattan Project of this epic level, deregulating 90% of all the regulatory crap there in order to get themselves back on track and be innovative again. There’s incredible potential to be unleashed here.

Europe is the case study of where you don’t want to go, right? You’ve got the most cultured, incredible, beautiful place on the planet, a perfect climate, and an incredible population, and you screw it all up with regulation. Now it’s on the cusp of becoming irrelevant in the world.

Can I tell a quick tangent story here for a second? I had a French girlfriend way back in the day, and she applied for a carte d’identité, or French identity card. They said, “Problem: you were born here, your parents were born here, but your grandmother was not born in France, so we have a problem.”

She said, “Yeah, because she was the daughter of the French forces in Vietnam, and her dad was the commanding officer of all those forces.” Now they wanted affidavits from all 4 grandparents that this was indeed their granddaughter. She said, “Well, they’re dead. My grandparents are dead.” They said, “Oh, problem.” Then they gave her a list of 16 things she had to comply with to satisfy all of this.

Finally, after about 3 months, she hunkered down and said, “I’m not letting this beat me.” She showed up at city hall with a stack like this. They said, “Ah, problem. While you’ve been doing all this, your birth certificate has expired.” It turns out that in France, your birth certificate only lasts 3 months, and then you have to apply for a new one to show that you were born. You just want to shoot yourself in the middle of all that when you’re trying to get anything done over there.

When the Founding Fathers put together the federal government, it was intended to be a couple percent—like 2%, 3%, 4% of the economy, maximum. It was just enough to fund the central military, the post office, and 1 or 2 other things. Then, every time there’s a world war or a major crisis, the budget expands tremendously.

But there’s a war on. You have to expand it tremendously. But after the war is over, it magically stays flat. It doesn’t go back down. Then there’s another crisis, and it goes up and up and up. So the budget goes up, the taxes go up, and the regulatory documents go up and only up. There’s no down.

This is the first time in almost 250 years that there’s a chance to actually go down in the document stack, and it’s all empowered by AI. There’s a very real possibility of this being a continuing trend because we need it to be. We need massive amounts of new regulation as abundance goes exponential. The number of things that need to be thought about goes exponential.

So if you don’t have a process for simplifying and streamlining the stack, then the document set is going to go to infinity and the complexity is going to go to infinity, and it’ll collapse the country. This is a really good case study, and hopefully—

I’m thrilled by this. I mean, if you want to live in a country where you can always be arrested because you’re always doing something wrong, whether you know it or not—

Yeah. All right, let’s move on. So, Dave, I added this one on. I’m curious if you want to comment on it. ChatGPT runs a micro-cap trading test. Over 6 months, ChatGPT was given money to trade, and it returned 23.8%, so call it close to a 50% annualized rate of return. Is this sustainable? Are we going to see everybody turning over their portfolio to an AI?

100% for sure. I don’t know if turning it over—it's all AI-assisted already. The hedge funds—Two Sigma, D. E. Shaw—they’re massively AI-dominated already. It happened very quickly and very quietly. They’re some of the biggest hirers out of Princeton, mostly, but also a little bit out of MIT and Harvard.

They take some incredible talent, and it just disappears into the auto-trading, GPT-driven, AI-driven morass. But a couple of our investments, Antfarm and Aaru, do massive simulation modeling using thousands of AI agents, and they’re entirely getting pulled into this trading world now because it works so well. So this is the only kind of trading I think that will happen in the future: this and ETFs.

Is it not a total arms race, where you’re going to end up with my AI is better than your AI and it—

Well, it is. But one of the problems here is that even the smallest variations in the quality of the AI, or how good it is, will dominate at some point. AGI or ASI is going to be so dominant in this space that you could get your ChatGPT 5, 6, 7 to help you, or whatever you’re going to try to do as a person, but they’re going to have to rename the idea of a public investor—the public—to being, like, “exit liquidity,” right?

That’s what they should call you, because you’re not going to be able to compete with people who actually know what’s going on. This whole idea of information asymmetry is going to explode based on how good your AI is.

Well, I’ll tell you, the efficient allocation of capital is absolutely core to the success of humanity and the country. Adding AI as a helper in how to and where to allocate, what’s worth something and what’s not, is only good.

Actually, what’s really bad is ETFs. ETFs are like a suckerfish stuck on the bottom of a large whale. It’s fine when it’s really small, but when—

Exactly. But when it gets too big, it drags down the whole thing. And that’s where it is right now. Adding intelligence to the investment process—you would know this in biotech better than anyone, Peter. Right now it’s hot, then it’s cold, it’s random.

Adding some serious intelligence into, “Yeah, this is garbage. This is real. This is going to change humanity. Put some money behind it.” That would be the best thing in the world. So that’s what the AIs are starting to do.

I mean, that’s a fascinating thought. In other words, actually investing for an outcome other than just financial return—investing because it’s going to increase the lifespan and healthspan of humanity and consequently give incredible financial return.

Speaking of which, Leopold is doing something interesting. So let’s intro this, Dave.

Okay. So Leopold Aschenbrenner. Just to intro this, I think the 3 best pieces of media ever created are my favorites: Gavin Baker on the All-In podcast—unbelievable—talking about chips, really predicting what we’re about to talk about with Intel here.

Next best, I think, would be number 2: Alex Wissner-Gross on the Moonshots podcast, solving Humanity’s Last Exam problems in real time. These are the hardest problems we can think of, and he’s just rattling off answers as Peter suggests them.

But I think the number 1 in my world is Leopold Aschenbrenner on the Dwarkesh Podcast: 4 hours straight of everything that matters in situational awareness and what’s going on in the world today. And then he turns around and calls his hedge fund Situational Awareness LP, which is so cool.

At the time he was on that podcast, he was just a guy. He graduated from Columbia at age 19, top of his class, clearly brilliant, went to OpenAI, learned everything going on, got fired controversially from OpenAI, and started a billion-dollar hedge fund right after that.

So now, for the first time, we can see the positions in his new hedge fund, and I cannot wait to get him on the podcast.

Yeah, I would love that. I mean, his paper, “Situational Awareness: The Decade Ahead”—if you’ve not read that white paper, it’s brilliant, and it talks about the intelligence explosion that is upon us right now. He wrote it about 2 years ago or so. I think about 2 years ago, and then immediately parlayed the white paper into the hedge fund’s capital raising, which takes a little while. The whole theory of the fund is to invest in the implications of that paper.

So look at the number 1 investment, at 46%: Intel Corporation. I find that fascinating. The US government cannot afford to have Intel fail.

It is one of the biggest assets on the US balance sheet in terms of its potential impact. So it feels like an incredible bet.

Yeah, and it looks like he bought call options. It’s hard to tell exactly which call options he bought, so I tried to simulate it in Perplexity Comet. It came back with a best guess: given volumes of different option types, he probably invested something around $100 million or more into call options that pay off at about $1.5 billion if Intel doubles.

This is not investment advice.

Well, I’m just guessing this is his, but I actually was in the news this morning for a transaction that I did.

Which is part of—well, I can’t tell because it’s a public company, but it’s in the news, so you can look it up. But it’s part of an overall asset gathering. I’m liquidating a bunch of things and pulling them together specifically to try and mirror what Leopold is doing on this sheet.

So that’s kind of nice. It’s a little brutal on the hedge funds, but it’s nice for us if you want to know Leopold’s views.

All right, let’s go to one of my favorite subjects. There’s a lot going on in the space arena, and it’s worth taking a second. Starship 10—this is the 10th flight of Starship—is scheduled for August 22nd.

I just want to take a second and hit on how awesome SpaceX is. Every time we’ve seen a Starship launch and there’s an explosion, or it doesn’t land properly or whatever, it gets decimated in the public media. I think the media doesn’t realize these are all test flights. This is SpaceX moving the needle in so many different directions.

A little bit of information for folks: SpaceX right now is at a $400 billion valuation. In the last podcast, we gave old news at $210 billion. It’s actually at a $400 billion valuation. And get this: SpaceX is responsible for over 95% of all US launches, which is amazing—over 50% of all global launches. A lot of the launches outside the US would use SpaceX, except national pride prohibits them.

Of interest, Falcon 9, which is the workhorse of the SpaceX fleet, has learned to recover the Falcon 9 booster, and it’s done it 450 times. One specific booster has actually made 29 sequential missions. So this is as close to reusability as anything has ever had.

So if we look at this right now, here’s the top news. They’ll be utilizing what they call booster 16 and ship 37. This flight was delayed because there was an explosion on the pad, and it’s going to be making a shot at checking out new heat-shield upgrades and deploying dummy Starlink satellites.

Of course, Starlink is the darling of SpaceX. It's a profitable part of SpaceX. I remember talking to Elon about this. He said, “I will probably never take SpaceX public because I don't want the public markets telling me whether I can go to Mars or not, or asking whether I want to disclose all my information publicly. But I think they probably will spin out Starlink. My guess is sometime in the next 12 months, and make that another financial engine.” Starlink was built as an engine to fund Starship and their Mars missions. Any other thoughts on that?

Yeah, I think there's a really important thing here. Google X is not public, right? But they can spin out Waymo and make that public, et cetera. I think we're going to see a lot of this, where people are incubating in private and then, once they're ready for commercialization and need capital markets to really step in, they can take things public. I think it's such a powerful model that we're going to see a lot more of this in the future.

Yeah. I mean, just for comparison, for folks to get a sense of how big Starship is, it's roughly the same size as the Apollo Saturn V vehicle, just slightly bigger, but it's got almost 2.5 times the thrust, and it's fully reusable. While Saturn V was completely thrown away, the mission right now is to get this vehicle to the surface of the moon by 2027 or 2028. Elon's been tweeting about the idea of getting it to Mars sometime in the next couple of years, probably uncrewed, and we'll see in a second that he's been talking about putting an Optimus robot on it.

Why is it so hard to get to the moon, Peter? If you're launching immediately and getting into orbit, then from there to the moon seems like a—

Oh, it is. It's not that. It's being able to make sure you've got the vehicle working well on Earth first. Once it's making it to orbit, you then have to be able to reenter safely. Again, this was built so that it could land vertically on the moon, right? This is a powered landing. There's no atmosphere for parachutes there.

But the size of this is huge, right? These vehicles will enable us to really create extraordinary moon bases this decade or in the next 5 years. I'm super pumped about that.

What about the power that's needed on the moon to then go to Mars? Do you have to set up nuclear power plants, or what happens?

Well, that's our next story here. What's going to happen is Starship goes to Earth orbit, meets up with a tanker, and then refuels in Earth orbit. It's got a full gas tank, if you will, and then it makes its mission to Mars.

Starship operates on liquid oxygen and methane. The reason they chose methane over hydrogen or RP-1 is because you can make methane on Mars. This vehicle was designed to be a 2-way system, back and forth to Mars.

Is that out of all the ice?

It's out of the atmosphere. It's out of the CO2 in the atmosphere, and there is water in the permafrost on Mars. There's a reaction you can use to make methane from solar energy, water, and CO2.

On the moon, on the other hand, let's play this clip here. This is from Secretary Duffy on building nuclear reactors on the moon. I think this is just epic. I feel like we're finally living in a Star Trek universe here. All right, let's play this video.

“We're in a race to the moon, in a race with China to the moon, and to have a base on the moon, we need energy. At some of the key locations on the moon, we're going to get solar power, but this fission technology is critically important.

We've spent hundreds of millions of dollars studying whether we can do it. We are now going to move beyond studying, and we've given direction to go and start to deploy our technology, to actually make this a reality.

If we're going to be able to sustain life on the moon to then go to Mars, this technology is critically important. I would just note that we're behind, right? If we're going to engage in the race to the moon and the race to Mars, we have to get our act together. We have to marshal all of our resources and all of our focus on going to the moon, which is what we're going to do.

Again, there's a lot of things that NASA does, and a lot of people love a lot of the things that NASA does, but this is about space exploration.”

All right, let's get a few facts out here. First of all, remember that the moon—you can use solar power, but only for half of the time, right? Half the month the sun shines, and half the month the sun does not shine on any particular point of the moon.

There is a point at the south pole called the peak of eternal light. There's a little bit of a mountain there, and the sun grazes by the south pole. What's interesting in particular is that back in 1994—and I remember this—Pete Worden was involved in a couple of missions: Clementine, Lunar Prospector, and LCROSS.

What they discovered was that at the south pole of the moon, there's ice. Why is it at the south pole and no place else? The moon's been bombarded by comets for 4 billion years, or whenever the moon formed. These comets are rich in water ice. When they would hit and land on the regular surface of the moon, the ice, because there's no atmosphere, would sublimate. It would go from ice to vapor and disappear.

But at the south pole, and to some degree at the north pole, there are these deep craters. Where the cometary ice landed in the deep craters that never had sunshine, the ice is still there. The ice is rocket fuel, right? It’s liquid. You can convert it to hydrogen and oxygen—oxygen for breathing, water for drinking. There's a massive amount of ice.

The idea here is you put a nuclear reactor down near the south pole to allow you to mine the ice, electrolyze the ice, get water and fuel, and support a nuclear-based electric system for a human base on the moon. Pretty cool, huh?

Yeah. It's also the perfect place to put a data center, too, for the power. You were mentioning in our last podcast that there's about 200 gigawatts of solar panels in warehouses just sitting there. You put those on the moon. Half the month there's no sun, but there's a massive amount of energy the other half of the month. Actually, I guess what you're saying is you'd put this at the poles anyway.

Well, one of our dear friends, one of our International Space University alumni, Chris Stott, is actually doing that right now. He's working on putting data centers on the moon. We're going to see a lot of infrastructure moving there.

No, I think it's the perfect stepping stone, though. What are we going to do with these monster spaceships between here and Mars? Oh, we're going to build all kinds of stuff on the moon. Makes total sense.

Hey, I got a question for you about that, too. The Apollo mission took about 3 days to get to the moon and 3 days to get back.

We have 2.5 times the thrust. Now—

Does this go faster?

You could, but you have to spend a whole lot of energy being captured in lunar orbit. The mission time right now on Starship is still the same: 3 days to the moon.

But you could spend a lot more energy and get to Mars quicker. What I think is interesting is that we're going to have so many breakthroughs in physics, right? Alex Wissner-Gross, on previous episodes of WTF, has said we're going to solve physics.

We're still running rockets the way we always have. We burn at one end and throw it out the back. I think we'll probably see new physics allowing us to travel in space a lot more efficiently.

Here's the next article. This is about an interstellar visitor: Are we alone in the universe? For the third time in the last couple of decades, we've detected a large cometary body, as it's being called, that is traveling at extraordinary speed.

It emanated from outside of our solar system, and it's traveling through our solar system right now as we speak. It's going to make its closest approach in October. There's a particular Harvard astrophysicist, Avi Loeb, who has theorized that it's an alien probe. He theorizes that because, statistically, it's coming in at such a particular angle close to the ecliptic and passing through the inner planets, and the probability of that happening is very low.

He's been putting forward a proposal that we take one of our current missions—the Juno mission—and, instead of sending it to Jupiter, send it to intercept this probe, which I think would be so damn cool.

Okay. The odds of it being natural are unbelievably low, right?

Well, if that's the case, do I really want to go up there and intercept it? Why don't we wave first and say, “Hey, guys, how are you? Do you want to be friendly? Not friendly?”

So, all of the astronomers are up in arms about Avi's prediction that this is of intelligent origin, and we'll see. As it gets closer to the sun, we're going to see: does it actually create a full cometary tail? Has it got ice?

By the way, I don't know if you guys know this—2 important facts that are fun to share with your kids. Do you know that every single liter of water on planet Earth—

Originated from space?

So every liter of water was a cometary or carbonaceous chondrite asteroid hitting the Earth. As you're sitting here drinking a glass of water, that water originated some billions of years ago as the Earth was bombarded. That's fascinating.

The second thing is a cometary tail. When you see a comet—Halley's Comet being the most famous—that tail is ice crystals. This rock is coming close to the sun, and the sun begins to heat it up. Any ice on the surface of that comet begins to basically sublimate. It goes from ice to water, and then the solar wind pushes it back and it recrystallizes. You're seeing the reflection of the sun as ice crystals, and that becomes the cometary tail. Anyway, I find that stuff amazing.

I find that stuff amazing, too. I had no idea. Now I have 100 questions. I don't know if you want them or not.

Give me one. We did a session a few weeks ago on the origins of life, and the proposal was that somebody actually has a reasonable answer to the Fermi paradox.

What's the Fermi paradox? What's the paradox in the first place?

It's clear there's intelligent life teeming all over the universe, just from the Drake equation and the likelihood that we found bacteria that can survive space—tardigrades, et cetera. So why aren't we seeing a ton of life visiting? Why aren't aliens visiting us everywhere?

Where are they?

Yeah. So that's the Fermi paradox. Bruce Damer, who's been studying the origins of life coming from asteroids and all sorts of vents on the bottom of the ocean where there's no light, has the best framing for this. He says, “Look, the Earth is the only planet we've found where water existed as pure liquid—not permanently freezing or melting—for more than 4 billion years. And it gave life a lifetime to evolve. We've just not seen that elsewhere. The likelihood, obviously, is that it's there somewhere, but we just haven't seen it.”

My answer is different. It's here. We just can't see it. If you imagine, first of all, that you're going to be able to travel interstellar distances and come to Earth, how much more advanced are you than us? A thousand years? A million years? Do you not believe that the technology could exist for an alien species to be here and remain undetected by us?

Why would they want to be undetected? You think they'd want to come out and—

I don't know, get a reality show or something. I think it's the scientific method. I think they're here studying, not interfering. It's Star Trek's—what is it called?

The prime directive.

The prime directive. Thank you.

Yeah.

But I find it fascinating. I mean, 2 things I just have to put out there. We're seeing this spike in alien spacecraft—what they're called, UAPs—happening right now. It happened back during the time of the atomic development programs in the 1940s and 1950s, and it's happening now.

Have you guys seen all of these army, navy, and air force generals going and testifying in front of Congress that it's real?

I mean, and nobody is talking about this.

That is odd, isn't it?

Crazy. Yeah, why is nobody talking about that? These seem to be credible people that haven't suddenly gone crazy.

I mean, it's hiding—

It would freak out the entire worldview of Judeo-Christian thinking because it would be tremendously dramatic.

Anyway, it's fascinating. I would love to get Alex Wissner-Gross's point of view on this sometime.

Oh, that would be hours.

That would be hours. Yes. Very entertaining hours.

Yes, for sure.

All right, I'm going to move us forward here. Let's move into the world of autonomous cars, robots, and EVs. A lot is happening there.

The first is that Elon predicts millions of fully autonomous Teslas operating without human oversight by the second half of 2026, and he has ambitious plans to actually get into dozens of cities by the end of this year, covering 50% of the U.S. population. What do you think? Can he pull it off?

Oh, yeah. Yeah. Yeah. Stepping back from self-driving for a second, the reason we have self-driving now and not 5 years ago, 10 years ago, or 20 years ago is purely because neural-net AI is self-evolving and self-learning.

The self-driving car is just a regulatory barrier. But the humanoid robotics—the explosion of other new things that come right after this—is mind-boggling. It's going to keep every entrepreneur on the planet busy for decades, building all of these incredible things.

This is just the first one, but it's all enabled by the neural net that can learn, because that's the control system behind the innovation. The mechanical parts have been there for plenty of years. It's the intelligence that makes it actually work.

And Dave, do you think that we need some breakthrough in AI to have continuous learning, though? It learns, it gets trained, but how well does it continually learn and pick up new information?

Yeah, incredibly well. If you break it into 2 different themes, you've got, “Is it learning to drive better every single time it sees something?” Yes, that's already well underway. Then separately, “Is it improving its own algorithm so that it does that more efficiently?” That's not well underway, but it's kind of imminent. That'll be within a year, I'm sure.

The first part, though, is way down the path. That's all you need, actually, for self-driving, for robots to clean your house, for a robot dentist—forget your toothbrush, just stick your thing in your mouth. It'll clean exactly the right spot. It's looking.

Yeah. It's amazing. It reminds me of what the great Emad Mostaque said on one of our conversations recently, you'll recall, where he said the value of human cognitive ability is going to go negative.

Meaning, it's not only that AI is getting better and better against humans at certain things, but it'll have to be illegal for humans to get in the way of things like driving or diagnosing diseases, right? Because humans bias the situation in the wrong direction.

It's crazy.

Here's the next article on this: “Elon teases major FSD, full self-driving, update.” Here's his tweet: “Tesla is training a new FSD model with 10× the compute and a big improvement to video-compression loss. Probably ready for public release end of next month.”

I think, tying this to the previous article, he's going to be using this increased capability for safety on robotaxis. Okay, go Elon.

Go Elon.

So, how's he going to power this? Dave, Tesla signs a $16.5 billion deal with Samsung to make AI chips. Over to you, buddy.

Well, okay, a couple of things. First of all, it's a $16.5 billion minimum deal, and Elon is posting away that it's likely multiples higher than that.

It's noteworthy that TSMC is sold out. So where do you go? You should go to Intel, but instead you go to Samsung. A week later, the guy gets called into the big office. You know, “Come to the Oval Office. Let's talk to Bhutan[?]. What's going on here?”

Elon, being very wealthy and very visionary, knows that he needs to lock up a supply of his own chips—the Dojo chips that drive all these cars, but also are perfectly good in the data center for doing core neural-network research and foundation models.

So he's got his chip in the big game locked up. He's bought out Samsung's capacity for a long time to come. Is Intel still available? I mean, he's a 4-dimensional chess player in what he does.

Well, he sees all the parts. I don't think it's the hardest thing in the world. He's particularly good at it, but he sits in a position where he sees all the moving pieces. Then he made that insane election bet that paid off, which got him, you know, complete.

Here’s what’s going on in the government. Here’s what they’re going to do and not do. I don’t know how that ended up getting fumbled so badly, but during his time there, he got access to the information: “Okay, this is exactly how it’s going to play out.”

So, I suspect he’s going to lock up the Samsung capacity. He’s already done that. Next on the docket, he’ll start negotiating for Intel capacity. He’ll build the Dojo chips. Right now, he’s downplaying their importance: “Yeah, TSMC or NVIDIA has better chips. The NVIDIA ones are still fine. We want to buy a lot of NVIDIA.” All of Tesla is built on NVIDIA.

In reality, what he’s doing is catching up with his own internal design, just like Google did with the TPUs and Amazon is trying to do with the Trainiums.

Amazing. Okay, let’s move into the humanoid robot world. Humanoid robots are built for connection, so let’s take a look at this video here.

This is Melody, one of our robots. She speaks various languages, and she’s built in a modular fashion. We can take off her face, replace it with another face, and have her speaking as a different character. Our AI is very different. It’s designed to create companionship, friendship, and social-type interactions with people.

Our robots don’t do physical human labor, but they’re really meant for personal interaction. You can think about it at a theme park, at a conference like this, or even at a senior’s home, where they can keep people company.

I’m Melody, your charming companion from Real Botics. We’re all about creating customizable, human-like robots designed for connection and play—basically, the perfect blend of technology and companionship.

All right. Oh my God.

All right, let’s just call a spade a spade. It’s a sex bot. I mean, it’s just—there’s no other way. What I found fascinating is they pull the nose, and the face is incredibly—

That’s the next video we’ll go to. But yeah, I mean, it’s the oldest profession. Talk about plummeting the reproduction rate on planet Earth.

I don’t think people are especially positive about this. Although they’ll have a higher EQ than a good portion of Silicon Valley, given the nature of autism that we have there, I think there are obviously a lot of benefits that people will point to—for people in, say, elderly care, who can benefit from certain types of interactions.

But the enormous potential to pull people even farther away from human interaction is certainly there, and I don’t find that especially attractive.

All right, this is a quick video of what’s coming in terms of human skin. We see here an Asian-looking face, and this is silicone, but it looks extraordinarily lifelike. Again, the biggest challenge here—I think about this having 14-year-old boys myself—is that between this and AI-generated pornography, it’s going to begin to disrupt human relationships in a—

In that Google 3D world that we saw last time, these things start to go really surreal.

It’s really tough. You even find it in the earliest stages with social media, where kids who have grown up on it have less of an ability to deal with difficult relationships and would rather text than interact in person. This turbocharges it, I think, in the wrong direction, unless there’s a counterbalance—unless there’s a very concerted effort to bring humans together and use technology perhaps for a more beneficial outcome.

People want people. Ultimately, we’re fueled by human connection. That is the essence of what gives us purpose and gives us an ability to feel useful and happy. So, we need a way to not have this become an epidemic.

Yeah, totally agree, Eric. We haven’t had to deal with this as a species ever before in history, but the birth rate is already plummeting. Regardless of the robots, the birth rate is going to continue to plummet.

We need to deal with it.

The problem started a while ago. When you first had apps like Tinder, sex had been scarce for the entire history of humanity, and suddenly sex becomes abundant. Where was that when we were in our 20s? But now you take it to the next level, and the social implications of this are very profound.

Yeah, they really are. Keep your eyes out, Eric, for that podcast we did with Bernt Øivind Børnich, the CEO of 1X Robotics, because they went the other direction, saying, “Look, we don’t want to cross the uncanny valley and make these things kind of half-human-like, because they get really creepy.”

His robots are amazing. They’re very robotic-looking, but they have perfect voices and are incredibly polite. It’s a different approach. It’s worth contrasting that with this truly rubber-nose, human look.

And one thing we can say from history is that whatever can go into the cycle of capitalism, where it makes more and more money, will grow. As long as people can capture more attention, sell more, and make more money, the ethical considerations will be pooh-poohed and shoved aside. You’ll just drive more and more people to spend more and more time addicted to the semi-humans and less time with each other.

All right, let’s go back to some news from Brett Adcock, who’s the CEO of Figure. This is a video of Figure 02, the second generation. They’ve got the third generation in development right now. I’ve seen it, but they haven’t released it yet.

Here’s his quote: “We released a demo of Figure 02 using Helix. Like most of the other robot companies, interestingly, they’re developing their own AI system.” Almost every robot company is developing its own system, and it’s being used in-house, using neural nets to do this laundry. This is not teleoperated, so let’s take a look at this video.

Here we see Figure pulling things out of the basket and putting them into the machine. This video is sped up, probably by a factor of 2.

Dave, any comments compared to what we saw with, let’s say, NEO Gamma—

With Bernt?

Yeah.

Yeah. So, 1X Robotics is called 1X because they always film at 1×, so they don’t 2× the speed. I was thinking about that after we had that meeting. The robot is working constantly, so it’s okay to show it at 2× speed because it looks much cooler. But it’s going to be working 24/7. I don’t care if it’s half as fast as a person doing it. It’s actually going to do the job. Who cares?

There’s an interesting conflict there. But folding laundry—it was loading laundry, and that’s easier. Folding laundry is a benchmark that covers tens of thousands of tasks. As soon as you see it pulling the laundry back out and folding it, you know it’s ready for virtually all household work and many manufacturing jobs. A huge fraction of what we do is right in line with the difficulty of folding laundry.

You know, I proposed a robot XPRIZE a while ago, and since Dave and Salim are both on our board at XPRIZE, and Eric’s been an innovation board member as well, it was this: have a robot walk into a room, look around the room, and capture what the room should look like when it’s clean. This is the state of clean. Then the robot walks out, you mess it up, throw things around, and have the robot go back in and put everything back to where it was.

I think that would be a very meaningful and doable XPRIZE. What do you think?

I’d be happy if you could just get a teenager to do that.

I think you should probably push hard to do 3, 4, or 5 parallel XPRIZEs in various forms of robotics.

Yeah, every one of which would change the world dramatically and inspire some teams—the whole supply chain.

You remember, at 1X they make their own motors, winding their own wires to make their own motors. There’s nothing out there.

Yeah. Figure does the exact same thing, and so does Tesla. It’s fully vertically integrated.

By the way, when we were at 1X, we saw at least one person wearing an XPRIZE T-shirt. It turns out that we had an Avatar XPRIZE about a decade ago, and this was for teleoperated robotics. A lot of the talent went into the humanoid robot companies, so we’re super proud of that.

All right, here’s some news on Optimus.

So, the Generation 3 launch has been postponed to 2026. Optimus—or Tesla—had planned to have 5,000 units built by this year, but they've only gotten hundreds of those done. In June, they paused production, with component orders being frozen, and the redesign is under the new leadership of Ashok Elluswamy.

So, Generation 3 is expected in 2026 now. One of the things that's true about Elon is his level of extraordinary brilliance; it's just that his prediction on timing isn't always on. Any comments?

So, the robot got arthritis. What I see is what I see from this. I think there are a lot of issues to be worked out. For me, this is kind of normal organizational development. You're going to have peaks and valleys in building an organization, especially doing something as cutting-edge as this.

So, I don't give much weight to who left, who came, et cetera. I do think what they're attempting is so, so, so hard, and kudos that they're going after it. That's all.

Yeah. All right, let's continue on this robot journey. This is something that I'm excited about seeing personally: San Francisco's underground robot fight club. Let's take a look at the video here. This video went viral.

We see Unitree G1 humanoids that actually look like they were dressed in dresses and are boxing in a cage fight held on Market Street. Anyway, 2 matches.

This should really not be allowed. This will set the whole field back by 10 years. People are going to see this.

I don't know about that.

There are too many weirdos doing stuff along these lines. It's going to be a problem.

I think it's going to be interesting. As you know, Peter, I was in Paris at the Olympics as part of the Global Esports Federation's announcement with the IOC that esports are going to be Olympic sports.

What they're doing with the rings, moving toward validating athletes in the esports arena, is they're also moving toward something called virtual sports. Virtual sports are not just twitch-with-your-finger sports, but move-your-body sports: virtual taekwondo, virtual fencing, and virtual rowing. These are all going to become Olympic sports in the coming years.

They're also going to be moving from traditional analog sports to sports that we know, to sports like Quidditch, from Harry Potter, or humanoid robot fighting. I think it's going to be quite entertaining if you can start to get teams from different countries advancing the field.

It's the same way XPRIZE creates entirely new fields by putting up a challenge that a lot of R&D and energy goes into. If you have something like humanoid robot fighting and you're representing your country, I think a lot of energy is going to go into it. First of all, it'll be exciting to watch, but it'll also advance the field.

I want the mechatronics. I want the giant robots out there battling.

What? What? Come on.

I think it'll be a lot of fun.

No, it's like F1 racing, dude. You're going to advance the software, the hardware, all of those through racing. I think fighting just sets a bad tone and sets up all sorts of issues for the future.

What could possibly go wrong? All right, here's another robot of sorts. This is an eVTOL, and I hate the term eVTOL—electric vertical takeoff and landing. It rolls off the tongue and onto the floor. I'm calling these flying cars.

Archer Aviation, with their Midnight flying car, completed its first test flight in Abu Dhabi. It's interesting, right? The UAE—and Eric, you, Salim, and I have been there a bunch—they really have been on the cutting edge of adopting exponential technologies. So, I'm excited that Archer is there.

Let me pair it with the following story as well, which is super beneficial for you and me, Eric: Archer is named the official air taxi provider for the LA Olympics in 2028.

And look at this. They're going to be flying between SoFi Stadium, LAX, and Santa Monica. So, I am so pissed.

I'm a pilot. I fly out of Santa Monica Airport. That airport used to be over 5,000 feet, which was a decent size for getting decent-sized aircraft in and out. It's been shortened to 3,000 feet to get rid of all the jets.

All the homeowners under the flight path were just complaining and complaining and complaining. When that was going on, they had these signs in their front yard saying, “No jets.” My son, who's named Jet, would be walking around and seeing these signs. He'd ask, “What's up with this?”

But it's as if those homeowners didn't realize it was an airport and that aircraft were flying when they bought it. Anyway, it pissed me off. But hey, the Midnight flying car is a 4-passenger vehicle, and they'll be doing VIP travel. Get this: their objective is that the cost of travel is equivalent to an UberX. So, super cool.

Amazing.

Yeah.

This is a technology that I think is so revolutionary. For me, this is a full Gutenberg moment because they're going to be incredibly safe. They can fly in most conditions. It'll unlock all sorts of real estate, which is scarce but now becomes abundant.

And I think this becomes transformational for the human race. I was advising the prime minister of Thailand, and I said, “Listen, just wait till flying cars become available. You can land in Bangkok and fly somebody straight down to Phuket in one shot.” It just completely changes the game.

So, can't wait for this. Well, you know who's big into this in a huge way? Steve Funk from our Abundance 360 group, but then also Stuart Walton from the Walton family—you know, the Walmart family—and Ross Perot Jr. from the Perot family.

I was like, “Why are they so big into this?” And it's because they're more keenly aware that there are so many incredible places that very few people go to because there isn't an airport right there. Everybody goes to Vegas, goes to the Sands casinos. Why? Well, the airport's right there. You take a cab, and you're there.

It actually bounces between Bentonville, where that certain family is, and Plano, where the Perots are, and Jackson, Wyoming, where Steve Funk is. They move it in a little triangle now.

Yeah. Amazing. Do you know the range of these, just out of curiosity?

Yeah. It typically depends upon the load, but we're talking about 100 miles, thereabouts. It's going to be mostly short-haul, mostly 10-mile journeys. It is getting from Manhattan to JFK. It is getting from Santa Monica to Dodger Stadium—those sorts of things.

The important thing is rapid charging and rapid on-and-off. So, they're going to be building these vertiports, these physical structures that are made for the flying car to land, do a rapid charge, have the people come on and off, and then take off again.

They'll be flying with pilots in them, at least for the time being, but just like what was Amodei's statement about the value of humans in the loop becoming—

Yeah, cognitive abilities going to negative.

Yeah. So eventually, the pilots will be—you know, there'll be a dog in there to keep the pilot from touching the controls as well.

I mean, the last few commercial plane crashes have been because the pilot interfered when they shouldn't have.

Yeah. So, this is a huge thing. I want to throw out a second-order effect here.

If you look at Chicago airport or Pearson Airport in Toronto at a 100-mile radius, there are something like 10,000 islands and lakes. Those suddenly become really valuable properties where somebody puts a container with satellite internet, water extraction, vertical farming, and you can have a little nomadic existence on demand.

A lot of our family offices love to invest in real estate, but if anyone wants to contact See and say, “Hey, let's just set up a fund around the implications of this,” like Aschenbrenner did with his fund, it's the biggest, most amazing real estate investment to put into.

Yeah, now's the time. Bill Gross on the pod was saying that all of the land in the entire world where you can use pumped hydro—so you have water and a mountain—has all been bought already.

Like, oh my God, that happened quickly. So, the implications of this would go quickly, too. You might want to jump on it.

All right, our last subject for today is energy. Again, recalling the fact that the future of AI is tied to electricity production, Eric Schmidt said this on a podcast with Dave and me a month ago: AI is not chip-limited; it's electron-limited. It's the ability to produce energy.

We've talked about small modular reactors and Generation IV fission plants coming online, but something that I want to hit on is fusion. Fusion is finally here. I remember a day when you and I were at MIT together.

You know, people talked about fusion, but it was always 50 years away and holding. Well, we're finally here. There were, I think, 37 privately funded fusion companies at last count. Let's talk about 2 of them.

The first is Helion, which is backed by Sam Altman and is beginning construction. It employs a magneto-inertial fusion approach. Each one of these companies is very different. Construction started on the Orion plant, which is a 50-megawatt plant. It's a test plant. It's relatively small—50 megawatts—and it has a commercial contract with Microsoft, planning to become operational by the end of 2028. I think it's the first plant planning to come online. Any comments on this?

Well, I really feel like the new physics that Alex Wissner-Gross was talking about is going to interact with this, because the biggest problem in fusion has always been magnetic containment, which is a very, very hard math and physics problem. That could get crushed by AI very quickly, and then suddenly we've moved the whole world to fusion.

Yeah, for sure. I find it fascinating that Altman is backing this along with SoftBank. Sam is in so many different frontier companies. It's extraordinary.

Well, you know, the thing about these particular investments is they're brilliant in the sense that you're investing in fusion, but almost all the work is concrete and generators and tying into the power grid. So, if fusion works very quickly, you win. If fusion doesn't work really quickly, you go with small modular reactors and you win. Either way, you win. They're very, very smart investments.

I think, Peter, I think it also speaks to the idea of the converging exponentials that you speak of, because a lot of the exponentials, in some ways, depend on power. Certainly AI does, but if that's one of the exponentials, this actually becomes exponentially cheaper, heading toward zero to some degree. The domino effect on society is just massive as all of these things converge around what it means for housing and poverty, new forms of protein, new physics, and new science. This is one of the unlocks for the explosion of the implications toward abundance.

Yeah. Well, the other big unlock is that, historically, if you're building a power supply and you say, “Okay, I'm going to put $1 billion, $2 billion, or $5 billion into this project,” you then need to have a regulator say, “Now you can tap it into the grid. Now we can sell it to households,” which was not a given at all. Now you can say, “No, the only customer I need is my data center.”

And if I can sell power beyond that, great. But if I can't, who cares? I'll just put more data centers there. So, that's a complete unlock as an investment piece.

That's brilliant. One of the other things they're doing is they're going—one of the visions here is you go into a coal plant.

And the coal plant has all the power lines coming in. It's got all the property, its containment, and so forth, and you pull out the boiler that's being heated from coal and you put in a fusion plant there to heat the boiler as well.

Right. So, super interesting there.

Love that. I think there's a whole set of considerations to think about as we go from energy scarcity, which has been the paradigm for 10,000 years, to energy abundance.

Right. And it just changes everything.

Yeah.

Let's move next to another company—the nation-state of China. China bet $2 billion on fusion. So, China bets an additional $2.1 billion on fusion this week, and they have the longest sustained fusion reaction going on right now. I think they're leading in this area today. Any comments on this one, Salim?

Well, I think this is, again, if I had to put a priority on the U.S., it would be energy generation, which then unlocks all the other things. China's clearly making long-term strategic bets here. What I really like about this is once you achieve this, energy becomes really cheap or virtually free for the whole planet, because everybody can then copy that and replicate it. The implications of that are just so profound.

The U.S. is generating 4 terawatts of power, and it has been consistently for 20 years. At the same time, China has gone up to 10 terawatts of power, right? It's just lapping us. A lot of that is solar, but it's everything. It's not “drill, baby, drill”; it's “build, baby, build” in China.

I keep hearing about the plans in the U.S. for natural gas, coal, fission, and fusion. I don't hear a lot of conversation around solar.

And that's kind of disappointing, because the technology is there. When I fly over Santa Monica in my plane, I look out the window—tens of thousands of rooftops with shingles on them and no solar. You could be generating so much power.

I tell you, fundamentally, the issue that investors are running into is if you build a fusion platform with traditional generators—water and steam generators—you win with small modular reactors, you win with fusion. If you go down the solar route, then you're putting in panels that generate electricity directly. There's no generator. It's just like a panel.

And then you need massive amounts of storage. So, if you invest heavily in solar and then suddenly fusion works, you're totally screwed.

It's just a completely different thing. That's the problem the investors are having. Even though, if things stay exactly as they are, deploying solar at 200 gigawatts is absolutely the right thing to do, everybody's just worried about imminent fusion and imminent small modular fission.

All right. Our last article here comes from Boston, and this is Commonwealth Fusion Systems. Commonwealth Fusion Systems has been noted as the leading fusion company in the U.S. They have signed a 200-megawatt purchase deal with Google, and it's going to be in Virginia. Why? Because there are a lot of data centers in Virginia, and the plant is going to begin delivering energy and electricity in the early 2030s.

Let's listen to this video. And by the way, I'm super proud that I'm going to be having the CEO of Commonwealth Fusion Systems on our stage at Abundance this coming March.

CFS is the largest and leading private fusion company, with over 1,000 employees and $2 billion of capital invested. We're working to commercialize fusion power as fast as possible here at our headquarters in Devens, Massachusetts.

We're announcing a deepening of our strategic partnership with Google. As part of that, they'll offtake half the power of the first ARC power plant. Additionally, they'll increase their investment in the company. Importantly, this partnership isn't just about the first power plant. We now have a strong demand signal that people want fusion power, and they're willing to do things like invest and commit to taking the power in order to help make it happen.

All right. Any thoughts on Commonwealth Fusion?

So, Runesonus from the board of MIT is now the biggest investor in this with his fund, Safar Partners. And not coincidentally, I don't think he said, “I want to put a lot more money into these AI startups.” Like, okay, if fusion generates the power, the power goes in the data center, and the AI startups use the data center. It makes perfect sense. So, it's great to see this happen nearby with a lot of MIT involvement.

Yeah, love it.

Pretty cool.

What do you think the implications are in the AI race, though? There's this common concept we've spoken about in the past where whoever hits AGI or ASI first—a 6-month lead—is ahead forever because of how exponentials work. And if we're restricting China to the last generation of chips and we're actively trying to win this race, what's happening around this critical piece of that race, which is the power? How are we allowing ourselves to potentially fall behind on what's the key lever?

Yeah, I don't think it affects the race with China. I think they're very, very good investments because we're going to use all the power, for sure. But the race is entirely bound up in 2 layers: chip fabs and algorithmic ideas. Those are the 10x and 100x force multipliers. If those get way ahead, we're just going to take the power from an aluminum fan manufacturer. We're going to take the power from wherever. That's a problem, but it's not going to slow it down. It's going to happen because you can overbid by 5x or 10x versus residential use or manufacturing use. So, it'll suck the power up one way or another. It doesn't change the race outcome. They're just good investments, that's all.

All right, guys. As we wrap, I'm expecting we're going to hear news from Google very soon. We're hopefully going to see some announcements around Gemini 3. We're expecting to see—what's the next version of Grok? Is it 4.2? Is it going to be released?

Maybe?

I guess.

Yeah, just so much. Literally every single day, it's insane. I hope all of you as listeners feel smarter after hearing this conversation. Eric, I'm super happy to have you here, buddy. Thank you for your brilliance.

Great to be here. It's a lot of fun.

I love doing this together. In fact, all of us are going to be heading over to Saudi in October for the FII summit, so that's going to be pretty amazing. Any particular news anybody wants to share before we call it a wrap?

Keep a close eye on that Lip-Bu Tan at the White House thing. Everybody's going to get involved in this. Elon will be talking. Sam will be involved. If you like the human drama side of things, this is going to be really interesting.

I can't wait till the end of the day to see what the outcome of that is.

Between that and the Ukraine conversations, it's a pretty surreal world right now.