一切的未来：Circle、CrowdStrike等公司CEO眼中的2026年

I think most of us agree that, going into 2026, stablecoins and AI—and crypto in general—are having a huge resurgence. We were lucky enough to get two of my friends, Brian Armstrong from Coinbase and Jeremy Allaire, who’s the CEO and co-founder of Circle. We’ve known each other for 30 years.

Yeah, it’s amazing.

What a long, strange trip it’s been.

I know. It’s great to be with you again.

Yeah. And this is not your first time at Davos. You’ve been here a couple of times.

Yeah, I started coming with my last company, Brightcove, back in 2008 or 2009. It was a very different time in the world. The great financial crisis was breaking out everywhere, and that was an interesting backdrop. I always reference that the same week was when the first block of the Bitcoin blockchain was mined, with the “Chancellor on the brink of second bailout for banks” kind of embedded in the blockchain.

Yeah, it’s fascinating. I think it’s “Game of Thrones”: chaos is a ladder.

Chaos is a ladder.

Yeah, and for guys like us who’ve been through this, I guess, 3 times now, we went through it in the dot-com boom. I think we’re too young to have—

You remember, obviously, Black Friday in ’87, but we were in college, I think.

And so we had our Great Financial Crisis, and then we had COVID.

COVID.

Yeah, which was also pretty spicy. That was a great disruptor. When you have those moments happen, I’m guessing for you and many of the founders I talked to, you just think, “This is the time to build.”

Yeah, totally. I mean, constraints have a huge impact on what an entrepreneur does. Even in the history of Circle, we’ve had extraordinary ups and downs, some of which are endogenous shocks and some of which are exogenous shocks.

Back to 2009—that was that year, actually—I got my company, Brightcove, profitable, and then not long after it went public. We deal with what we’re dealt, with constraints.

I think the old expression is, “Constraints make for great art.”

Let’s talk about your journey with Circle. Stablecoins are obviously top of mind because we had the GENIUS Act. My bestie, David Sacks, who’s our crypto and AI czar for America, is here with me at Davos—or I should say, I’m here with him. Why is this technology important?

Look, when I got started working on this almost 13 years ago, Bitcoin had emerged. From my perspective as an internet technologist, I was thinking, “Wow, this seems like a new infrastructure layer for the internet—a missing infrastructure layer of the internet.”

The internet had ways to represent data, media, audio, video, and software, but there was no notion of money on the internet, and there was no protocol for money on the internet. It was very clear at the time that that was going to happen.

When we started, I wasn’t convinced that everyone in the world was just going to use a new commodity money like Bitcoin. My view was that we needed a bridge. We needed to connect the existing fiat system to crypto and to these new networks and build what we called an HTTP for dollars on the internet. That was the idea, and eventually that was called stablecoins.

Early on, we talked about building fiat digital currency, fiat tokens, and all this, but stablecoins stuck. The basic idea is that we now have a general-purpose, general-architecture form of money on the internet: digital dollars that can move just like everything else and can transact peer-to-peer, which is super powerful and provides a huge amount of value to people.

Really importantly, because of the technology of blockchains, we actually have programmable money. We’re right at the front edge of a renaissance in how money is used in the world. We’ll come back to AI, I’m sure, because it ties into that.

Fundamentally, we need a native way to have money on the internet. We need a very safe form of dollars on the internet, and that’s what stablecoins provide.

And volatility, and not knowing how much your dollar is worth when you open your wallet or your bank account, was a blocker for consumers.

Yeah, I mean, volatile cryptocurrencies—no one’s going to buy a cup of coffee with Bitcoin, et cetera, right? So that’s where the word stablecoin came from. It’s like, well, it’s a coin, but it’s stable.

How you actually achieve that is a whole different thing, and that’s where the model that we built—which is fully reserved with ultra-safe assets, and we have regulators and auditors look after it and make sure it’s all done in the right way—comes in.

And I remember talking to you offline about this. It was the easier decision for most people in crypto to go to Zug here in Switzerland, just be based in Zurich, or maybe go to Dubai and YOLO it. Don’t worry about regulations, audits, or regulators.

You made a different decision. You told me, “No, I’m going to do this buttoned-up and proper.” Why didn’t you go for the quick, “I’ll just do this offshore and give up my United States citizenship”?

Why not?

Yes, we have many crypto people right now watching the show. From San Quentin and—

Yeah. I mean, look, the show’s popular there. Going back to the founding of the company, when I looked at it, I thought, okay, if we want to establish a new way to put what we think of as regular money on the internet, and we want to actually build a new internet-native financial system that the whole world uses—businesses are going to use it, people are going to use it, and we’re going to do loans in it and all these things with it—if you want it to work that way, you have to integrate with the existing system and work with policymakers to figure that out. There’s just no other way.

So, I testified to the Senate in November of 2013, and if you read the testimony, it’s saying all the same stuff now as I did then.

It was received slightly differently.

It really was. I got a lot of hate very early on.

Take yourself back to that moment in time, where you’re in the right, you’re trying to explain how to do this correctly, and you get a reaction that is different from what you maybe anticipated.

Well, I think with a lot of technologies on the internet, I was around—and I think we were both around—in the early stages of the internet and the early stages of the web. There’s a very hard libertarian view, and then there’s obviously the statist view at the other end of that. I lean libertarian.

But I think also, in the early days of the internet, it was like, well, if we want to allow companies to get online, we need regulation for ISPs. We need secure transactions and SSL. Then we’re going to need people who are making decisions about who’s a valid site and who’s not a valid site.

These were really controversial things at the time that the more anarcho side of things was against. Crypto itself—cryptography and crypto—kind of birthed out of that.

These are people who wanted to be outside the system.

Wanted to be outside the system. I think Circle has always tried to find this middle way: open public networks, permissionless innovation, building on public blockchains, using open-source infrastructure. What you can do with something like USDC, as a dollar and as a technology, is far more open than the legacy payment systems and the legacy money systems.

There’s a very deep commitment to those fundamental internet ideas. At the same time, if you want BlackRock to use it, or you want the biggest tech companies in the world to use it, or you just want someone to hold it in a digital wallet and say, “Yeah, this is actually a dollar and I can rely on it,” you end up needing to have this kind of structure around it.

For trust.

Trust remains a key thing. People talk about “in code we trust” as the thing you kind of put around it—

Easy for a developer to feel that way. Easy for somebody who’s extremely technical to feel that way, right?

But for a civilian—

Right.

Non-anarchist, right?

Yeah, maybe you want to understand how the dollar is actually backed. That’s exactly right. I think we just took that path, and that was a harder path.

It took more capital. I had to hire—my first executive was a general counsel and chief compliance officer.

One of the stories, actually, is that before Jim Breyer and General Catalyst would give me capital when we were starting the company, they were like, “We really need to know that if you do this, it’s not illegal. We’re not going to have some kind of liability issue.”

So I personally, on my own, with my own money, hired the top regulatory advisory firm in the world to look closely at this—to talk to people at the U.S. Treasury Department and others—to really figure out: Can we do this? And actually, we could. There was a path to do it, and that’s because the U.S. Treasury Department had given guidance about how to deal with virtual currency in the banking system. This was in March 2013, so super early.

We needed to know: Is there a legitimate legal pathway to accomplishing what we want to accomplish? That’s always been important.

Well, you had to bring those in-house, I suppose, because if you went to a law firm, they would be like, “We don’t know. Don’t take this risk.”

And as we’ve seen, there are things that are legal, and then there are incumbents. Incumbents will use the legal system to try to stop innovators. You’ve also faced a little bit of that.

I would say, in the entire history of building this, there have been huge uphill battles with regulators and incumbents.

What’s interesting is that I found over the years that if you come to a policymaker or regulator and say, “Hey, there’s a new technology. It can improve things in this way. We’re trying to figure out how to deal with the risks. There are real risks—let’s not pretend there aren’t risks—and let’s come up with ways to address that,” people are pretty interested in talking.

But generally, when you go to incumbents and say, “Hey, we want to work with you or integrate this because we kind of need to work with you to make the whole thing happen,” there’s a lot more skepticism and constraint there. So that’s changed now. We’re in a very different world now.

Now it’s gone from a threat to an opportunity, but there is still a little reticence toward stablecoins and crypto by, I think, some of the major banks and big players, because—my perception is—they’re a little concerned that you’re too good at what you do and might be too far ahead.

If they had started 5 years ago, they might feel differently about it. Is my intuition close to your assessment, or am I wrong?

I think there’s some truth in that, for sure. But it’s interesting: You and I have this kind of juxtaposition against these other eras of the internet.

When digital media happened and it was, “Hey, you could stream media, or you could put up content on the web, and you could do all this,” there were all these digital media startups, and the media companies were like, “Well, we can do that too. Our business model might have to change,” et cetera.

And digital advertising: “Okay, well, you can transform advertising into this. It’s more targeted,” all this stuff. But then there obviously were companies that executed with fundamentally different technical execution—software execution. They built completely different types of utilities with completely different unit economics that kind of turned the product, user experience, and economics upside down.

Craigslist for classifieds, Google’s ad network—

Google’s ad network—

The New York Times, Amazon Marketplace, and many, many more like that.

So I think we’re in a similar place right now, where a lot of media companies, communications companies as well, enterprise software companies—there are big buckets here, retailers—have sort of said, “Okay, this is a new paradigm. It’s better for customers. I can deliver a better product. The economics are better. We just have to do this, and we’re going to”—innovator’s dilemma, blah blah blah.

And they make that transition, and it takes 5 to 10 years.

It takes them 3 times longer than probably the innovator. Walmart and Target actually do an exceptional job. Now, the United Airlines app is not terrible, right? I mean, it’s not Uber, but—

And this is the difference between a software-driven company and a non-software-driven company. We’re a software company. Fundamentally, of course we’re going to go faster. Of course we’re going to understand UX.

So the banks—my understanding is some of them are threatened by the concept of a stablecoin having the ability to generate points, incentives, or interest, essentially. That’s kind of the sticking point.

The GENIUS Act says, “Hey, you can have rewards, but you can’t have interest,” and that was, I guess, where this was able to get through the legislative process. Explain that to the audience.

Yeah. So, a couple things. If you go back a few years, when stablecoins sort of emerged, all the biggest bank regulators around the world got together in what’s called the Financial Stability Board and said, “We have to have regulations around this. This can’t just go crazy.”

That was also when Libra was coming out and all this. It didn’t come out, but it tried to come out.

Meta tried to do its own project, and they gave up on that project, based on my insider information, because the government felt they had too much power already with their platform. Zuckerberg said, “I have enough heat on this company already. I don’t want to have another group of people thinking I have too much power,” and he shut it down.

That’s what I was told by insiders. Is that your understanding as well?

I have a variety of insights about it, but I think we’re kind of a neutral company, right? We’re not giant big tech, et cetera.

But coming back to your question, regulators got together and said, “Hey, there is this payment-system innovation. It is this thing called stablecoins.” And they all kind of said, “Here are some recommendations for how you could regulate this.” Actually, that happened all around the world.

It happened first in Japan—there are stablecoin laws—and then it happened in Europe years ago, actually: stablecoin laws. Then in the UAE and Hong Kong, and then in the U.S. with the GENIUS Act.

In all of these, stablecoins are designed as a cash-like instrument, a payment instrument. They’re designed as this form of money to be used in the payment system, and that’s really at the heart of it. That’s how these laws have been written, and it’s pretty consistent around the world.

The GENIUS Act does that, but as a result, under the GENIUS Act, Circle, as a payment stablecoin issuer, is prohibited from paying interest directly to stablecoin holders. It’s the same thing in Europe—we’re regulated in Europe. Same rule. Same thing in Japan. All these markets have the same rule.

But at the same time, we’re building a business. We do generate revenue, and we work with lots of different types of platforms, markets, distributors, and brokerages. We work with everyone from Robinhood to Revolut to Coinbase to Visa and lots of companies.

Those companies—stablecoins are really important to them, too, because they are how people hold money, make payments, trade, and do a lot on those platforms. They want to be able to pay rewards or have loyalty programs or incentives and other things. If they’re making money from the relationship with us, they want to be able to do that. And that’s what the GENIUS Act captured.

I think it’s good. I think it’s a good model.

So you’re happy with it as is?

We think it’s a very solid model. I think there’s some ongoing discussion about whether there’s a specific set of things that qualify for those kinds of loyalty and rewards programs and other things, and how prescriptive the law should be. It’s kind of getting relitigated a little bit.

The banks, my understanding, maybe want to try to kill it—maybe some of them who feel the most threatened. Is that the back channel here at Davos?

I think banks see stablecoins as both a threat and an opportunity. I can say we’re having more engagement with more banks in the world than we’ve ever had before.

Major banks, global banks, regional banks, banks all over the world want to integrate this and use it in payments. They want to use it in how capital markets work. Think about, “I need to post collateral on an exchange to trade, and I’m trading different assets.” Stablecoins are a better way to do that.

Whether it’s in capital markets, in trading, in wealth management, or in payments, there are uses for this. So definitely, banks see a lot of—

Because it’s faster and cheaper.

It's instant and it's free, essentially.

Yeah. If you talk to a global money-center bank, you could imagine one of those—I might say JPMorgan, something like that, or Citi, or one of these banks—and you ask them, “How many different payment networks are you integrated into as a bank?” They'll tell you they integrate with over 200 different payment networks. It's all over the planet, and it's all different standards, I would say—all different standards, all different systems, et cetera. So if you go to them and say, “Well, stablecoin networks are like a new payment network,” they're like, “I get it. This is—oh, it's on the internet. It has the attributes of how the internet works. We understand that.” And this will be your fastest, cheapest, and most trackable version.

That's right. And so, actually, we have a global systemically important bank we work with that's moving its own money between its own global branches using USDC because it's faster than going through the correspondent bank, and they trust it more.

Yeah, in those cases. So, bottom line, there's a lot of opportunity for banks.

But it is not—it's not a black-and-white thing. I think we'll figure out this rewards thing. I'm very sure.

Yeah. Amex figured it out, and video games figured it out. There was a little rattling even in those cases where they're like, “Hey, is this a currency?” People were trading it on eBay—trading my sword from World of Warcraft or whatever these nerds are up to—and they were like, “Is that like having your own currency?”

They're playing video games. Who cares? Can we just move on? There are more important things to do in the world than sweat Amex points being traded somewhere and pretending it's a currency.

You're up against an offshore platform that has, I think, maybe 3 times the amount of assets under management: Tether.

Tether—who's counting? 3, 4, 5 times. But you've been catching up pretty quickly.

Yeah.

USDC, among the largest stablecoins, has been growing faster.

You're number two.

For 2 years straight. Yeah, we're the second largest.

Tether's first.

And we are by far the largest regulated stablecoin network, right? Our growth in the amount of transactions happening with USDC is growing quite a bit faster as well. So, yeah, we feel really good.

I think our view has been, to the earlier question, that if this is going to be part of the actual economic system—if households and firms and corporations and financial institutions are going to use this and depend on this—whether you're a remote worker who's a software engineer working in Pakistan, a small business that's importing products from Vietnam that's in Brazil, or a hedge fund in the U.S. that's trading derivatives, you're going to want to know that you have something that is safe, fully reserved, audited, compliant, liquid, and available in the global financial system.

When we think about that opportunity, that's an enormous opportunity. The TAM of legal electronic money today is about $120 trillion and growing because of monetary easing and things like that. Of that, there's about $60 trillion, which is physical cash and non-interest-bearing demand deposits—kind of working-capital money sitting out there, right?

There's a huge amount of value stored and used as a store of value and as payment-system money, and stablecoins can grow into that. There's going to be an infrastructure for that that's well integrated into the global financial system.

And the business is—there's a float on this large base of capital that gets to underwrite your business. If interest rates are high, it's boom times. Tether's making over $10 billion a year on its float. You're—I'm assuming—making billions of dollars, or low billions of dollars, on your float. I'm not sure if that's public knowledge.

We're a public company. You can look it up.

Yeah, so I'm assuming it's billions. I'm not sure if you break it down by that.

We talk about reserve income as a line, and it's billions. It's a significant amount.

Yeah. And so, if we go into a zero-interest-rate phenomenon, this could be challenging for the business—or not?

Well, a couple of things. If you actually look at what's happened, when interest rates were very low, we saw 1,000% year-over-year growth for 2 years straight. Growth was off the charts. When interest rates started to rise, we actually saw declines in circulation.

Oh, why?

It's really the price of money, the opportunity cost of money. Interest rates set a kind of cost to holding money, so the incentive matters.

Yeah, the incentive matters.

What's interesting is that if you look at December 2023, when the forward curve—which was sort of the short-term price of money, in a sense, the market's view of the short-term price of money—started to fall because of the expectation that interest rates would come down, USDC started to grow.

From the peak of whatever it was—5.25%, 5.5%—down to where we are, 3.5% or whatever it is now, that's a 35% to 40% decline in the interest rate. We've had a multihundred-percent increase in the amount of USDC in circulation. So there's an inverse correlation there.

I've said this publicly in the media many times: When interest rates are really high, my view is that we really need interest rates to come down. That will help us grow. It will put more velocity into money.

The pie gets bigger.

Yeah, the pie gets bigger, which also impacts adoption. More people are investing capital in technology to transform their businesses and grow.

Right.

Technology, right? And so that is a catalyst. My view has been that this is really important.

I think our view is that there is some conceptual neutral interest rate. Given the persistence of inflation at around 2.5%, some people argue that the neutral rate should be—and the baseline of inflation should be—2.5% or 2.8%.

You know where the 2% came from? I talked about it in our previous All-In episode. I don't know if—

I do, but go ahead.

Yeah, New Zealand. There was a politician in New Zealand talking to the central bank, and he asked, “What should it be?” They said, “I think 2%.” He asked, “How did you come up with the number?” They said, “My instinct.”

America's been at 2.8%. 4% unemployment, 2% inflation. Both of those are going to be challenged because of very significant technological and macro forces.

Yeah. Job displacement and globalization have had impacts on this, depending on the country. The unemployment rate also depends on how many people choose to be employed. When you have a society that's doing as well as ours is—or the European Union is—some people just choose not to work.

The labor-force participation rate is now 61% or 62%. When you and I were coming up, it had peaked at 68% or 69%, so it was about 15% higher. It's really interesting to think about how that changes.

Tether, as a competitor, would not be able to operate in the U.S. There are lots of reports—I'll be judicious here—claims, reports, and concerns from politicians, the press, and regulators that they're doing things in dark areas that maybe you wouldn't want to touch in terms of the dark web and the dark economy.

The rumor is they're going to do a U.S. version of Tether and maybe try to go legit and get audited. How do you think about them as a competitor, given that they've been banned in multiple geographies and there were some very big concerns that they might have a run because they weren't dollar-for-dollar backed?

People had all kinds of claims about, “Hey, maybe they have Chinese paper,” and all these different concerns. They wouldn't do an attestation, or they would do an attestation with a bank somewhere in the BVI that nobody's heard of, and maybe a big KPMG doesn't know a lot about it.

I was not a Tether truther, but I did go down the rabbit hole because I reported on it on the pod and talked about it. They are a significant player. I'm wondering how you think about them coming to the U.S. and becoming more legitimate, and maybe trying to catch up with you in that regard.

Look, the beauty of having the GENIUS Act is that it creates a level playing field. It defines a federal law under which you have to come in and not just be audited, but be regulated by—if you're large—the national bank regulator, the OCC, which is a very serious, credible regulator.

We've received conditional approval for something called First National Digital Currency Bank, N.A., which is a national trust bank that we're setting up to be part of how we operate USDC as well.

But again, it's a level playing field. If you want to comply and build a product under that regulatory framework, you can do that, and I think many people will. This is like when net neutrality happened and common-carrier rules came: anybody can get into providing data services to the internet; anyone can build these things.

That's great. Free market, clear rules. Love it.

So my view is there's absolutely going to be more competition, and my view is also that the structure of this market—stablecoins are network businesses, meaning they actually exist as platforms and utilities on the internet. The USDC and our stablecoin network is literally the software protocols, but it's also the tens of thousands of applications that have integrated to the APIs.

Every time an app integrates—Cash App just said, "Hey, we're adding USDC support"—that's great. Now Coinbase has it, Revolut has it, banks have it, and Visa is using it. Every time someone adds it, Stripe merchants and Shopify merchants can use it. Every time someone adds that, it adds utility to the network and it adds network effects.

Then the next developer who comes along and says, "Hey, I want to build an app that uses digital dollars, et cetera. Which one should I use?" Well, I have interoperability with all this stuff. Great. So you have these network effects that are really key.

Then you also have what I call liquidity network effects, which is the ability for one to easily get it and use it within banking systems all around the world. So we've built out this incredible liquidity network by being regulated in Singapore, in the UAE, and in the European Union.

And this isn't an easy thing to do. I mean, this is a huge build.

This is a huge build.

Yeah.

It's a huge build, and now it's at a point where—

And it's defensibility.

It is. It's defensibility, and we have pipes, in a sense, where some of the biggest capital-markets participants in the world—companies like BlackRock—can enable a tokenized fund that enables USDC to come in and out of it, and their institutional participants know that'll work, right?

So I think it's a long-winded answer, but the answer is essentially that we feel like we've built a great platform and a great network, and we have really strong network effects. They're accelerating, and we've been publicly audited as a corporation.

Yeah.

As a New York Stock Exchange-listed, publicly listed, SEC-supervised company now for a period of time, but more broadly for a long time.

And when companies are going to choose what they're going to build on and what they're going to use, they're going to look at all that.

Yeah. The footprint matters.

It does.

And so the marginal value of a net-new dollar stablecoin coming into the market is essentially zero. All these things are needed.

What do you think about this concept that everybody's going to have their own stablecoin? There'll be an Amazon one. Obviously, PayPal's added one. Do you think that's realistic, or do you think people are going to be like, "That's just—I might as well just use Circle"?

Yeah, I don't see that. I think this is a lot like other internet platform markets. Everyone doesn't need their own data center. Everyone didn't need their own vertical search engines. Everyone didn't need to build their own video platform.

Internet-scale utilities achieve network effects, unit economics, developer flywheels—all these things.

And they tend to lower their prices over time.

It tends to get more and more economical and more and more capable. There's a lot one would need to do to do that, and I think even the biggest banks, who have pretty big technology budgets, may reach that same conclusion.

What worries you now? What's 2026 forward-looking like? What keeps you up at night? The opportunity is obvious. I think you've explained it perfectly.

Yeah, I think we're at an interesting place where we have a big piece of the regulatory clarity done. There are more pieces that are needed, so there's work related to markets and related to how digital tokens work. That's really important because we believe in tokenization as a phenomenon and smart contracts as a way that people are going to intermediate things on the internet.

So there's more that's needed there. I think one of the things that concerns me is the changing geopolitical and geoeconomic landscape. There are lots of different points of view being taken on where economic alignment is, et cetera. As a global company that's building technology that we want to make available globally, that introduces new complexity, right?

Yes. We are entering an era of new alliances and national champions.

Yeah, for sure.

And you're an American company, and America first, and so, yeah, you might be looked at differently based upon how the American enterprise is generally done.

Yeah, that is very true. Our headquarters is Freedom Tower in New York City. At the same time, the technology of these networks presents an opportunity to develop versions of this that are more geopolitically neutral, geoeconomically neutral, so that if you're from India, Brazil, Southeast Asia, or the Middle East, you can build on this.

These are high-growth markets that you're comfortable building on, with a variety of different nexuses of economic relationships. So that's something we think about, actually.

Yeah, like a euro-based stablecoin.

We have the largest euro-based stablecoin.

How many currencies have you done that with so far?

We've really only focused on dollars and euros.

Why?

Well, I think those are 2 widely adopted currencies, both as a store of value and as freely floatable, circulated, tradable currencies. We have a big commitment to the European market, and so it's also part of just being committed to that market. We think it's a big opportunity.

For everything else, we've really built a technology stack for other stablecoin issuers from other markets, now that there are laws around the world.

Ah, so if they want to do their national stablecoin, absolutely centralized. And are you doing that with Bermuda or some sort of partnership?

The Bermuda partnership is related to helping them have digital dollars and have economic activity, commerce activity, and treasury activity all happen on-chain.

Got it. So that's an infrastructure play, not—I don't know what their dollar is.

Well, it's their dollar. It's a dollar-backed currency. They're already their own stablecoin. But in other places—in the Philippines, Mexico, Brazil, Japan, and other markets, or Australia and Korea—there are stablecoins coming, and we want to make sure that we can provide technology so that they can do all the same—

To the government—

Or to banks.

Well, more to the private sector—

Because stablecoins are private-sector-issued, but these companies all need to be regulated by the government. They need to be following the same rules.

So it's easier for you to partner with a major partner in those markets.

Yeah. We want to see that grow and flourish. This idea of an internet financial system, where all the economies of the world can be on-chain and all of the contracts and financial contracts and markets and capital formation, lending—everything—can happen on-chain, is obvious.

Small businesses and poker players—people who are concerned about their fees and are focused on that—are being driven to stablecoins. They get it. They're like, "Yeah, I'm doing my design job. I'm a designer in the Philippines working for somebody in India." They're starting to figure out, "Hey, I don't need to pay these." And then consumers are starting to figure it out. Maybe if they're—

Like me, a poker player, you're sending a lot of wires. Every year I send tons of wires back and forth for different poker games.

Yeah.

Well, now, on both Polymarket and Kalshi, you fund your account with USDC, right?

Right? Speaking of that kind of—

And, well, that's because that audience understands. They're always looking for an edge.

Speed, speed—

But they're looking for an edge in a poker game. If you were a novice and I was an expert—I was Jason and you were Phil Hellmuth, whatever it is—you know, it's got a bigger edge on them.

It compounds after 100—

Poker games to be a bigger number.

It does.

This is why the rake matters, and gamblers and people like to wager things.

You know who the biggest users of USDC are?

Yeah, that's what I was kind of getting at.

It's actually a form of gambling.

The biggest electronic market-making firms in the world.

Got it. The guys who need every edge in terms of milliseconds, cost, and capital efficiency to be able to move capital in markets because, at the end of the day, they are algorithmically trying to figure out the very next best thing to do. If they have dollars that operate with the physics of the internet and the cost efficiency of data, they have an edge. And so they love it.

They love it. And then the people who are sending money home to their family, they love it, too.

They love it, too.

Yeah, because they're like, “Wait, I'm paying Western Union 12% of this?”

Right. They can have a wallet and directly send to their counterparty, and it's like, “Wait a minute. This literally arrived.”

There's a proliferation of these digital wallets now that can also hold USDC but spend it through Apple Pay, which are being issued all around the world, too. People can store value in digital dollars, use it through payment terminals, and that's pretty cool.

Intuit thesis there, because that seemed like there was something underlying that I wasn't getting to. I mean, it's obvious that there's money flowing through there, but what's the big picture?

Well, Intuit's an amazing company. Obviously, they've been around for a long time. They have franchises that are huge. Credit Karma is a huge franchise—actually huge—and they're growing what they provide to people through that franchise. QuickBooks, obviously, we all know: if you run a small business, you use QuickBooks. Actually, people invoice through QuickBooks, and it's trillions of dollars of transactions a year that are invoiced through QuickBooks.

If you can invoice people and settle in USDC, that's better than ACH, and it's faster. You can imagine a world where small businesses can gain an advantage from something like that.

Reconciliation: you're a TurboTax user, you file your taxes, and you get a refund. Imagine being able to get your refund instantly.

So I think there's a lot of cool stuff in it. It's a very innovative company. It's a tech company that's in these financial technology adjacencies, and so we're super excited about the collaboration that we have.

I don't want to say there's a loser in this, but there are people who are going to be challenged by it. How do American Express and Bank of America look at Circle and stablecoins, in your mind? Because they must call at some point—or do they not?

I think all these companies have an opportunity to use this technology. They have an opportunity to use this technology to improve how they provide payment utility. They have an opportunity to use it if their customers want to interact with the whole on-chain economy and on-chain investments, and what's happening there from a wealth-management perspective. I eventually think that there's going to be more credit products that are actually built using stablecoins.

How would that work? And what's the advantage?

Well, so already today, there have been trillions of dollars of loans made in stablecoins through DeFi protocols. Yes, you take your USDC and effectively deposit it into a protocol, and on the other side of that protocol is a borrower. You're paid an interest rate for the amount of time—

You become the house. You become the bank.

The individual does, actually. I mean, as someone years ago said, these are like self-driving banks. They're software machines where the risk management, the collateral management, the liquidation—all of the liquidity that's there—is all just smart-contract machinery that we can all observe in real time, perfectly auditable in real time and transparent. You don't have that with a bank, so you have a really interesting thing there.

Right now, a lot of that is lending to people who are borrowing to do things like investing, like margin loans. If I own $1,000,000 in Bitcoin and I want to spend $100,000 of it, but I don't want to liquidate any Bitcoin—

Yeah, I want to spend $100,000 of it, but I don't want to liquidate any Bitcoin. Somebody else can make some money.

And you can borrow USDC against your Bitcoin and so on. But I think this is the early stage of that. Our view is that if you have these digital cash things like USDC, you can create lending protocols that are lending for, “I want to hire a new employee,” or, “I need some new equipment for my kitchen and my restaurant.”

Factoring, equipment leasing.

All these forms of lending can be done, and the risk that's taken—there's real risk there—can be underwritten. The insurance on the risks of failing can be priced. You can build credit markets entirely in software, entirely with software machines—no storefront coming into play. AI plus these sorts of smart-contract machines and stablecoins, I think, creates a really interesting little cauldron for credit-market innovation.

Yeah.

I have a gleam in my eye, which is: imagine a credit market that worked like AdWords. Imagine something that was that efficient and could clear and settle credit decisions at the speed at which an auction happens for attention.

Yeah, and so those kinds of things will become possible, and that probably would be pretty good for people who are on the other side of credit need.

Yeah. If you think about monetary velocity in the economy, you can get more things moving.

Exactly.

More jobs created, more chances, more swings at bat, more shots on goal. So money velocity increases. I think these kinds of credit-intermediation models powered by AI and blockchain networks can actually further increase money velocity.

We have a mission statement to increase global economic prosperity through the frictionless exchange of value. That is literally our mission statement. People a lot of times think exchange of value means, “Oh, like making a payment.” No. Value exchange is sort of time-value-of-money transformation. I have value I don't need right now; you have a need for that value, and I'm going to use time to transform that and then create new things from it. That's where actual growth comes from. So that's what we want to see.

One of the interesting things, since we're here talking politics at Davos—you and I are talking tech, but politics is one of the big topics here—we'll sort of end on that a little bit and get your ideas and your thoughts on it. There's a movement toward socialism in New York City, my hometown. It's kind of heartbreaking for me to watch.

California is seizing billionaires' assets, literally asking them to write down and audit everything they own, put a value on the painting or the piece of jewelry they bought for their spouse, and we'll just take 5% of it once. They lost 200 billionaires and $1 trillion worth of wealth, and they don't seem to have any problem with it. As a technologist, a builder, and a capitalist, as long as I've known you, what do you think about this moment in time that we Gen Xers are living through, watching this socialist movement?

It's interesting. I just moved to New York City.

Just in time. By the way, that's costing you, I think. You're going to have to pay 2% more a year, but I don't think you're an income guy. I think you're more of a capital-gains guy.

Yeah. Look, here's what I would say. I was with Dario Amodei from Anthropic a couple of times over the past couple of days, and he obviously talks a lot about the economic disruption, labor-market disruption, and what's going to happen with AI.

He's not a doomer, but he has deep concerns.

He has deep concerns, and I also do. When I look at the very positive thing, I think we're going to see an acceleration of GDP growth. Pick your time frame: 3, 5, 10, 15 years. We're going to see, I think, an extraordinary period of economic-growth acceleration.

Even more than the internet and PCs?

More than the internet and PCs. This is a very, very strong, very, very big transformation, but most of the labor is actually executed by AI machines.

Yeah.

And the capital accrues to the capital owners—

Right?

And that's just kind of there in front of us.

Yeah. It's the fundamental nature of capitalism.

Fundamental nature. I think that is just going to challenge us in ways that we haven't dealt with in a long time, since other periods.

Labor was necessary for capital to succeed.

That's right. The Industrial Revolution was mechanized human labor, right?

Yeah.

This is mechanized AI.

By the way, the CEO job seems ripe for AI doing a better job than we would do at times, right? I don't know if you've—

I use AI all the time in my work.

Do you ask? I don't know if they realize what they're interacting with.

No, but I mean, literally, when I was talking to Brian Armstrong, our friend, he said he's got all of his data in a proprietary LLM internally. He said this on the program; I'm not speaking out of school. He asks it, “What do I need to know about my organization?” It's like, “Oh, there's nonconsensus about this important issue in the corner over here.” And he's like, “I didn't know that.”

Like, whoa. Is that a coach, or is that like a parallel?

Yeah.

Like another CEO? Is it a duplicate CEO? Did you clone yourself? How do you think about that?

I think about it all the time. I'm pushing very hard to encourage all of us inside Circle to do more with AI. We're making sure, obviously, that we keep safety and compliance in mind because we're a regulated company.

Compliance matters.

Yeah. So there's a lot of work. We invest a lot in that. Nonetheless, creating the avenues so that awareness and understanding can be there, and so that as leaders we can turn to that, is important. I put a Slack post in a Slack channel recently that sort of said, if you want to become a great manager at Circle, probably one of the best things that you could learn how to do is manage AI.

Yes.

And so, invest in how you manage AI and AI agents. Actually, that's how you might become a better manager and leader through that, and I mean that—

It is one of the—if you think about the tasks of these product managers who would run the standups, or that mid-level manager historically, it's to keep tabs on a group of people and their targets and their time frames, who's excelling, and who's got blockers. AI is kind of better suited for that.

Really good.

And so, the great middle manager—which is a derogatory term, but let's just use the word manager—a great manager should be able to manage 10 groups using, say, Claude Cowork. I don't know—have you been playing with it yet?

It's incredible.

Wow.

It's really incredible. It's been out for a week, and I'm like, I'm sitting there like—

We haven't deployed that inside the company yet, but I've looked at it and used it outside.

What impresses you about that?

The ability to take tasks that I have to do, teach them, and then have it just do them is extraordinary. I'm trying to get this deployed in my household, too. It's really amazing.

An interesting way to do it is—because I went down this rabbit hole of estate-management stuff, like when you have to deal with this after you get a second home, or however many you wind up having. Notion is, I think, the best solution right now because Notion put AI into it. Do you use Slack personally, too, for your domestic affairs?

You don't have a domestic Slack?

I don't have a domestic Slack.

Okay. I have a domestic Slack and a domestic Notion. So I took the stack from the office and just did it at home. What that did for us was, the people who bridge the family office and this can kind of—they know the same tool stack. Then Notion has been investing in its AI so much that when you do a query of your Notion and you're like, “Hey, this house with this HVAC issue,” it's like, answer. You're like, “Oh my God, that would have taken 3 phone calls and an hour.” I think it's going to change everything about how we work.

Yeah. What else did Dario think? I might have cut you off before you—

Yeah. I think the big picture there was just that, whatever your politics are, whatever you think about taxes, whatever you think about the social contract, all of that, I think, is going to have to be thought about again in the coming years.

Yeah, in different ways.

And I don't think it's going to be the straightforward answers that we've turned to.

No, it can't be.

And so that's why my answer is, yeah, there are things California, New York, and others are doing, but I think when we look at that in comparison to the broader political economy that we're going to have in front of us, those will seem modest in comparison to what we've got to think about.

By the way, we fixed it in startup land long ago. It's called stock options. If you look at the Trump Accounts, which Invest America, Michael Dell, Brad Gerstner, and a number of our friends worked on, 42% of Americans don't own an equity and are not part of that. Let's just say it's 50%, because some of them own equities kind of passively. Half the country doesn't own equities.

If everybody had equities in a superannuation fund like they have in Australia and they were watching them go up, then you're not like, “Well, right, screw those guys at Amazon and Apple and Google.” Being stakeholders in the success of this technological transformation is critical.

Yeah, that would solve so much of the problem. Jeremy, I could talk to you for hours. You and I have talked for hours in our lifetime about all these paradigms. Continued success. It's just a pleasure to spend time with you. It's always one of the great discussions I have.

That's awesome. I can't believe it. George Kurtz is here. He's the CEO and co-founder of CrowdStrike. This is your second Davos.

Yes.

Why are you at Davos? Obviously, everybody knows CrowdStrike.

Yeah.

We'll get into it. Why are you spending the week at Davos?

Well, I have all the world leaders here, and all the best company CEOs are here. It would take me a year of activity and flying around the globe to meet all the people that I need to meet, so it's just a fantastic opportunity to meet customers and get business done.

Yeah, it's efficient.

Very efficient.

It's this tiny little ski town.

Yep.

It's not fancy at all. This isn't Aspen, and everybody's within 5 blocks once you get through security.

Yeah.

Speaking of security, AI has had, or is going to have, a profound impact on defense and also offense—black hat, white hat, everything in between. What is the attack vector that AI is most effective at, sitting here in 2026?

There are a couple of different areas. One is, if we think about the adversaries, I kind of create a pyramid where you've got nation-states at the top: very sophisticated, but fewer of them. The middle band of the pyramid is eCrime: more of them, a little less sophisticated than the top. Then the bottom is just hacktivism.

Essentially, you're minting new adversaries because you don't have to have all the knowledge that you had to have in the past. You can ask any number of LLMs, and you can get answers back. What's happened is that the attack timeline has been compressed. You can automate all of these attacks, and you can do it with a level of sophistication that looks like a nation-state.

That's one of the greatest areas of exposure: the speed and sophistication have dramatically increased.

Because the answers have been compiled.

Yeah.

The scenarios can be run even by somebody who is a level-6 hacker out of 10. I'm just making a number up here, and they can become an 8. So just like a developer using Copilot goes from being an average developer to an above-average developer, the same thing is happening with hackers.

Absolutely.

One of the things that we're seeing is autonomous malware. Most people are probably familiar with malware that runs on your computer. Typically, people would use antivirus to protect against that malware. But what we're seeing now is prompt-only autonomous malware, meaning I can drop the prompt on your computer by a variety of means. Then it will autonomously interact with an LLM and give a unique fingerprint every time it runs.

Your computer is different from the next guy's. You have different data, and it will begin to prompt its way to getting what it needs without ever phoning home to someone controlling it.

Oh, wow. Just to unpack that and translate it, I did a little hacking in my day, in the '80s. A little phone phreaking, mainly. We'll get into it offline. I think—

I had a Hayes 2400, guilty as charged. I started with the Ventel 300 baud. I think we're of the same generation.

I had a 300 Hayes.

You had the 300 Hayes.

It was a tank.

It literally had such a great form factor that's never been repeated, this great rectangle. But we're old.

We are old.

People are now saying, “Hey, I'm going to use the Comet browser by Perplexity.” ChatGPT has a browser. Claude has an extension. It's really wonderful to say, “Hey, get all my emails from LinkedIn, put that into a database, and add it to my CRM.”

Wonderful. You watch it work: “Get me the cheapest flight.” The same thing can be used by an adversary to get on your computer.

One of the great ways is detection in security. You're in the business of detecting the attacker. But what you've just informed me of, which kind of blew my mind even though it's completely obvious, is that the LLM can say, “I'm going to do 1,000 attacks today, and I'm going to make each one different and unique.”

And unique. Absolutely. They're unique.

So scary. They're not actually code because they're prompts, right?

Traditionally, malware would phone home. There would be a tether, right? So you'd always be able to see the signature of something phoning home. Now these can run autonomously without ever phoning home—

Like a sleeper agent. You drop them into America, 20 years later they start pursuing, but they never have to return to base.

Exactly.

How does one counter that?

You need AI to counter it. That's been a big part of the success at CrowdStrike. When I founded the company in 2011, it was using, say, AI—machine learning at the time—to detect malware that had never been seen before.

Now, obviously, that's evolved into Gen AI, and we're really leveraging the large data set that we've amassed over 14 years to train our own models to be able to counteract what the adversary is doing with the speed that you need. So, if you think about it, there are only a few ways to break into a bank. I'll use this analogy, right? You can drive there, you can walk there, you can use a gun, you can blow the safe, but at the end of the day, you've got to get the money and you've got to get out.

Yeah.

And it's the same way in the computer world. There are certain things we call indicators of attack. We've trained our models to look for these, so it doesn't matter what you look like or what car you're driving—we can identify that. A big part of our success has been the AI we've built over the years with this tremendous data set.

Yeah. So you might be able to get into the bank, but at some point you've got to leave with the diamonds. You've got to get to the safe.

Exactly.

That's where you can catch them.

Exactly.

You mentioned state actors, and you mentioned for-profit. When we look at a country like North Korea, they need revenue. Hacking is a pretty great revenue source. They were even using AI, my understanding—and I'm sure you've been following this—to get developer jobs in the United States and convince Americans to put laptops in their homes so that they could work remotely. Talk to me about that attack vector, actually infiltrating companies as employees because of this remote-work nonsense.

Well, we were one of the first to ever find that. We were developing some new AI algorithms.

Yeah.

We saw something that was called signal, which basically strips the noise from the signal, and we saw the signal and said, "This is really weird." We investigated it and said, "Okay, this looks like somebody using remote tools. What's going on here?" We went further, and then we said, "We think it's North Korea, and we think it's an employee who—or the company thinks it's an employee."

When the R&D team came back to me, they said, "Okay, we want to notify our customers and say we think we found this, but we have to tell them that the employee they think is an employee isn't really their employee."

Yeah. The phone calls are coming from inside the building.

Exactly. So for us, when you have a piece of malware, it's like, okay, that's bad. You can detect it. But when you have to tell a company that their employee may not be their employee, you have to be gentle. You're letting them know just how incompetent they are when it comes to security.

It's a little gentle.

You have to be gentle. You're letting them know just how incompetent they are when it comes to security.

I said it, not you.

Okay. Well, it's hard to find these guys, but in any event, we found 40 of them. We were right—40 of them.

That's for the first run. We found hundreds now—hundreds in America. Yeah. Now, they were doing it, my understanding is, for the high-paying salary to fund stuff in North Korea. Is that correct or not? Were they doing it to get trade secrets?

They were doing it to buy access. Why break in when you can just log in?

Oh my God. Yeah, you know, so it's a lot easier to get somebody hired.

True story. We actually said we thought this employee wasn't an employee. The company investigated it, and they said, "Yeah, you were right." I said, "Well, tell me the story." They said, "Well, we went to this person's boss and said, 'We don't think that's a real person.'" They went through all the reasons why. Finally, they said, "We think it's a North Korean." The guy's boss said, "Well, do we have to get rid of him? He did such good work."

Oh my lord. So the value of a top-tier developer is so high that you'll take a spy.

Yeah, exactly.

It's like, you know, we kind of need him to ship.

Yeah, he was doing such good work. He was shipping. This is one of our best performers. What now?

Remote work has opened up so many vectors. You believe in remote work. Do you support it in the company?

We do. Yeah, we started the company remote-first.

But this opens up all these attack vectors. What's—

Even without remote? Well, best practice, really, and what we see companies doing is—let's take a simple one: meet whoever you hire.

Wow. Shocking. Shocking.

But with COVID and remote work and things of that nature, there are a lot of people who get hired that people never met. You'd never see them on the screen, but the work would get done. Emails would get written, and then these things happen. So what we're seeing now is that the best companies are actually embedding a security person in the HR group, where they're prefiltering all these resumes because the resumes are AI-generated. The LinkedIn profiles are AI-generated. So you want to catch them up front rather than in the interview process.

A crazy idea: for your final interview, can we fly you in? Or, hey, week 1, you're going to be at headquarters?

That's it.

That's it. You solve the whole problem because then they don't take the job. They move on to the next person.

That's it.

Okay, so we're looking at state actors. Russia is still the number-one state actor.

China. Russia. I mean, it depends. If you look at pure intelligence, pure—

Ability.

Russia.

Russia. Yeah. Why? Why are they so good?

I mean, like the Cold War, you could say it's over, but you have really smart people who went to work in various areas, right? Obviously, from a nation-state perspective, but also from an e-crime perspective. They're smart. They know what they're doing. Typically, they're what we call low and slow. The Chinese over the years have been a little bit noisy, kind of a smash-and-grab. They've gotten better.

Not subtle.

Not subtle, but much better now. But the Russians are very targeted. They're patient, and they're not always gathering IP information to steal it. They're gathering intelligence for reconnaissance for nation-state activities, and then they moonlight for e-crime.

And let's go to China, then. If we were sitting here 10 years ago, we'd be talking Kumbaya—China, we're all going to win together. The decoupling has happened, essentially. Do you have operations there? Do you work with Chinese companies?

We don't have operations there. We don't have any revenue. We don't sell into China.

Never have.

Never have, from the start of the company.

And what's your take on the decoupling? Let's go a little geopolitical here.

They are our adversary, obviously, and a competitor. You know, their technical ability is pretty strong, right? We're seeing that with AI.

Are they the future competitor and the future attack vector that you have to worry about—China?

Well, I don't know that it's the future, because I think it's already there. It's here, and it's been here for the last 20-plus years. Some of it may not have been talked about in the early days, but they're very, very prolific. They're very good. Obviously, nation-state is always going to be nation-state versus nation-state. That just happens. The big thing with China is that there's a commercial aspect to what they do. If the government breaks in and steals a trade secret from some company in the U.S., they'll just give it to another company in China.

Right? That doesn't happen necessarily with all nation-states. Yeah. We're certainly not breaking into China to steal their trade secrets and giving them to a U.S. company. The CIA is not breaking in to figure out what BYD is doing and give it to Tesla or something, correct?

So it's a one-way relationship there.

Now, there's also something very unique: they make hardware. They ship a lot of hardware, our laptops in some cases, Huawei, other companies. They're putting spyware or back doors into all that hardware. In your estimation?

Over the years, there have certainly been cases where people have had concerns about what was actually shipped. Supply chain can be problematic, whether it's actually shipped with something or, in some cases, packages get diverted somewhere else. The malware gets installed, and it gets reshipped out, right? So you've seen that.

It's important to have good hygiene on whatever equipment is being brought in, and to know the provenance of that manufacturer and the software that goes on it. That applies really around the world.

Do you trust Signal and your iPhone to protect you as a corporate executive? What measures do you take?

Well, because you're a target.

Yeah, we're a target. Our team spends a lot of time making sure that we try to do the right things.

So WhatsApp and Signal—what about those two? Do you think those are attack vectors? Can people do insertions there?

Well, I think you have to look at privacy versus the actual attack vector. If you're clicking on links in Signal, that doesn't mean the link is secure, whether it could be Signal or something else, right? So there are a lot of ways to get implants on these devices. I mean, I would say an iPhone is safer than just a regular computer because of the way it runs.

But if you look at the nation-states and some of the creative ways they can actually get into a phone, as an example, if they want to get in badly enough, they're going to get in.

How do they get in? What's the attack vector?

They get you to click on a link. They insert something.

Typically, what they'll do is—there's a whole black or gray market on the value of a zero-day. There are zero-days in all software, and if you have one, say, for Apple iOS, that's super valuable. Generally, if you click on something and it's an unknown zero-day, you can be compromised just by going to a browser.

Explain in plain English to the audience what a zero-day is.

Sorry, talking technical jargon.

Not everybody's got the hacker bona fides.

I know not everybody's a hacker guy, but if there's a software vulnerability that hasn't been found and patched, it means they can exploit it and run code. In some cases in the past, there was a vulnerability where they could simply send you an SMS message—you didn't even know; you didn't have to click on anything. The SMS message would process this information, and then they would put an implant on the phone.

Biometrics, two-factor—this has changed everything in the industry and made it much more secure.

More secure, but not much more. If you look at a lot of the attacks, they're identity-based attacks. What you see is that even when you have a credential, or if you steal a credential, or if you have a session token—which means that you've already authenticated—

Yeah.

Right. So even if you have two-factor, if you have the session token—

Yeah.

You can replay that and get access to a system.

The crypto people learned this the hard way.

Yeah. They were basically using two-factor over SMS.

People call the phone company.

Reset the SIM.

Yep. And then they got you.

And game over. Now your Bitcoin wallet is emptied.

Correct. Where there's money, the hackers will figure out a way.

They figure it out. Humans are normally the weakest link. The way I was able to get my first exploit was I called the New York Public Library. I knew they had a VAX system with dial-up.

Okay.

I just said, “Hey, it's Joe from IT at the 34th Street branch. I need the number for the dial-up.”

Yeah.

Literally, the person gave it to me over the phone. They wanted to be helpful. I dialed it, and then I could search the stacks and search all the books. That was the only thing you could do.

What year was this?

1987.

Okay.

Yeah, I was at Fordham, and they had BITNET. They had the internet there, and I was just starting to see these things. But it's almost universally about calling somebody on the phone, meeting somebody, or compromising somebody.

That's it. That's it. The weakness is between the keyboard and the chair. The keyboard and the chair. What is the best practice, then, when you're advising your customers and explaining to them just how the human-factors part of this is the most important one to focus on?

Well, there's certainly an educational element, and you also have to look at how people get paid and the motivation. Typically, they're going to call a help desk. How does a help desk get paid, and what are their metrics? Their metric is: get a call, open a ticket, and close it as quickly as you can.

The incentive is speed, right? Particularly if it's a third party out in another country.

Right. Their whole job is to open a ticket, solve a problem, and close a ticket. If you follow the money and the incentives, you can see why there's a problem. People want to be helpful at the help desk, and hence they get into these situations. First, start with the education piece, and then have the right controls. Identity protection, things like what CrowdStrike builds, are going to be additive to making sure those sorts of things don't happen.

People using their own devices—bring your own device—is the other major attack vector. Corporations make this mistake.

Yes. Your own device could be a cesspool if you let your kids use it.

Or it could be pretty good, but that's the reason why now there are enterprise browsers to contain what people do, even if you bring your own device.

Yeah.

The enterprise browser runs on a virtual machine at some headquarters. You provide that kind of situation. That's sort of the old-school way of doing it. So what we've done is actually acquire a company called Seraphic Security.

I saw that. Explain the state of the art there.

The state of the art is not switching out a browser. If you use Chrome or Firefox or whatever browser you want to use, you can keep the browser you have. It's actually running beneath the browser and will support any browser. It basically looks at the interaction and what the browser is doing, and if there's malicious activity in the browser or an identity that's compromised, you can stop it at the browser. You can wall off what you do with the browser, including what data you take out of it.

It's the front door now for how people work, and how adversaries get in is through the browser.

How should companies implementing AI in the enterprise put their data into these clouds and work with partners? Claude Code this week kind of blew people's minds. I don't know if you saw the announcement or played with it.

Yeah.

I was using it earlier this week, and I'm authenticating on Gmail, Notion—oh yeah, my desktop files. I got halfway through this, and I was like, “This probably isn't a good idea,” but it can be pretty scary. We don't want to throw them under the bus, but it was a lot of authentications I did.

Exactly. They're doing great work, but there was a customer who basically created a whole suite of AI agents to help with automation in their IT department. They had one agent looking for IT problems and software bugs, and it found something while the code was being checked before it was committed. The agent said, “Hey, I found this bug. I want to fix it,” but it didn't have access to fix it.

So it went to the Slack channel that had the other 99 agents and said, “Hey, all, does any other agent have access to this thing? I need it fixed.” An agent raised its hand and said, “Oh, I have access, and I can fix it.”

Do you see how scary this is?

This is unintended consequences. These LLMs are essentially guessing what you want them to do. They're reasoning it: “It is reasonable for me to go ask for help.”

Right?

It is reasonable for me to give help. Now, what if it pushes the wrong code? What if it makes a mistake? How do you ever track that down? Who's monitoring these agents? The agent technology has unlimited upside, but my lord, you're going to be in business for a long time.

Well, this is it. It's called AIDR.

AIDR.

Yeah, AI detection and response.

Got it.

There's a concept that's been around that we helped pioneer, which is endpoint detection and response. On your computer, most companies have it: you can monitor and see what's happening and prevent bad things from happening. We have to do that now with an agent, and this is why it's a huge opportunity for us. On average, each employee is going to have about 90 agents they control.

We're going to have protection and visibility across all of those agents, whether it's from a third party or whether it's a homegrown agent. That is a massive TAM opportunity for us.

Do you trust the LLM operators with CrowdStrike's data, or do you want to stand up your own large language models?

We have our small language models, and we work with just about all of the large language models. We've built our own guardrails around what gets shared, how it gets shared, and how we process any of the data back.

Are you concerned about it? I can see you're thinking, “Hmm, yeah, maybe there are some concerns here.”

No, there are. But we tried to engineer for that.

Got it.

Right, because you want to leverage—and this is the thing—tokenomics. There's an economics to how you use the tokens as well. Plus, you want to get the right data. We've built small language models for things that we control with our data. Then, for other bigger, frontier-type models, you're not going to replicate billions in R&D. You want to leverage what's out there and take the best of it.

Yeah. You're also a race-car driver.

Yes.

And you've done well in your life. You got to buy a piece of an F1 team. I've been going to F1 races—I went to 3 this year. Tell me everything about it.

Where do you want me to start?

How did you get the bug? When did you get the bug? This has been an American thing in the last couple of years. I can't believe the racket these F1 guys—well, you guys—are doing. Somehow, you made this the most important thing in America ever.

I knew to survive was—

That is a big driver.

And then it’s a luxury experience. So now I’ve gone to Miami, Austin, and Vegas. I’ve been to 3 in the last year.

I hadn’t even heard about it 5 years ago.

That’s fantastic. Well, a little backdrop: I grew up—I had zero money as a kid.

Where did you grow up?

New Jersey.

Oh, really? I grew up in Brooklyn.

Yeah.

You know, we used to play ball.

Yeah, exactly.

Okay.

So, I always liked cars and fast things and never had the opportunity to do that. My first car was a Toyota Celica.

Toyota Celica. Yeah. A ’73 Mustang Grande, 351 Cleveland engine.

351 Cleveland engine and slightly more horsepower than yours.

I know. Mine was just—I had a commute in the thing.

Yeah. Okay. It was functional.

It worked. It was functional.

But I never had the opportunity to do it. I always followed sports car racing. Mostly it was Indy, and then I got into racing later. We’ll talk about that. But I’ve been in Formula 1 since 2018 because CrowdStrike became—well, Mercedes became a customer of CrowdStrike.

Nice.

And then I met Toto Wolff, who runs the team, and he said, “Hey, why don’t you think about working with us?”

You got in at the right time.

Good. That was a good trade. It was a good trade. So, we got in then, and over the years I built a great relationship. Then, just last fall, I put a deal together and now own 5% of the Mercedes team.

Congrats. And now you race yourself. You’re in the Bronze league, I hear.

Yeah. So, I race. In fact, I have to go back for a race in Daytona.

Nice.

Still got the 24 Hours of Daytona.

What do you drive? What kind of car?

LMP2.

Explain.

It’s a full-fledged race car. I mean, it looks like a—

Like an F1 car?

With covers over the wheels.

Got it. So, it’s safer.

Yeah.

What do you think of this new Corvette ZR1X? Have you seen this beast?

I’m trying to get one.

Yeah, fantastic. I know. I’ve got a contact at GM.

Okay.

Everybody’s got a guy. They gave me an E-Ray to play with.

Yeah.

Which is kind of the precursor. They took the E-Ray and put it in the ZR1X.

Yeah.

My lord, off the line, it’s like having Tesla acceleration. Then you have the V8, 600 in that one. I think it was 600 horsepower plus 150, which gets it to about 800.

This new one is 1,000 horsepower plus 250.

The thing costs $200,000.

And for the money, it’s a tremendous car. I still think to this day—I mean, someone keep me honest—that the ZR1 will still have a warranty even if you track it.

Really?

Yeah. It’s one of the rare cars. At least it used to be.

Yeah. They said you can take the ZR1 and the E-Ray version, and you can change the driver profile, lower the RPMs, change when it shifts and everything, so you can actually talk on the phone and have a conversation without making the passenger throw up.

You want to hear the engine.

I think that’s right. I had a C6 convertible in that generation, which was just dynamite. Canary yellow. What color are you looking at? This is where we’re going to get real. You saw that crazy green.

I like the blue.

You like the blue? It’s beautiful. They have a really beautiful, vibrant blue.

Yeah, I was looking at the silver. I don’t usually like silver on cars, but it looks tremendous on that car.

I keep all my Mercedes silver.

You do?

Yeah.

Interesting. All right, listen, the question everybody wants to know: Greenland, what should we pay?

How do we make a deal?

I’m sure there are lots of deals—that’s above my pay grade—but I think whenever you see geopolitical tensions go up, there’s always more security activity.

Yeah, good for business.

It’s good for business.

Seems like something we should—

The Trump administration is spicy on the margins, as you may have seen.

But they’ve been good for business. Yeah, supportive of the business community.

I think so. When you think about what the administration is doing with security, they’re taking a business-first approach. You know what that means? They want to save money. They want to consolidate. They want platforms. They want better outcomes, right? And they don’t want this piecemeal buying across the government.

So, absolutely, I think they’re doing a great job on security.

I’ve been asking people’s perceptions, and it’s overwhelmingly been—

Yeah. They’re calling us up. They’re at the table. They’re engaged.

So, whatever you think politically—all these cultural issues—they’re listening and engaging with the business community, which I think is a great thing.

All right, brother. Thanks for doing this. Talk soon.

Transportation, of course, is one of the most important parts in the business sector, and eVTOLs—we were promised flying cars. We still don’t have them. But my guest today, who’s spoken and is a friend of the All-In podcast, is going to tell us how close we are to getting rid of these noisy helicopters buzzing around and having eVTOLs, nice and quiet ones. Adam Goldstein, CEO of Archer Aviation. Welcome back.

Thank you. Good to be here.

Two years ago, you were at the summit. You promised us we’d get flying cars. When can—I mean, let’s just get to brass tacks. This is all anybody wants to know. When can I go from Manhattan to JFK? When can I go from San Francisco to Oakland Airport? When is this going to happen? And is it going to happen in the United States first, or is it going to happen in the UAE or Saudi Arabia, or maybe China?

The hardest part about bringing these aircraft to market is certification. We have to prove that these aircraft are really safe. You can’t blame the regulators. It’s not a regulatory thing; they actually have to be very safe and reliable.

So, the rules got put in place. But the most important thing that happened in recent history was that President Trump issued an executive order that was fast-tracking the program through the regulators and really starting to create a platform for us to launch. There are now 5 cities going to be announced in the first quarter, and then we’ll start flying in the summer.

That will be the first time you’ll see these aircraft flying around the cities on a regular basis. That will allow the general public to get comfortable with this, and they will certify us sometime after that, because the political side of this is not a Democratic-Republican thing. Consumers have to feel comfortable with this. They have to watch this, and we don’t want to be fighting that. They’re giving us a chance to go do that.

Okay, so it’s 2026. We’re sitting here in January of 2026. You guys are going to announce 5 cities.

The government will. DOT will.

Oh, here in the United States, they’re going to announce them. These are for all eVTOL companies—Joby, yourself, all the contemporaries. They’ll get to do those 5 cities first.

Correct.

You have any guesses about what they could be? What are people thinking, and how are they selecting them?

My big push has been around Huntington Beach, right around Los Angeles, because we won the exclusive for the LA28 Olympics. That was a big deal for us, so we need to start trial operations and really ramp up operations, because it’s challenging. We also recently bought Hawthorne Airport, right outside LAX, to help give us a hub.

Hawthorne, the private airport. I’ve flown out of it many times.

Exactly.

It’s a great platform.

It’s a great platform.

Wait, let’s just let this soak in, folks. Archer Aviation bought an airport.

We did.

Can you just buy airports? It’s a privately owned enterprise. What did that cost you?

Yeah, you can. I’ll call it a sort of buyout. You can control it. There’s a runway that’s owned by the municipality, there’s all the property around it, and then there’s control of the master lease. It is very hard to do. They come up for sale every 50 years.

What did that cost you?

It’s around $170 million for everything, including the FBO and all the real estate around it.

So, you own all the real estate. That’s got some residual value. But this gives you a massive opportunity, because Los Angeles is known for its traffic. That would be a place where you could have a home base for these.

Exactly. It also happens to be home to a lot of the Elon companies, and it borders SpaceX. The original Boring Company tunnel was there. The Tesla Design Studio is there. You have a lot of good connectivity around people trying to change the future of transportation.

Oh, yes. I’m sorry. I was thinking Van Nuys in the south.

Yeah.

Funny story. Elon was like, “Hey, I’m thinking about renting a place for my rocket ship company before it had a name. Want to come down and see it?” I went down to see it. It was at Hawthorne. He had just gotten the Falcon, and I said, “Can the Falcon land here?” He said, “Yeah.” I said, “Get this office space immediately.”

So, you own the SpaceX office, or you’re the—

That’s on the other side of the fence.

Got it.

And so we have the airport side.

They have some hangars.

What a perfect location, because if memory serves me, it's just south of LAX.

Yeah. About 2 miles from LAX, about 2 miles from SoFi Stadium. So the goal is to make that a hub of Los Angeles. Imagine a Grand Central-type hub: you go down to San Diego, Laguna, just so many places to the south, Manhattan Beach.

And let's push our imagination here. The Boring Company is also based there, too. Maybe we can convince Elon to start digging some holes around LA, and all of a sudden you could really transform LA. What do you think the other cities will be in America, and how are they picking them?

Well, the FAA and DOT are going to be the ones that choose. The companies have submitted, in partnership with the cities, these different bids. And I think it's going to be a mix of urban areas and rural areas. I think it's probably going to be a lot of red states. You'll see. My guess is you'll see something in Texas. You'll probably see something in Florida.

Red states because Trump's a Republican, or because they're easier when it comes to regulations?

I think both. It's also just easier to operate.

Got it. So if you want to land a helicopter in California, there are a lot of rules. You want to land one in Texas, you put it on the grass wherever you want.

You know, it's interesting. When I got my ranch in Texas, they were like, “This is a perfect place to put a helipad.” And I was like, “Hell yes.” They said, “You can really put a pad down.” And they're like, “Yeah, it's your ranch. You can do whatever you want.”

Now, in LA, it used to be that people would land helicopters in their backyards. I don't know when that stopped, but what are the rules in LA now?

Each municipality has different rules around the different permitting you need, the amount of time it takes to get done, how you certify the helipad, and whether you need to certify the helipad.

So you basically can't do it.

Yeah. It makes it very hard. That being said, this is a safer platform than what already exists: helicopters. So I think there'll be a lot of loosening of that because it's just increasing safety for something they already do.

Now, New York City, that's the big one. I think Joby made an announcement that they're going to be at one of the ports. Yeah, you're contemporaries. I don't know if I would describe you guys as competitors now because, like the early days of, say, EVs or self-driving cars, if this works, man, we're going to need 10 Archers and Jobys to do it. But they're going to be out of Manhattan soon, yeah.

Yeah. We both will operate out of the city. I mean, it's a wonderful place. It's already the biggest helicopter market in the US. So there are 3 big heliports that exist already: the West Side, East Side, and Downtown Wall Street heliport. So it's already naturally configured. It is a very complex, congested airspace.

And so I also think the way that this will come to market will be in relatively low volumes. We'll gain the trust of the public, and they'll allow us to keep scaling this stuff up. So you can start with maybe tens of aircraft, and then maybe over 5 or 10 years, you scale it up to hundreds of aircraft.

Robotaxi and Waymo story: you start slow, build trust. Are you guys on the clock? You've been working on this for a decade. Yeah.

Yeah. The tech actually goes back to NASA 40 years ago: how to use multiple electric engines to fly airplanes. And then, really, thanks to Larry Page and the early work at Zee Aero, they really helped bring a lot of that technology.

Yeah. Larry bet on 2 or 3, right?

Yeah. So he's been super involved. He was a huge part of the industry.

And then if you go back, really, Uber did a great job when they had the Uber Elevate platform, which really put it into the mainstream. That was around 2016.

With Blade, right?

Yeah. They really were almost a research project.

Right, I remember when Travis did that.

Yeah. And then in 2018, Morgan Stanley had a 100-page initiation report saying it's the next $9 trillion market. That opened up the capital markets to everybody, and that allowed us to do this.

You guys SPAC'd right after or during COVID?

2021. Yeah.

Okay. So right after COVID started to wane. Was that the right decision? I know these SPACs have been up and down. Being public can be a distraction, and you guys are a deep-tech company with a long runway. What's it like trying to deal with shareholders in a public company when you know you're in a multidecade rollout of a product like this?

Well, I actually give a lot of credit to the Reddit community because the retail army really helped allow Archer to raise the money it needed to raise in order to get to where it is. So we've raised around $4 billion of capital to date.

$4 billion. Wow.

Yeah. And we'll probably, over time, raise more. But the stock is super liquid because it has such a huge fan base in the retail market. It creates liquidity, which allows the institutional investors to play, which allows the whole cycle to keep going.

So as long as we keep performing, there's this pot of gold at the end of the Morgan Stanley sort of pot of gold. We should be able to keep stair-stepping up the valuation, keeping everybody happy, raising the capital, keeping it going. Deep tech is extremely capital-intensive. And this was the right avenue for us, for sure.

It's really interesting when you think about it. Who is a better base of investors for the long term: public markets, institutional investors, venture capitalists, private equity, or the lunatics on Reddit? It turns out the lunatics on Reddit—all due respect, I say that with peace and love—actually probably get it right because they know what consumers want.

Yeah.

They're also willing to look past the quarterly earnings issues and really say, “This is a technology they want.” I mean, who wouldn't want it? It's safer than helicopters. They're super quiet. They cost less. It's convenient for everyone. Nobody loses here. This is a win-win.

Where is your product at? You guys are flying runs, obviously. Where are you flying them currently?

Yeah. So 2 core places we fly. One is Northern California in the Bay Area. So we fly at Salinas Airport there, which is about 90 minutes south of the Bay Area, and then also in the UAE. So we've had a great partnership with Abu Dhabi. Mubadala has been an investor. IHC has been an investor.

Oh, IHC is an investor.

Yeah, they've been a wonderful partner.

Yeah, it's really opened up the country.

Yeah, they're thinking in 50-year cycles, I think.

How often do you fly this thing? What's the distance that it flies? And are there humans in it?

Yeah. So we fly most days, piloted flights. And so you can fly them without pilots. It's actually pretty easy to fly the planes autonomously or remotely, or automated is a better way to really describe it. But we fly them piloted, and the goal is to work toward enough flight hours, enough confidence from us and from the regulators, that we can start operating commercial service. We're getting there. It takes time to do that.

I mean, how often do you fly in it?

I haven't flown it. Just the test pilots do.

And why won't they let you? You're the CEO.

No, you could. I mean, if we're just honest about flight-test programs, there's a dangerous part until you cross every T here. And if you go back in time, most of the big aerospace companies have had crashes with early platforms.

So you really try to be careful with just the test pilots, very serious testing, until you get to the point where you're close to the certification side. Virgin Galactic, another SPAC, had a tragic incident. Aviation is dangerous in the early days.

When will you be willing to get in one, I guess, is the question we all want answered, because we're not getting in until you do it every day.

Later this year.

Later this year. Now, are you married? Do you have kids?

Married with 2 kids.

Okay. So you have the conversation with your wife, I assume.

Probably tell her after.

What's her position? When will she let you go in it?

She's a huge believer and supporter of everything I've done. I think she'd trust me. When I think it's time, I would do it. She'd be okay.

Listen, continued success with it. It's obviously going to work. It's obviously going to change the world. What do we need to know as we wrap here about safety and why these are much safer? Because the idea of even having a pilot seems a little bit performative. Is it just to make the passengers feel a little bit better? These things are obviously going to be flown by computers much better than pilots eventually—and by eventually, I mean by the end of this year.

Yeah. The challenge with autonomy is regulation and infrastructure. So even if you could do it, there are no rules in place to get that done. And even if the rules were in place, how does the system work? Because today, air traffic control is very, very manual. If you listen on ATC when you fly somewhere, they're guiding you in: turn 10 degrees to the left, drop 1,000 feet a minute.

That makes total sense.

So the infrastructure doesn't allow for autonomy—

Yet.

Yet. Even though we would both agree that autonomy would be safer this year than a pilot, would you agree with that statement?

Assuming you could work your way into the air-traffic-control system? Definitely. Humans make mistakes; computers make far fewer mistakes, and maybe no mistakes. So it certainly would. That's the dream, that's the goal.

With the advancements of LLMs, it actually allows this interesting period of time where you can now communicate with a machine in a way you couldn't really before. There's probably a middle ground where you have pilots talking to machines, machines talking to pilots, and you can start to implement different systems. That's where I think it goes first, actually.

Ah, that's fascinating. So when you're clearing with the tower, it's just that the eVTOL is talking to the tower?

Yeah. Think about it: Instead of looking at your map, looking at the weather, and understanding all the different trackers you have to follow, the machines can just do that, make the decision, and say, “Do this.” So it's actually easier for a machine to do it than a human.

Again, I would encourage you to listen to air-traffic control. I'm obsessed with it. There's an incredible channel on YouTube called Blancolirio. Have you ever heard it?

No.

So go on YouTube and Patreon and throw this guy $5 or $10 a month.

Yeah, he's a commercial pilot based out of the Lake Tahoe area.

It's amazing, and I've become a little bit obsessed with it. He breaks it all down. Almost universally, the pilot makes some series of incredibly poor judgments. I'm talking about private aviation as opposed to commercial aviation. It's just great that there's somebody like him out there. Now, how many rotors are on the thing?

Twelve.

Twelve.

Twelve. Now, is that 6 with 2 in each, or is it like 12?

It's 12, which means 12 motors. There are actually redundant motors—there are actually 24—but for simplicity, we call it 12.

So, there are 12 redundant motors—24. And then how many blades? Are they double blades, like 1 on the bottom and 1 on the top?

There are 5 blades on the ones on the forward part of the wing and 4 blades on the ones on the back.

Why? Why is there a difference?

There are lots of differences. The ones in the back are used just for the lifting portion. The ones in the front are used for both lifting and cruise, so they have different configurations.

What scenario do you have the most concern about and work on the most? Because with 12 of these—and 24—and your technology, I'm sure it's even smoother and better than a toy drone. Anybody who's flown just a toy drone knows that if you come up to it and push it or hit it, it immediately gets back.

What do you worry about? Some catastrophic electronic failure? Are the electronic systems redundant? Battery failure?

Everything is redundant. You have to—

What do you worry about?

From a safety perspective, today I would say the biggest thing is the pilots. That's probably—

Take the pilots out. Now what are we worried about?

You're always worried about different cascading failures.

Got it. So, there was an incident in the industry where one of the companies had a propeller blade dislodge and it crashed their aircraft. You don't really know that unless it actually happens.

The blade comes off, hits the other blades, and then the whole thing cascades. So you don't really know if that's going to happen until it happens. You have the math behind it that you could try to predict, but when scenarios happen, you try to protect against cascading failures.

How many of these 12 could go out and the aircraft could still safely land?

One of the beauties is that we're also building it to be able to take off and land conventionally, so you can lose a lot. We have a big, 50-foot wing. You can glide. The glide ratio is huge—you can get up to a maximum of 10 miles of glide. You can do a lot of things here.

That's the only way to really certify to the standards the FAA wants you to meet. So it gives you a pretty big—

You have those nice, wide wings you can glide in.

Yeah.

What height? What's the ceiling on these?

They can go up to 11,000 feet. There's no real reason to. It's not pressurized. Do you really want to fly in the helicopter range, 500 to 2,000 feet?

What do you take from the helicopter industry when they have accidents? What happens? It seems like those pilots are particularly really skilled, but also maybe on the margins—maybe I'm reading into this—a little bit cowboyish and eccentric.

Yeah. It depends. Every scenario is obviously pretty different. The beauty behind these aircraft is that the redundancy allows you to certify with near-zero single points of failure, or zero single points of failure, whereas a helicopter doesn't have that.

If you think about 1 big rotor, there are a lot of parts that go into making that 1 big rotor work. If 1 part fails, you have a catastrophic event. They're mechanically complex machines. With the eVTOL side, you reduce that complexity.

If you were to say it's X times safer—without us holding you to it—if you were to ask the 10 engineers working across the 10 companies in your field, not you, what do you guess those 10 engineers would say? How much safer is it currently than helicopters?

That will be a standard that the FAA makes us certify to. The question will be, what's the ultimate standard? Is it 10 to the minus 7? 10 to the minus 8? One in a billion? One in 100 million?

Yeah, but you think it's twi—what would your—

It would be an order of magnitude safer.

So, 3, 4, 5 times safer is a pretty—

Yeah.

—reasonable goal. 10 times is a reasonable goal.

Why are you here at Davos?

Um—

Are you virtue-signaling about being all-electric? What's going on?

Selling aircraft.

That's—you're here to do business.

Yeah.

Who are you selling to? Just nation-states?

Yeah. Geographically, it's obviously easier to sell stuff in the US or close to the US. Here, you get to meet different companies and countries. The Middle East and the GCC have been very active. Africa has been very active. Asia has been very active.

There are a lot of great people to go see, and you can line them all up. We'll announce deals here. We'll sign things here that will ultimately get announced, so it actually is very beneficial from a business standpoint to come here.

Most of the companies here are AI companies, so there's a lot of AI talk. We're a non-AI-predominant company. There's a lot of AI in what we do, but we don't sell AI—we sell infrastructure. It's the perfect place to meet the ministers of transportation, the heads of state, and the sovereigns. So there's lots of opportunities to bring Western tech to this region.

I think they see having eVTOLs as a point of pride for their country. Some countries see it as something they can point to and say, “Look, we got here first.”

Yeah. I could see the UAE, Saudi Arabia, Qatar, and Kuwait all feeling that way, like, “Oh, we have this.”

Absolutely. There is another element where Davos is sort of non-defense-oriented. They don't like you to talk about defense here. We do have a really strong partnership with Anduril, and we've been building new autonomous attritable aircraft. We've talked about 1 of them—the second word, autonomy.

Attritable, they call it. So—

Attritable. What does it mean?

It's not a $20 million plane. It's not expendable like a 1-time missile, a 1-way type of product. They call it the space in between. They call it attritable.

If you lost it, it wouldn't be a big deal.

Yeah. So the goal is—1 of the programs we've talked about is called Project NYX. It is an autonomous, collaborative attack-helicopter drone. Think of it as sort of an Apache-type platform. There will be a big, manned asset, and it will have a bunch of drones that fly with it.

That's a program we're working on with Anduril. We build the core aircraft, and Anduril missionizes it—sensors, munitions, those kinds of things. It would fly with a bunch of drones around it to protect it.

We are effectively—the drones aren't little, but these are huge aircraft. Think about it this way: If you have an Apache, which is a $50 million to $70 million asset, and it has a person in it, you don't want to risk it for lots of reasons. 1, there's a person. 2, it's super expensive. 3, it's very hard to replace.

If you're not willing to risk it, it's not that much of a deterrent.

And so when you do risk it, you want to be certain you can have things come back. But what if I could make an aircraft that does the same thing—the same fighting power, maybe even more fighting power—at 90% lower cost, with no pilot?

Wow.

That’s key because these pilots are worth $25 million each. I literally have a friend who was in special forces, and he told me that they put a number on each of them: how much they invested and the replacement cost of a Navy SEAL. He wasn’t in the Navy SEALs, but one of these kinds of groups. You could actually know the replacement cost of a person. They called them assets, these top elite folks.

So you have no problem working in defense. That’s awesome. Yeah.

No, I’m pro-country. I’m a patriot. I’m a super believer in what the U.S. is doing. They’ve been, as a country, extremely helpful to new industries like Archer and eVTOLs.

Maybe more so than the last administration.

A lot more so. A quick example of that was that I could not get a meeting with the former secretary of transportation.

Couldn’t get a meeting.

Wouldn’t take a meeting. And I know he’s busy.

Wait, wait. A publicly traded eVTOL company that’s meeting with the Gulf monarchies and being courted?

Zero meetings.

And you requested meetings.

Requested meetings.

How many times?

Multiple times.

This is literally the Brian Armstrong story. He wanted to meet with the SEC, and they were like, “Yeah, no, we’re good.”

Secretary Duffy comes in, and all of a sudden it’s like, monthly. It was, “We want to make sure we can reindustrialize America. This is important. Aviation leading the world is important, and modernizing air traffic is important. If you can help with all that, we want to hear from you. What can we do to help you?” That’s what resulted, by the way, in the executive order that came out.

That helped the industry because I showed them a path. We won the exclusive for the LA28 Olympics to fly air taxis around that city during the Games. We’ll sort of control the air during that. I said, “I need a path to make sure I can do that. This will be a huge opportunity.”

I think the Democratic Party has not taken this to heart yet. It’s not about being subservient to the technology industry or capitalism or corporations. It’s about winning the future. Part of winning together for American companies is at least meeting with them and hearing the vision.

How does it hurt the previous secretary of transportation to meet with the eVTOL companies? How did it hurt the SEC to meet with Brian Armstrong or any number of crypto companies? In hindsight, it seems they were in contempt of the entire technology and business industry.

Yeah. In the end, if this works, imagine the jobs we will create. Imagine the GDP contribution we will create. The FAA administrator publicly keeps saying that 11% of GDP touches aviation, and so it’s capped because of air traffic. We have to upgrade that system.

I’m red-pilling myself as we speak here. If you believe that, then you would think you want to unlock that, and it will help everybody. Literally no one loses. So you would think you’d want to do that.

It was almost like they had—

Like an axe to grind with capitalists for some crazy reason. The opportunity to embrace the industry was always there for them. Yeah. Whether it was incompetence or intentionality, the result is the same. They set back the industry years by not engaging.

Greenland: how much should we pay? What do you think? What do you think it takes? I’ll tell you—

Jump into it and make yourself an interesting guest. We should take it, right? We should just take it, or we buy it.

I saw something that said there was an offer. That was the buzz here.

That was more than the GDP of Denmark.

The offer was big. I mean, there are 55,000 people. Let’s make a deal. How much do you all want?

Well, the funny thing about my meetings here is that I come to sell aircraft. That’s what we come to do. I will get asked that question in nearly every meeting. Literally, back in Silicon Valley, no one is talking about it on an every-hour basis. I mean, yes, we’ve heard it on the news, of course. But I come here, and at every meeting there are people commenting on it.

And Trump negotiates by tweet, or by Truth Social, whatever they call that platform—by post. Say what you will, this has been on the agenda for America for a century or 2, and it’s a critical deal. If President Trump gets incredible enjoyment out of negotiating a deal and solving a long-term problem for America, I say let him cook.

Yeah.

Let him cook.

I agree.

Greenland, we’re here. Fifty-first state. Let’s do it.

All right. We’re very lucky to have Chase Lochmiller with us. He’s the CEO and co-founder of Crusoe Cloud—C-R-U-S-O-E, not Kuso, Crusoe Cloud.

You got it. You got it.

Which is, or was, the first neocloud, really. You were kind of—

We were part of an early contingent of neoclouds, between us and CoreWeave and Lambda.

How did that occur? You had anticipated buying a bunch of H100s before everybody else, and you had the inventory. How did that start? How did you get such a good lead?

We had started well before Hopper came out. Crusoe is a business that has always taken this energy-first approach to developing computing infrastructure and really changing the notion from trying to collocate compute in these network hubs to focusing on where we can access abundant energy.

We were experimenting a lot with the Ampere generation, which preceded Hopper, so we had a lot of A100s. We built out this high-performance cloud platform that was meant to enable AI innovators to do incredible work. We launched before ChatGPT came out, so sometimes you’re in the right place at the right time.

Yeah, timing was good.

Yeah, timing worked.

And your biggest customers by far are Oracle and OpenAI.

Correct. We’re a vertically integrated business, which means we’re focused not just on the compute layer, but also on energy development, data center development, as well as the application layer, where we’re running things like our managed inference service.

As this boom really took off, the scarcity of data centers became very apparent, and our ability to build those large-scale AI data centers very quickly became a very in-demand skill set.

So why West Texas? Is it because the politicians there are pretty permissive in terms of giving permits, and you can build quickly, like Elon experienced with his big factory in Austin? Or is it because of energy?

It was really an energy-driven decision. In Abilene, Texas, it’s an area where a lot of actual renewables had been built out on the back of production tax credits. There was an abundant amount of wind and solar that had issues with transmission, so power prices were frequently negative.

There were renewable energy producers that were having to curtail. They could be producing power but were shutting down because there was no marginal demand.

So the grid didn’t need their energy.

Exactly.

That’s wild to think about. It’s also wild to think about that Texas has the largest solar base of any state in the union. Yeah.

Absolutely. We went there and said, “Man, you guys have too much energy. Hey, buddy, I have demand for you.”

We have a 1.2-gigawatt substation there. We’ve also built a 350-megawatt gas plant on-site to energize one of the largest clusters of GPUs in the world.

Was there gas under the actual data center, or are they shipping it?

No. We had access to a pipeline that feeds into 10 gas turbines that are on-site there.

Amazing. The gas turbines were a blocker for a while, too, huh?

They still are. Gas turbines are a massive supply-chain constraint as people look to energize compute infrastructure.

I think we were really one of the first groups focused on natural-gas-powered data centers, doing things behind the meter and off-grid. A lot of people have followed this pathway, which has created a lot of supply-chain challenges with some of the major producers: GE Vernova, Caterpillar’s company Solar Turbines, which we work with very closely, Siemens, and Mitsubishi.

We recently did something with an incredible company, Boom Supersonic.

Yes, Boom Supersonic was making the Concorde replacement.

Correct. They had reengineered this turbine for supersonic jet travel. Colleen, my co-founder, and I were talking to Blake and said, “Hey, could we use that to generate power instead of transporting people over the ocean at supersonic speeds?”

Now we’re their first large purchase order, for $1.2 billion of gas turbines to power critical AI infrastructure.

It’s wild to think about gas turbines and jet engines. They’re not super dissimilar.

No, I think this is the full playbook of GE historically, right? The reason GE Vernova exists is its leadership in terms of air transportation as well as power generation.

Yeah. I wonder if the really interesting thing about that opportunistic move for Blake to make with Boom Supersonic is that it will now allow him to fund his Concorde replacement.

Absolutely. Power is a great business, and we think he can make a lot of money as the scaling up of AI infrastructure occurs. Hopefully, he’s able to build those incredible supersonic jets that get us to Tokyo much faster.

And the Stargate project—

Yep.

—is a $300 billion project. What’s the number? There was a big announcement at the White House, and there’s been a little bit of a wink-wink that these numbers were kind of estimates. What’s the realistic footprint of this?

I think Stargate has been a term that’s been used to describe a lot of different things at this point. Initially, our campus was called Stargate, and then Stargate was a company for a while. I think OpenAI sort of described all of their spend on compute as being broadly labeled Stargate.

I think the number is $500 billion, and this incorporates chips, data centers, and energy to ultimately power this intelligent infrastructure that’s running and scaling both ChatGPT and all their other core services.

One of the blockers, in addition to the turbines, has been electricians.

And construction workers.

Yes. My understanding is you’re paying 2 to 3 times what they were getting paid before the data center.

No comment on exactly what we’re paying the electricians, but they’re very well compensated.

Am I in the right zone that their salaries in the industry have doubled or tripled?

Look, I think it’s an incredibly exciting career path for anybody looking to do work with their hands and get well—

Hundreds of thousands of dollars a year.

Yes. So—

I mean, let’s think about that for a second. No college degree.

Yeah.

You just have to be an apprentice as an electrician, and you can make hundreds of thousands of dollars a year. And you need how many—thousands? Hundreds?

Correct. I’ll get into it in a second. I’ll give you an example. In Abilene today, we have 8,000 people on-site every day. They’re working day and night to bring this facility online as fast as possible so we can energize this intelligence factory.

And that’s 3 hours west of Dallas.

Correct. So, it’s in West Texas. Abilene is a town of 120,000 people. When you’re hiring 8,000 people to work in that town, you can’t source everybody locally, so we’ve actually had to bring in a lot of labor from all states.

And housing, I would assume.

Yeah, housing. The market’s actually fairly efficient. A lot of people who won’t be long-term permanent workers there bring mobile homes and collect a stipend. It’s a very efficient process in that regard.

Long term, we’re going to have about 2,000 permanent workers at the campus. So, there’s both a short-term job spike and boom in terms of demand for the construction and installation process, but there are also great, long-term permanent jobs to operate the facility. We have a power plant there, and there are a lot of mechanics who are going to be operating this facility.

This is just one campus. We have another campus in Armstrong County, Texas, that has close to 3,000 people on-site every day. We have another project outside Cheyenne, Wyoming, that we’ve announced we’re planning to build as a 10-gigawatt campus.

Why Wyoming?

Wyoming is a state that’s very rich in natural resources and energy. There have been really favorable public-private partnerships in terms of making this project happen. There’s a tremendous amount of gas there to support what we’re doing.

One of the very cool things that we’re focused on is trying to bend the arc of energy production toward sustainable resources. Even with the gas, Wyoming is a place where you actually have primacy on what are called Class VI wells. They’re post-combustion carbon capture and sequestration disposal wells, where you can take the carbon that comes out of the combustion process of a natural gas turbine, pump it underground, and permanently sequester it.

Ah.

There’s actually an incentive there called 45Q, where you get paid by the government to do this. It doesn’t quite cover the cost of doing it, but for customers that care about the impact and sustainability of the campus—

They’re not putting the carbon into the atmosphere.

Exactly.

Putting it underground—

They can produce power from gas in a carbon-free way, which is a really exciting value proposition.

How are batteries? I think we saw Elon—he’s got his own lithium refinery and is building his own batteries, and they’re one of the larger battery manufacturers. How do batteries play into all this currently?

There are a bunch of different ways batteries play into it. You need batteries in these low-voltage UPS systems that help operate the electrical system of the plants—

To normalize and baseload, or—

Yes. Actually, one of the issues that we’re seeing is, when you think about these large-scale AI factories, when you’re deploying giant clusters of GPUs, the entire data center acts as a single computer. It’s running one single workload that’s training some breakthrough foundational model. What that results in is massive load fluctuations in the actual power draw, because the chips are basically running compute cycles and then publishing the data over the network to all the other GPUs so they can be in sync with one another.

This causes a massive fluctuation in overall power draw. Utilities hate that. It’s terrible for the turbines. So, you really need a way of normalizing that power draw. Exactly. We’ve actually been able to solve that with a 1-hour battery. That’s a medium-voltage battery energy storage system. That’s something that we’ve deployed.

There are a lot of other use cases for batteries. I’ll give you another example. We’re actually working with Elon’s former co-founder and partner at Tesla, J.B. Straubel.

Oh, he’s doing the recycling of them.

Correct. He has a business called Redwood Materials that’s focused on recycling batteries. He has this massive supply of end-of-life EV batteries, batteries coming out of digital cameras—all these batteries that are sort of at the end of their life. As a demonstration to show people what we could do in terms of cost competition as well as deployment, we deployed a fully off-grid solar-plus-battery energy storage system that powers an AI data center 24 hours around the clock with a power price lower than the power price in Northern Virginia.

We’re able to do that because we’re basically taking these batteries that were in an electric vehicle where maybe the range was 300 miles and now it’s 250 miles. The owner trades it in and says, “Okay, this car is done,” right? But there’s still a lot of juice left in those batteries. There’s a lot of life left in those batteries, so we’re able to make use of them in a second life to power these AI data centers.

You’re under massive pressure to deliver these data centers, correct?

Yeah, yeah, you could say that.

Compared to where you were 2 or 3 years ago—maybe let’s say 3 or 4 years ago—you were knocking on doors saying, “Hey, do you need a data center?” Now your door is being knocked down.

I think it really speaks to the demand for compute. People are constantly having this conversation of, “Are we in an AI bubble?” I think there’s just an incredible amount of demand. Nobody has enough compute. None of the leading labs or the leading application companies can get their hands on enough compute, and that’s driving an incredible amount of urgency to deliver infrastructure.

Both the chips and the data centers. So, even this year, going into 2026, are you seeing the same amount of inbound—“Hey, we need more, we need more, we need more. What do you got?”—or was last year when people put in their big orders for the next couple of years?

No, we’re seeing things accelerate.

Who are the new customers? Obviously Elon, Sam Altman, I’m sure AWS, I’m sure. Well, I think Google does their own. Yeah.

Google, I think. I don’t want to speak for them directly.

This is All-In. If you want to be on All-In, you’ve got to speak candidly. Just tell us what you heard at the cocktail party.

Google does a mix of self-performed development as well as outsourcing.

So they do some outsourcing? Yeah.

Correct. But I think all of the leading labs are seeing a tremendous amount of demand: Anthropic, OpenAI, Google.

Elon builds his own. What was your take when you saw him build Colossus in that short time frame? Your team must have thought, “That’s just not possible,” and then he did it.

No, I don’t think we thought that at all. I was really rooting for him.

But my understanding is that he did it in half the time anybody else had ever done something like that.

It depends on how you measure these things. I think Colossus 1 was a unique case where he had this large-scale industrial building and power to the building. What he was doing was what I would call the tenant fit-out, which is basically the data hall buildout of the cooling distribution units, the RPPs, the electrical systems, and the hot-aisle containment systems. Then you roll racks of GPUs into these.

They were able to execute on that incredibly fast.

Jensen said he’d never seen anything like it. He seemed to think it was a unique thing that occurred in your industry.

Yeah. Elon's the GOAT of modern industrialists, so hat tip to him.

You’re running a company that’s not competing with him but is building the same things. You have to look at that and study it a bit. How did he get it done so fast? What do you think enabled him to do it so quickly?

I think Elon does an incredible job of breaking down a large industrial process into a lot of subprocesses, understanding the constraints, and taking a first-principles approach. How do I build things as quickly as possible? How do I parallelize things as quickly as possible?

Got it. Has that informed some of your thinking?

Absolutely. He’s been an inspiration, from building the Gigafactory to everything he’s done at SpaceX and Starbase. It’s all incredibly inspiring, and we try to channel that same sense of incredible urgency by parallelizing a lot of the work.

We self-perform a lot of procurement and engineering functions, and then work with a lot of very ambitious folks on the construction side. There are actually a bunch of people who worked on the Tesla Gigafactory in your new hometown of Austin who are working on our campus in Abilene, Texas. There’s a lot of overlap in the methodologies.

There’s been a lot of talk—and I think Brad Gerstner started this discussion on his BG2 podcast when he had Sam Altman on—about how a company with a $20 billion, or at the time $12 billion, now $20 billion run rate pays for a $500 billion buildout. In your contract with OpenAI and your relationship with them, I’m assuming this is being done in stages—not one giant $100 billion contract, but in stages.

I think it’s important to understand that, in a lot of ways, my role as CEO of Crusoe is like half risk management. The amount of capital going into this is enormous, and all that capital is not going to come from dilutive equity capital that we’re raising at the parent company. We have to raise project equity and a lot of debt.

What debt providers are focused on is, “How is this person going to pay me back?” So it is very much something that’s being evaluated in the capital markets. These large-scale, blue-chip, investment-grade businesses play a major role in catalyzing the capital formation you need to build these giant infrastructure projects.

So when you have a company like Google, a company like Microsoft, or a company sitting on $100 billion in cash, and they throw $10–20 billion to the bottom line every quarter, they could fund it from their existing businesses. But Elon and Sam have new businesses, and even Anthropic has to come up with that money somewhere.

Correct. In our case, I’ll give you an example in Abilene, Texas, because that’s the most public one. We have a long-term lease agreement with Oracle—a 15-year offtake agreement—so they’re committed to paying us for that time frame. That helped unlock a lot of the construction debt and the capital we needed to build the project.

We worked with JPMorgan, and a number of other folks were in the syndicate: Bank of America, Apollo, SMBC, and a bunch of different institutions.

They must be very excited about this opportunity.

Yeah, I think everybody’s really excited about the opportunity to build the infrastructure that’s going to power the economy in the future.

This is collateralized by a data center, which has inherent value. If a customer of yours were unable to meet its commitment, there would be somebody else to take that capacity.

Yeah. I think that’s the biggest argument I make when people ask about the AI bubble. Imagine a scenario where OpenAI went out of business, which I don’t think is going to happen.

There are people saying that could happen.

Sure. Okay.

I’m not saying that. You didn’t say it.

Imagine they did go out of business. The reason they’d go out of business is because some other model company—whether it’s Anthropic, Gemini, or Grok—really blew them out of the water.

They’re kind of in fourth place right now, according to some of the benchmarks.

Put all that aside. They have incredible adoption and an incredible platform. But if that were to happen, the group that massively surpassed them would have so much demand for that product that they’d be jumping for joy to step into the seat and take over that compute capacity.

If you believe that AI is going to be an important aspect of the operating system of the economy in the future, this infrastructure is going to be very useful and valuable to whoever the winners are in that future state.

People don’t know this, but you’ve also made a bet on startups. You’ve been incredibly generous in giving credits to startups. Maybe you could talk a little bit about how you made that decision to play the long game, because I would think there are probably some people on your board or in your organization saying, “We’ve got some big fish here. We’ve got whales. What are we doing with the minnows?” You’re specifically targeting up-and-coming startups to be their provider.

I really think about Crusoe as a vertically integrated business. We offer three core things. We can build the data centers and rent them to customers. We’re really focused on a small subset—maybe 5 customers—of very large tech companies, and that’s really their key bottleneck. They’re not renting out capacity from clouds because Meta doesn’t need to know how to run a GPU cluster. Of course they know how to run big clusters of GPUs. They need data centers. We’re trying to unblock that critical pain point for them.

On our AI cloud platform and our infrastructure-as-a-service platform, we offer managed clusters of high-performance GPUs. We’re a very large partner with NVIDIA, and that’s really critical infrastructure for the startups you’re talking about. This is the development of new AI-native applications, coding assistant tools, managed inference solutions, video generation, image generation, chatbots, and all sorts of AI applications that are proliferating in the world.

We offer managed GPU clusters there, and we also offer serverless AI services like our managed inference product, where we charge people on a dollar-per-token basis. We can charge a dollar-per-kilowatt rent on the data center, a dollar per GPU hour on the GPU, or a dollar per token on the managed services.

Yeah. And that’s thousands of customers, potentially tens of thousands, and that’s where you’re going up against Azure, AWS, and Google Cloud.

Correct.

Those are some pretty significant competitors. How do you compete with those kinds of companies?

When you look at those really large hyperscalers, they’re incredible platforms, and they’ve been able to accomplish so much. But they really are the outsourced IT solution that’s meant to be everything to everyone, which means their lowest common denominator isn’t AI; it’s everything.

Servers, storage—it’s everything.

We are relentlessly focused just on the AI use case and the AI application to deliver the most reliable, highest-performance computing infrastructure directly for the AI use case and application. That’s all we care about. Which means all of the optimizations we’re making on the high-performance networking system, compute side, as well as the storage and how people access their data, are entirely in service of AI use cases.

There’s going to be some technological advancement in the coming years that really ramps up what we’re able to do in building large language models and doing inference. What do you think that will be? A lot of people have talked about optics and the transfer and transport layer. Some people are talking about the raw horsepower; other people are talking about cooling.

When you talk to some of the hardware providers, where do you think we’ll see the next step function in the next, let’s call it, 3-to-5-year window? You must be thinking about skating to where the puck’s going, because you did that with your current offering, I should say.

Yeah. So, look, I think what’s so fascinating about this space and what gets me so excited is that it’s actually the culmination of so many different engineering disciplines. It’s the pinnacle of human achievement across so many different engineering disciplines, from the cooling systems and the chemical engineering, and the electrical systems and the physical electrical engineering that needs to go into these chips, to the silicon systems and the chips, and the networking and the software engineering. It’s really a full-stack solution that ultimately produces intelligence.

Now, getting to specific challenges that we see and specific trends, we’re seeing everything go to higher-density configurations: more power in the rack. A traditional data center maybe 5 years ago was 15 kW; 20 years ago was probably 4 kW in a rack. The current generation of Blackwell chips is 130 kW in the rack. The next generation, Vera Rubin, is going to be 250 kW, and Vera Rubin Ultra is expected to be 600 kW.

We’re ultimately going to get to 1-megawatt racks. 1 megawatt, to give a perspective, is roughly the amount of power of 1,000 homes.

Yeah. It’s like a small town in a data center, right?

So a rack will be a town. Exactly. A couple dozen of those will be a city.

Yeah. It’ll be a New York City in a data center.

Crusoe has over 45 gigawatts in our pipeline. I know you’re from New York; that’s about 8 to 10 New York Cities’ worth of power, depending on how you measure it. So it’s an incredible amount of energy infrastructure that’s going to need to come online to help energize this layer of compute. And, like I said, there are challenges in every domain, from cooling to networking to—

Yeah, that density arrives—this 10-to-1 density you’re talking about—and the heat also arrives.

Correct. Yeah. So, a lot of exciting stuff is happening.

And what do you think of small modular nuclear getting close to hydro? Obviously, gas is the layup, and gas is everywhere. We’re leaders in that. Solar feels like a layup, and battery—that combo to add—but hydro, I don’t know if there’s much left.

We’re actually doing a lot with hydro in the Nordics, in Norway and Iceland. In Iceland, there’s abundant geothermal. It’s this sort of geological phenomenon, and there’s ultra-low-cost geothermal energy and also a lot of hydro there.

So that’s a one-two punch.

You get the one-two punch there.

I don’t know if you guys heard, but America just acquired Iceland. That was the rim shot off of Greenland. That wasn’t a mistake. We’re taking both—I mean, or you’re giving us both. It’s like Greenland. We don’t ask for much, okay? It’s a very simple request.

We have data centers. You may have heard of big data. Big data centers, in a few words. Listen.

But to your comment on small modular nuclear reactors, we are very bullish on SMRs. We have signed 4 contracts. We’re actually hoping to energize the first AI factory powered by an SMR in 2027.

2027. That’s next year.

I know. And one of the ways we’ve been able to do this is that it’s actually going to be at the Idaho National Laboratory, where you’re outside of the regulatory domain of the NRC. It’s considered experimental technology.

Oh, yes. Who’s the partner on that?

I don’t think we’ve made it public.

No, it’s okay. They’re going to make some news here. Journalists are here. No, I know because I was told this exact same insight by the company that’s doing it, but I won’t say the name of it.

Yeah.

Yes.

I think we can announce it. We should announce it.

Okay.

Let’s announce it. The partner is Alo Energy. They’re an incredible partner.

Isn’t it amazing that, for solar, it was something we hadn’t been able to do since the ’70s, and now it’s like, “Yeah, we’re going to do solar.” It’s absolutely necessary, so we’re going to do it. What do you attribute that to?

I think there’s this human ingenuity, the passage of time, and the relentless pursuit of efficiencies. I just think it’s really incredible if you look at the cost curve of solar, how much it’s come down over time. I think you’re going to see a similar thing play out in SMRs and next-generation geothermal.

We’re really excited about innovations like what Fervo has made in terms of being able to produce geothermal at scale at a very competitive price point, leveraging a lot of the technology from fracking and oil and gas.

And we’re not going to have this impact on consumers’ electrical bills?

Yeah, I think that’s such an interesting story. When we look at this problem, we say, look, a lot of the power on the grid is very saturated. A lot of the data center capacity is saturated. So it just makes sense for the technology industry that wants to bring online all of this new infrastructure to also bring online the power that goes with it.

The incredible opportunity, from my perspective, is that when we bring on new power generation to support an AI data center, we’re sizing it to the peak demand of the data center, which means we’re only using peak demand 0.1% of the time. So you might have some excess. 99.9% of the time, we have excess power that can support the local community, create an abundance of energy, and drive down the overall cost for ratepayers in the local communities.

People will have lower-cost power. We’re going to have advanced, intelligent infrastructure that’s driving massive efficiency gains in the economy. It’s going to be an incredible future we’re building toward.

This is something that I think the technology industry could be self-aware enough to understand: if we’re making this incredible new business, it’s a great way to share it with other Americans. Maybe your energy bill will get lower or eventually free.

Absolutely. You’re already seeing this trend unfold. We’ve taken this energy-first approach. You saw Google recently make the acquisition of Intersect Power. Sheldon and his team there are an incredible group of energy developers, and they’re doing that because they know they need to build the energy infrastructure that’s going to support their compute needs in the future.

Yeah. Well, I really appreciate you taking the time, Chase, and continued success. What an amazing story, man. You really got there early, and one of 2 things can happen when you get there early: you can just fail fabulously, or you can absolutely crush it, and it’s been the latter for you. There might have been some moments where you stared at the ceiling at night as an entrepreneur and said, “Are we too early?”

Yeah, it’s mostly been up and to the right. No, I’m just kidding. There have been tons of complex problems and challenges and moments of doubt throughout the company’s lifetime, but I wouldn’t have wanted to do anything else with my life.

It’s nice to move from “Will the customers arrive?” to—

“Okay, we’ve got too many customers. We really need to deliver.”

Yeah.

It’s actually a whole organizational mindset of “Will the customers ever show up?” and this fear, to “Oh my God, I hope my customers are happy and delighted.”

Totally. But scaling people, scaling culture, scaling technology—it has its own set of challenges and problems. The culture part is important.

It’s difficult. Yeah. It’s difficult to go from a small startup of tens of people to more than 1,000 people.

And adding what—what are you adding, 1,000 a year?

We’re going to add 2,400 people this year.

Full-time employees, and then tens of thousands of contractors.

Yeah, I’ve seen that movie before.

And no, I watched it with Uber and Robinhood as they were adding one person a day, then it was 5 new people, and then now you've just got training and recruitment and just trying to keep that culture

Tight and make sure you hire the right people. Well, listen, you have to go, and we're way over time. I appreciate you taking the time.

Man, I appreciate it.

All right, give it up for Chase.

Thanks, Jess. I'm going all in.