2025年预测：科技、商业、媒体与政治！

It’s 2025, and we’re doing our Bestie Awards. It’s going to be amazing. With us again, for the second time, is a truly amazing Bestie, Gavin Baker from Atreides Management. Gavin, welcome back to the program. Let us know briefly: what does Atreides do?

Thanks for having me here, Jason, Chamath, and Dave. Atreides is a crossover firm. We invest publicly and privately in consumer and technology companies, and we go from Series A to megacap.

Got it. So you’re a capital allocator. You place bets on technology and on the most important new companies in the world. We’ll get into that. This is our prediction show, and we did our Bestie Awards. How big is your firm, Gavin?

Roughly $4 billion.

All right, so that’s about a half an inch bigger than Chamath’s.

He’s really worried now. Not that he’s a size queen or anything.

Let your winners ride, Rainman David. Instead, we open-sourced it to the fans, and they’ve just gone crazy with it.

We’re going to do a really great prediction show today, and we’re going to do some super predictions this year. Each Bestie gets to make a super prediction or two during the show, and we’re going to take those super predictions, Chamath, and put them on Polymarket.

Yes, that’s right. If you don’t know Polymarket, it’s a prediction market where people can place a wager, a bet, or an investment on one side or the other.

We’ve obviously talked to Shayne at Polymarket for some time, and we set up a deal with them where we could put our own markets up on the site. We’re going to talk about a couple of our predictions this year, and then Polymarket will post them on its site. People can trade them, and we’re really excited about it because it will allow us to track how we’re doing and give us the opportunity to create markets on an ongoing basis.

There will be an All-In section on Polymarket where you can go in and see the All-In markets. It’s going to be really cool.

Fantastic. We’re all wrapping up our ski trip here. Nick, I understand there was some footage—some found footage—from the Besties skiing. Is that true?

Here I come down the mountain. Gavin, I want you to rate everybody’s skiing. Here I come, looking great. There comes Chamath, looking good. Then here comes somebody else. I think the technical term Joe Lonsdale uses for that third person is “beep beep.”

I made a very strategic decision to stop snowboarding so I could learn to ski with my kids. It is so hard to learn how to do something in your 40s.

Not if you’ve never done it before.

I do appreciate what you and your brother did for me last season. I learned a couple of things, so this has been my third season with the Blackcomb. It was great.

I’m absolutely amazed at the progress you’re making. Friedberg, I’m amazed you made it onto the mountain. You were out there a number of days. You’re getting it done.

I needed new boots, and I tore my MCL last season, so I got really tweaked when I went out there. But I’m good. It was great. I had a lot of fun.

You skied for 4 days?

Yeah, 4 days.

I finished my 16th day yesterday, but I took today off just to get prepared here.

I got 16, too. I do the executive program. I go off for an hour and a half to 3 hours, zip, zip, zip, 10 runs, and I’m done. I only ski in the morning.

That’s the way to do it. At 12:15, have lunch, relax, and get a little work done. You need a healthy mix of après-ski and skiing.

Gavin, how would you compare yourself to what you just witnessed in that video? Where would you rank yourself?

I’d make 2 observations. First, I’m in the bottom percentile in terms of natural athleticism, but I do have many thousands of hours skiing.

Great. Then we’ll be going out with you next year, new Bestie.

We’ve got to get to it. We’re going to start off with politics. We have to keep this moving because there are so many predictions.

Last year, we got it kind of wrong. Friedberg, you said independent third party in the United States, and we did see some of that with the breakout of Robert Kennedy. Chamath, you said independent centrist, and we didn’t get there with that one. I said dark-horse presidential candidates, so maybe I get a quarter-point credit there.

Hold on. What do you mean we didn’t get there? Independent centrists won the election.

That’s what you believe?

Independent centrists won the election. I think independent centrists swung the election for Biden in 2020, and then I think they swung the election for Trump in 2024.

Friedberg, is that statistically correct? Do you think independent centrists decided the election?

I think it’s right. There were a lot of people—not just centrists, but people who had been lifelong Democratic voters—who voted for Trump. I do think Trump won the centrists.

Trump won everything, essentially. All right, Chamath, what’s your prediction for the biggest political winner of 2025?

My biggest political winner for 2025 is fiscal conservatives. I think we are going to test a very important concept in 2025, and I hope it works: austerity.

The reason austerity has to work is that the only thing left after austerity is to cut entitlements. I think that, in doing this, we’re going to figure out how much waste, fraud, and abuse exists in the United States federal government. I think that’s going to spill over to a lot of state elections, and I think the fiscal conservatives who have been clamoring for a more restrained approach to spending will have their day in 2025.

Fiscal conservatives. Friedberg, who’s your prediction for the biggest political winner of 2025?

I also took a class-based approach. I chose young candidates. Trump’s cabinet picks have an average age of 40 to 45 years old, compared with the Biden cabinet, where the average age was a little over 59—almost 60.

I do think this marks the beginning of a new trend in the age range of political candidates shifting younger. I think this is something we should expect as candidates start to emerge for the midterms. By the end of 2025, we’ll start to see younger, new names pop up that deliver resonant messages and aren’t part of the old guard of the aging political class.

Excellent. Young candidates. Gavin, what do you have for the biggest political winner of 2025?

I would say Trump and centrism. For my choice for 2025’s biggest political winner, I went with something similar to you, Friedberg: Gen X and the elder millennials.

If you look at the notable Gen X appointments, you have Elon Musk, you have Sacks, obviously, and Marco Rubio. It just goes down the list. Then, if you look at the elder millennials—J. D. Vance, Vivek Ramaswamy—there are a lot of young people.

This is going to be fantastic because they’re going to start thinking not just about themselves, as the boomers are doing with Social Security, taxes, real estate, and all the different issues they tend to pick for themselves. They’re going to start thinking about their own kids and themselves. A sea change is underway.

We have 4 predictions for the political winner. Let’s go with political loser. We’ll start with you, Gavin, and do this in a round robin. Who’ll be the biggest political loser of 2025?

I think Putin. I think Putin is going to lose bigly.

If you are Xi Jinping, Russia is a client state of China at this point. What’s happening is a disaster for you because Europe is starting to rearm, and that will only accelerate this year. That will allow America to take resources out of Europe and put them in Japan, South Korea, and all over the Pacific.

That makes it a lot harder for you to do what you most want to do, which is reunify China and Taiwan—or invade Taiwan, let’s call it what it is. I think Xi is going to begin decoupling from Putin.

If you’re Trump, you want to show that you’re independent, that you’re not beholden to Putin in any way. I think Trump is going to be a lot tougher on Putin than people think, and I think he’s going to get a deal that’s very bad for Russia and Ukraine. You’ve lost half a million people, and for what? For what exactly?

It’s more than 1,000 days into this. Nobody from the West—nobody from NATO or America—has lost their life. We’ve just given them weapons. It’s a humiliating defeat so far for Putin.

I’m going to predict the pro-war neocons, who are going to go head-to-head with J. D. Vance, Elon, and others in the administration. I think they’re going to lose, and I think there’s going to be a big crack in the establishment of this neocon movement that’s been very pro-conflict around the world.

We’ve heard it in the speeches and commentary from J. D. and others. I think this is the year it’s all going to come to a head, and they’re going to end up on the losing side.

Can I take the other side just a little bit?

I think that’s why you’re here.

I think it’s right in reality, but Trump said something very interesting about John Bolton. He said the guy was absolutely crazy, but it was awesome having him in the room when you negotiated deals because people looked at this guy—angry, red in the face, so excited to hit the nuclear button, so excited to go to war—and you ended up with much better deals.

I think you’re going to see a lot more bellicosity from the Trump administration than anyone expects. That’s just to get a good deal between Russia and Ukraine, and then to get China to decouple from Russia.

In reality, I think you’re right, David, but I think there will be a lot of rhetoric at odds with what you’re saying before you end up being right. We already see it with Canada, NATO, and Taiwan. There’s a lot of “we’re going to do this” or “we’re going to do that.” The tariffs, obviously—there’s aggressive posturing leading up to negotiations that are hopefully going to get the United States good deals.

Chamath, what’s your prediction for the biggest political loser of 2025?

The biggest political loser of 2025 is going to be progressivism.

In November of last year, right after the election, I flew to London, went up to Oxford, and spoke at the Oxford Union. My speech was a full-throated defense of MAGA, but it was mostly an explanation of MAGA. It was the antidote to progressive instincts that had been riddling the Western G8 countries and were starting to come undone.

When you look at what’s about to happen in 2025, Justin Trudeau is going to lose massively to Pierre Poilievre in Canada. In Germany, the AfD looks like it will win. In France, if there’s a deadlock and it goes into an election, more than likely Marine Le Pen is going to win.

Then, in the United Kingdom, you see this unfolding child-rape scandal, where allegedly hundreds of thousands of young girls over the course of more than 20 years were being raped by organizations of Pakistani Muslim men who were not prosecuted for fear of stoking Islamophobia, allegedly under the current prime minister, Keir Starmer.

If all of that comes to pass in the U.K., I think you’re going to see the Labour government fall, and I think you’re going to see Nigel Farage win.

What will all of these countries look like by the end of 2025? It will be very much a repudiation of this class-based identity politics. I think that has enormous ripple effects throughout the world. I think the biggest political loser for 2025 stands to be progressivism—what we labeled “progressivism.”

I took something very similar. I said the racist vocal minority of each of these parties. There’s a little bit of this on either side. You have DEI on one side, and then you have outright racism on the other side. It’s probably 5% of MAGA and 5% of the left.

To recap last year’s political predictions for 2024, I said Netanyahu, Friedberg said Ukraine, and Chamath said the Koch family.

The GOP nailed that one.

I nailed that one.

Friedberg, any thoughts on last year’s prediction? Anything on the Republican side more than the Koch family?

I think essentially 2024 was the end of the Republican Party as we knew it. I think what stands in its place is what I would call the MAGA reflection of a coalition of people that will be housed under the label of republicanism.

These folks aren’t necessarily Republicans. They’re believers in the MAGA philosophy who are using the Republican vessel to run their candidates and get elected.

In that overturning of the status quo, you had one family at the center of the political machinery for the last few decades. It decided candidates, agenda, and policy. That was the Koch family. They spent an enormous amount of money to get what they stood for—frankly, literally the day after Citizens United happened at the Supreme Court.

I think, in that lens, those billions of dollars of investment have essentially gone to zero because I don’t think it means much of anything anymore. If you look at the new class of donors who will decide Republican policy, it’s going to be the Miriam Adelsons of the world and the Elon Musks of the world. That’s very different from how I think the Kochs used to decide things.

Do you think there’s too much money in politics now, Gavin?

It’s a good question. I might feel very differently if I didn’t agree so profoundly with the largest donor in this political cycle.

The reality is that, as long as the money raised on each side is roughly equivalent, I don’t think it really matters. That would be my take.

You just want some equivalence. All right, let’s do our 2025 predictions for the biggest business winner. Friedberg, why don’t you start us off?

I feel like we’re at a really interesting inflection point that is going to make 2025 the year of autonomous hardware, or robotics.

If 2024 was the year of compute buildout and the rollout of AI systems in software, I think 2025 will be the year of the robot. There’s a company out of China called Unitree that we placed an order with today. Gavin, have you looked at this company?

Yeah, it’s pretty wild. It’s an incredible business and an incredible product.

Their Go2 robot is $1,600 and has an API. It has lidar and intelligent guidance systems on it. This is the robot system that was used in some of those videos we looked at earlier this year, where machine guns were mounted to the back and it was essentially a new kind of field soldier.

You can use it in scientific applications. We’re looking at using them on our test farms, where it can wander the farm, take images, and report data back to us. It’s such a low cost—less than $3,000—that you can do some incredible things with it.

This business raised a couple hundred million dollars last month from mostly Chinese investors. I think there are going to be other similar businesses. It takes a long time for things to work, and then all of a sudden they happen faster than you could have ever imagined.

I think this is going to be the year when we all look at humanoid robots and autonomous systems and say, “Oh, my God, I can’t believe this is here.”

Such a strong choice for biggest winner. The quote you’re looking for is, “How did you go bankrupt?” Slowly, and then all at once. That’s how these technology changes work.

I would love to have one of these for the ranch to run around and do perimeter security. Nick, pull up the Go2 with the wheels on it, managing terrain. Everyone thought this was a fake video, that it was CGI or AI-generated, but this thing is incredible.

Gavin, is this real?

Absolutely real. I want to get the G1, which is their humanoid robot.

Pull up the humanoid. It’s really cool.

It’s a little more expensive.

$16,000. You can basically command it to do things for you in your house, factory, or workplace.

I wonder if this thing could deal poker.

I knew you would do it.

Maybe we could get a poker dealer.

Do you guys remember that robot that was onstage for our All-In Holiday Spectacular?

My brother made an observation. As soon as that robot came out, everyone just wanted to abuse it. Humans have this interesting nature where the robot emerges and all humans want to do is hurt the robot. It’s like, “Here’s something I can dominate. Here’s something I can tell what to do.”

We’re so threatened by it that we want to make sure we’re at the top of the species. It doesn’t feel like it’s going to be that way for much longer. These things will kick our ass pretty soon anyway.

I profoundly agree that this is the year of the robots, including full self-driving. FSD works today, and it’s going to cross into mainstream adoption.

If I’m taking an Uber late at night, I really prefer to have a Tesla. Sometimes you get an Uber driver who’s tired, and I feel a lot safer if they have FSD running. Using it for yourself is amazing, but I think it’s going to continue compounding at an accelerating rate.

Broadly, for a while it’s going to be big businesses that are the winners—big businesses that use AI thoughtfully. With what o3 showed us from OpenAI, the combination of reasoning and test-time compute is profound.

If you’re a big business and can pay $1 million to let an AI think for 6 weeks about the most important question for your business, that’s going to be a profound advantage relative to small businesses that can’t afford it.

So you’re also long Nvidia and chipmakers like Groq because we’re finding new uses for all that compute.

Absolutely.

Very good. Chamath, do you have a prediction for your biggest business winner of 2025?

I think the biggest business winners of 2025 are going to be dollar-denominated stablecoins.

Two critical things happened in 2024. The first is that stablecoins became uncoupled from crypto volatility and started being used for wholesale, useful functions in running businesses. Independent of crypto volatility, stablecoin usage has risen up and to the right. That’s an incredibly important decoupling.

The second point is that stablecoin usage at the end of the second quarter of 2024 was about 1.1 billion transactions, summing to $8.5 trillion in transaction volume. Compared with Visa over the same period, that was more than double Visa’s transaction volume.

I think we now have something that has fundamentally crossed a point of no return. Similar to how I thought Bitcoin would be the big trend in 2024, I think the big trend in 2025 is stablecoin usage.

I think we’re finally going to attack the duopoly of Visa and Mastercard. You’re going to see an enormous number of use cases using stablecoin rails. When Donald Trump becomes president, I think you’re going to see him go after incredibly high credit-card transaction costs. You’re already seeing cracks in consumer credit because of high APRs.

All of this is going to come to a head in 2025. Stablecoins could quadruple or quintuple by the end of 2025. It’s going to be an enormous market.

Should they be regulated, Chamath? At some of these congressional hearings, Tether was dragged for being the primary transfer and monetary tool for terrorism, getting around sanctions, and human trafficking. Should stablecoins be regulated, and how?

I don’t think I’m qualified to say how they should be regulated. The point is that there are immutable logs that sit in between and on either side of all these transactions, so the knowledge is there. There are also third-party services that add an intelligence layer.

The thing to question is: If you took 300 basis points of drag out of the global economy, how valuable would that be? I think it would be valuable in the United States alone to the tune of $1 trillion.

The idea that you wouldn’t do it at this point is somewhat quizzical to me. The economic justification is profound.

Gavin, let me ask you a hard question. Should the United States government give up what is, let’s face it, a bit of a stranglehold—a monopoly on U.S. dollars? Stablecoins are a competitor.

We’ve been talking about the BRICS countries starting their own currency. Aren’t stablecoins similar to that in some ways? Wouldn’t they take people off the U.S. dollar standard?

I think they might be very good for the world, but I do think they would be very bad for America if some constellation of stablecoins became the new reserve currency.

It’s such an advantage to borrow in your own currency. To give that up lightly would mean giving up control of the real size of your debt. We have a big debt, but this is a little bit like what we just did: Let’s let inflation run hot, and suddenly that debt is effectively a lot smaller than it was.

Friedberg, any thoughts on stablecoins and the monopoly we have on the dollar?

I don’t know enough about stablecoins.

That’ll be one of our big topics for the year. We should invite Jeremy Allaire from Circle and USDC.

I have a product that I exposed to the world which is just our research, and we use Stripe. Jeremy reached out to me and said he could rebuild all of our payment rails using U.S.-dollar stablecoins. I’m spending hundreds of thousands of dollars in transaction costs that I could save and use to hire somebody else or pay the team more.

My biggest business-winner prediction for 2025 is Tesla and Google.

I think this is going to be the year of AI and robotics. I’m absolutely amazed at the comeback Google has made with AI, and I’m in awe of what Elon did with Colossus. Jensen Huang and Michael Dell participated in building out xAI, and they were shocked that Elon stood up 100,000 GPUs in under 45 days.

That cohort—Elon, Tesla, xAI, and Google—is investing heavily in AI. Those names are going to be the biggest winners next year. They’re going to come out with things people didn’t anticipate.

I don’t think people are anticipating what Google is going to come out with. Gemini Deep Research is so impressive. I’m using the app, and it is a step-function improvement over anything else in the market. It blows everything else out of the water.

It’s a great explanation, but shouldn’t you explain what it is?

Good call. Let’s say you wanted to create a model based on what it would take to replace all the cars and trips in the United States with full self-driving cars. If you ask ChatGPT, Grok, or Claude, you’ll get a decent answer.

What Deep Research does is figure out the subcomponents of each question, fire off multiple threads, search the web in real time, summarize the results, provide citations, and produce a report that looks like something Gartner, McKinsey, or Boston Consulting Group would have spent 30 days and half a million dollars producing.

It’s bonkers how good it is. The amount of compute it takes is extraordinary. That’s why OpenAI released something similar that costs $200 a month, the o1 Pro, while Gemini’s is $20.

Given what Google has access to—YouTube, your Google Drive, Google Docs, Gmail—it’s going to do things you wouldn’t anticipate. It might look in your Gmail for possibilities to answer a question, or at YouTube videos from channels you subscribe to.

This is going to knock people’s socks off and solve problems people didn’t know they had.

I think it’s a great explanation.

Last year, Chamath said bootstrapped or profitable startups for the business winner. That was a great pick, and it came to pass. We’re seeing some of those companies go public and raise money at higher valuations.

Friedberg, you said commodities businesses. How did that do?

Pretty flat. We didn’t have a lot of the inflationary pressures underlying that pick, so it ended up being a loser relative to the other indices you could have bought.

I said training-data owners such as Reddit, The New York Times, and Google. Reddit stock was up 24% in 2024, and they launched Reddit Answers. The New York Times stock was also up, and they filed that lawsuit against OpenAI.

Now we’re going to move on to the biggest business loser predictions for 2025. In 2024, Chamath said pro sports teams because they had reached peak valuations. That’s looking like a pretty good prediction.

Friedberg, you said vertical SaaS companies because of AI disruption. That also seems like a great prediction. I said smartphone manufacturers. Apple stock was up 30% in 2024, but we’ve all seen that smartphones aren’t advancing much. People are taking their time, and sales are down while services are up.

I’d take the other side on pro sports teams. It feels like that market is about to be institutionalized, and funds are going to start buying professional sports teams.

That ultimately means capital markets have far more resources than the wealthy individuals who have done it to date.

Exactly. The amount of money that can go after professional sports teams is about 10 times or 100 times larger.

The reason I said that last year was that it was pretty clear to me at the time—and the NBA was the canary in the coal mine—that there was a viewership problem in professional sports, specifically in the NBA.

The game has devolved into essentially rebounds and dunks or 3-pointers, and that becomes meaningfully less interesting to watch. At the same time, TV deals determine the discounted value of these sports franchises. The TV deal was so enormous that it creates no reliable rivalries anymore because players hopscotch teams every year for compensation that is obscene.

Sports will have a decent run until the next TV deals get done. If you take pharmaceutical ads outside of television, that pool shrinks. If you have less viewership, you can sell the remaining ads less effectively because there are fewer of them and fewer buyers. The dollar pool that television networks and streamers will be willing to pay for sports will go down.

The group that will be the most price-sensitive is exactly who you mentioned, Gavin: the non-trophy buyer. When I bought into the Warriors, I bought it purely as a trophy asset. I was price-sensitive, but I agree that now that private-equity firms are on the cap tables of these sports franchises, they’re all discounted-cash-flow models and Excel models.

They aren’t buying things for emotion. They haven’t grown up thinking, “I want to own this thing,” using institutional LP dollars. That’s why I said it last year.

On behalf of Phil Hellmuth and myself, we appreciate that you made that vanity investment for the number of times we got to sit in your courtside seats.

I never went to the games. I should have gone to more games.

I went to your games and almost got thrown out of your seats. I got a red card one time.

You did?

I got a call, I think from the Warriors or the NBA, saying something to the effect of, “Chamath, this guy who was sitting in your seats was almost kicked out.”

I knew it was during a game, and I said, “Excuse me?” The guy was jawboning our own players. I was getting into it with Andrew Bogut and David Lee because I told Steve Kerr, “You’re up 30 on my Knicks. Sit these guys down. This is bush league. What if Steph Curry gets hurt out there?”

Bogut told me to shut up. Andrew Bogut basically told me to go pound sand.

This is what happens when you give Jason one of our good friends’ seats.

My wife was mortified. The guy handed me a card that said I had been warned one time and one time only about abusive behavior. If I had to be warned one more time, I would be escorted out.

I looked at the guy and said, “But—” He said, “Read the card.” I read the card, and he gave me a thumbs-up. That was it.

Three years later, I was sitting in the same seats, interacting with Draymond during the Finals. Everybody was okay with it.

When you say interacting, what do you mean?

Hold on. Gavin wanted to make a point.

I just want to come back to Chamath’s point. These private-equity firms are DCFs. The NBA has been terribly managed. Even LeBron said they have a big problem. I’d separate the NBA, which is the worst-managed league, from the NFL, which is probably the best-managed.

The counterpoint on TV rights is that Google bought Sunday Ticket and is extremely happy with it. Amazon and Netflix also bought NFL games and are happy with them. To those companies, it doesn’t really matter if pharmaceutical ads go away.

In 18 months, you’ll be able to dynamically create an ad for each individual person. If Google knows that you’re about to book a vacation, it can dynamically generate hotel destinations and show you a “click here” ad.

As long as sports command the eyeballs they do, they’ll be valuable. If the NBA doesn’t fix its problems, the value of those franchises will decline, but all the biggest tech companies are happy with the sports rights they bought. They’re going to buy them all.

Why wouldn’t Apple, Amazon, or Google just back up the truck to the NBA and say, “We’ll take the whole thing”? Everybody with an iPhone gets the NBA for free, everybody else pays, and they figure out the economics in the background. They could lose $5 billion and make $2 billion; it’s a rounding error for Apple.

I agree. They’ve all stuck their toes in the water, and I think they feel, “The water’s warm. Let’s dive in.”

The new deal is an 11-year deal for the NBA, so these things don’t come up that often. Maybe somebody will buy one of these media companies and inherit the rights.

What’s your prediction for the biggest business loser of 2025, Gavin?

Government service providers. You do not want the United States government, at any level, to represent more than 35% of your revenue.

Good choice. In the age of DOGE, they might not be spending wildly. They might actually look at the bill.

Chamath, who do you predict will be the biggest business loser of 2025?

I don’t know what the percentage drawdown from here will be, but I think when we look back, the absolute dollar drawdown of the megacaps will be in the trillions of dollars.

I’m worried about the general concentration of the top 7, 8, 9, or 10 companies in the indices. I think it’s approaching 40%. When you look at these historic concentrations, they’ve generally foreshadowed a big drawdown.

Independent of the quality of these companies—they’re exceptional businesses—I think there has been too much concentration, and it’s a setup to retrade and give back. It could even be 10%, but 10% in the megacaps would be a couple trillion dollars.

I’m on the same wavelength as Gavin. I went with the old defense and aerospace providers: Boeing, Lockheed Martin, and Raytheon. Defense 1.0, driven by China dominance and U.S. defense budgets, is going to need to shift toward a more technology-oriented and drone-driven rationalization of pricing and spending.

Palantir and Anduril will obviously benefit. There’s also a lot of failure at scale with these businesses. They’ve become too clunky and bureaucratic, as we’ve seen with Boeing, from its space-program failures in 2024 to the challenges with its airplane business.

This government-contracting business across the board is going to be deeply challenged this year with all the new blood.

So the cost-plus businesses are going to suffer, and the Andurils will thrive?

That’s right. The efficiency they gain will also be felt by their customers, and their customers might not hire them when they have software entrepreneurs advising and informing the federal government on how to run operations.

I’m pretty sure some of these service providers are going to get washed out. You could add the traditional consulting companies to the mix—the Wipros, Tatas, HCLs, Accentures, and Cognizants.

Cost-plus is a fancier way of saying time and materials, and time and materials is a fancy way of saying human-labor arbitrage. I think all of that will be scrutinized in 2025.

If you’re buying $800 wastebaskets and $9,000 umbrella hangers because of time and materials and all the other crazy things in the system, and all of that gets washed out, a lot of these business models will be put under severe stress.

The incentive of cost-plus is to remain inefficient and wasteful. These defense contractors are taking 60%, 70%, or 80% on the price of each piece of munitions or vehicle.

Cost-plus is terrible.

No, cost is terrible. That’s all.

For my biggest business-loser prediction, I had to go through a lot of the people I criticized or started fights with over the past year.

Commercial real estate came to mind because leases are going to keep coming up. I think they kicked the can down the road a little bit. Then I looked at MicroStrategy, and it made no sense to me that it was trading at 2, 3, or 4 times the book value of its Bitcoin.

I looked at Truth Social, which was doing $4 million or $5 million in revenue and valued at $7 billion. That makes absolutely no logical sense. But the one that is the most overpriced and that I think is going to see its peak valuation is OpenAI.

The headwinds for OpenAI are being absolutely underappreciated. As I said in my previous prediction, Google is kicking ass, and xAI is just getting started and building out a tremendous amount of infrastructure. Microsoft was essentially laughing at OpenAI on another podcast, saying it didn’t need OpenAI anymore because it had the source code and owned all the big iron.

I think OpenAI’s valuation makes no sense. I don’t think it will be able to keep charging the prices it’s charging, and I think AWS, Apple, Google, and xAI are going to make that $57 billion valuation the peak valuation of the company.

There’s also a nonzero chance that OpenAI could lose the process in these court cases involving the transfer of $157 billion in value from a nonprofit into a for-profit. That whole thing could blow up.

Unlike a lot of the other things you’re talking about, OpenAI is a real business with real revenue, real scale, real growth, and real technology. It’s distinct from being a meme stock.

That’s why I’m picking it. It’s easy to pick a meme stock; it’s harder to pick a real company. But I think that revenue—which is a lot of consumer revenue and a lot of developer revenue—will be challenged.

The developers I know all want to use open source. They don’t want to be beholden to OpenAI and Sam Altman. They’d much rather use open-source models, and they’re already setting their queries across multiple different stacks and trying different ones.

I don’t think there’s any loyalty to OpenAI, Gemini, or any of these services. Eventually, they’ll just use open source in many cases.

All right, let’s move to the biggest business deal of 2025. Chamath, what’s your prediction?

I said Starlink would go public last year, and I totally whiffed on that one.

I think this is the year we will see the collapse of the traditional auto OEMs. The deal that happened at the end of 2024 with Honda and Nissan is a signal of what the industry has to do, which is go through a massive wave of consolidation.

Tesla is in an incredible position with the quality of its vehicles, software, and autonomy through FSD. After a couple more meaningful product releases, I think the public capital markets will realize that these auto OEMs are uninvestable. The result will be a wave of auto megamergers.

For people who don’t know, Honda and Nissan signed an agreement to merge, and Mitsubishi is involved because it’s part of Nissan’s alliance. This would obviously eliminate a lot of redundancy.

The European OEMs are in real trouble. What does Volkswagen do? It’s not clear. What does Stellantis do? It’s not clear. These are businesses that are effectively melting icebergs.

Typically, when melting-iceberg businesses are put under pressure from smart investors like Gavin and his ilk, they’re forced to merge.

Gavin, do you have a spread trade? Are you short these names?

I’d rather not talk about specific positions, but I agree with Chamath 100%.

They’re going to lose their Chinese business because they don’t make competitive products anymore. If there isn’t massive protectionism, they’ll be caught between Tesla and the Chinese OEMs.

The only risk to Chamath’s prediction is government intervention because these companies are such big employers and are often seen as national champions. Absent significant government support, they’re all in deep trouble.

Last year, I said there would be blockbuster deals for rights holders licensing data for AI training, and Reddit did sign at least 2 of those deals.

I’ll follow on my year-of-the-robot theme. I think there are going to be massive funding deals similar to what we saw this past year for compute buildout. We’ll see massive funding deals for hardware-based manufacturing buildout in the United States.

Those deals could take the form of traditional private-market equity, or they could have a component of government support to motivate and accelerate onshore manufacturing.

The United States isn’t going to return to making things like last century. We’re going to need to move manufacturing to the next decade and the next century of production. That means making autonomous and robotic systems that will become critical for us, particularly with China’s massive ramp-up in drones, robots, and autonomous vehicles.

We’ll need to onshore a lot of this, so you’ll see a large amount of capital move into hardware buildout in the United States.

I think the biggest business deal is that there are going to be deals. After 4 years of not being able to get anything done, you’re going to see a tidal wave of M&A. There’s an enormous amount of pent-up demand.

Something will happen with Intel, and it will be big. Hopefully it’s good for America.

I also think you’ll see a lot of the independent frontier AI labs become quiet. The ultimate AI winner will be the one with the lowest infrastructure cost and the lowest compute cost. You can’t be the low-cost provider if you rent your compute from someone else because there’s a markup.

If you buy compute from Azure, AWS, or Google, you’re at a disadvantage relative to their internal services over time. The full stack wins.

Last year, my prediction was that ByteDance would go public or TikTok would be divested. We’re 18 days away from figuring out if the Supreme Court will do just that.

For this year, I was looking at all the media companies. There were so many possibilities: Warner Bros., Apple, Disney. But I think the age of autonomy is here, and there are going to be partnerships between Amazon, DoorDash, Uber, Tesla, and Waymo.

I wouldn’t be surprised if Tesla could buy Uber right now for 10% of its market cap. Waymo could spin out and partner with Uber. Amazon could buy DoorDash fairly easily.

With the wrath of Lina Khan over, it’s possible that megadeals like these could go through. If a couple of them do, whoever teams up could win autonomy, delivery, food delivery, and e-commerce.

This is an enormous space. Tesla buying DoorDash and Uber, or Amazon buying DoorDash and Uber, could be the greatest service ever created. If you wanted to build a super app, that would be it.

The other big deal potential in 2025 is autonomous drones. There’s a company called Zipline, which my firm is an investor in. Autonomous drones really are the best way to deliver almost anything to suburban America, and over time it will be sorted out so they can deliver in cities as well.

Amazon is doing this in Texas. My understanding is that they already have 60,000 SKUs being delivered by drones. They drop them in your backyard, and they’re there in 45 minutes.

I saw this in China with Meituan. They have food delivery happening with drones.

You could see something happen with Waymo this year. Waymo launched in San Francisco in an open-market way in August 2023. At the time, Uber had 66% market share and Lyft had 34%. In 15 months, Waymo had reached 22% market share in San Francisco, the same as Lyft, while Uber had fallen to 55%.

Now Waymo is launching in Los Angeles, Austin, and all over the country. It’s already in Phoenix. They also moved the hardware platform over to a new device that is supposedly going to bring capex significantly down for new launches. You’ll see much-improved ROIC metrics, which means there’s a more efficient way to use capital to scale.

The system works. It’s scaling. It’s opening in new markets. You could see a massive financing, IPO, merger, or acquisition involving Waymo this year. I don’t know whether a merger or acquisition with one of the big ridesharing companies makes strategic sense, but I do think you could see a big deal with Waymo.

If you haven’t been in a Waymo, it’s an incredible experience. Everyone I know, of every age group, who has taken a ride in one comes out saying, “That is the future. This is going to absolutely dominate how I get around.”

Some people say it’s slow, monotonous, and takes weird routes, but that will obviously be fixed over time.

I have exposure to almost all of these companies in a significant way because I believe in the entire space. About 1.5% of rides in the United States and less than 1% globally are done by ridesharing. That total addressable market is going to 20% in a very short period of time.

There are going to be a lot of winners. Uber has deals with 8 different companies, and BYD produces cars for half the price of any other car manufacturer. It also has full self-driving, and it’s pretty good from what I understand.

This is going to be a global competition. If you want to win that competition, companies like Uber, DoorDash, Waymo, Amazon, and BYD are going to form some interesting partnerships very quickly because there’s so much at stake.

I think I have a good one for the biggest business deal: consolidation in the transportation space.

Last year, Chamath predicted that the enterprise value of OpenAI would go down. It roughly doubled, but I picked it as my prediction, so I think we’re simpatico on what we think long-term.

Friedberg, you said there would be an increased probability of a nuclear weapon being used for the first time since World War II. Thank the Lord that didn’t happen.

I didn’t predict that it was going to happen. I said the probability went up.

It’s a tough thing to call.

I picked Apple as the player in AI. They launched Apple Intelligence, but it sucks. It’s terrible. I just bought the Google Pixel Fold, and Gemini works perfectly. It does everything Siri is supposed to do. If you say, “Play this song,” “Download this app,” or “Add this to my calendar,” it does it in one-tenth the time Apple takes, and Apple gets it wrong every time.

Apple Intelligence is even worse than Copilot, which is saying something.

Would you like to announce that you bought the domain “disgraziad.com”?

I own it, and I’m not selling it right now. It redirects to Jake Paul.

Gavin, do you have any idea what word I’m saying when I say “disgraziad”?

It’s “disgraziad.” It’s Italian American slang for somebody who is disgraceful. You spell it D-I-S-G-R-A-Z-I-A-D.

I bought it while watching the Tyson fight. I said to my brother Josh, “The fight is disgraceful.” He said, “Disgraziad,” and I wondered whether the domain was available.

I’m redirecting it to Tim Cook and Apple Intelligence.

Do you think Tyson threw the fight as part of the deal?

I would love to know. Somebody said Saudi Arabia was offering to host a rematch where the winner would make $50 million and the loser would get nothing.

That’s a good use of the Kingdom’s money.

The people want to know.

I also heard that Logan Paul is fighting Conor McGregor in India, and Conor McGregor is going to get $250 million.

It’s supposed to be boxing, and it’s meant to bring tourists to India and show them the country.

How do we get in on this grift? Friedberg versus Baker. Polymarket versus Calacanis. We need to do our own celebrity boxing.

Chamath, what’s your most contrarian belief for 2025?

I think you’re going to see a banking crisis in one of the major Mainline banks.

A banking crisis? Why? What’s the Black Swan?

If I had to build the case, it would be along the following lines.

If you added up the total indebtedness of Pax Americana—U.S. government debt, corporate debt, and mortgage debt—and sensitized it to rates around 5%, what you quickly realize is that the amount of debt has massively exploded.

Five-percent rates today on roughly $70 trillion is equivalent, on a dollar basis, to 10% rates 25 or 30 years ago, because we only had a fraction of that debt. The pain you feel at 5% or 6% can very quickly ripple through the economy the way it did when rates were 10% 25 or 30 years ago.

People forget the total dollar impact. When we collectively have to come up with $3 trillion or $4 trillion, how do we do that?

I think there’s a nontrivial risk that, if you have a mark-to-market problem or a credit-default problem among corporations or individual consumers, it triggers a reserve issue. The reserve issue could emerge in one of the major banks. I have 2 that I think are more obvious than others, but I don’t want to name them.

If you use Gemini Deep Research to look at total Pax Americana debt outstanding, apply the current market interest rate, and compare the interest expense with GDP over time, that would be an interesting chart.

I don’t disagree with Chamath. There could be a problem at a big bank. I don’t know that I would say it’s likely, but anything is possible.

My most contrarian belief is that, at some point over the next 4 years, America will print at least 1 year of greater than 5% real GDP growth.

I think productivity is going to go vertical because of AI and deregulation. It may not sound like a big difference—5% or 6% versus 2% or 3%—but it is massive. At 5% or 6%, the economy doubles roughly every 12 years, versus 24 years at 3%. That’s a massive difference in the wealth of the country and individual Americans.

As a specific prediction for 2025, I think the frontier labs will stop releasing their leading-edge models to prevent knowledge distillation and having their IP effectively stolen. DeepSeek was impressive, but it thinks it’s GPT-4.

The party line is that socialism was defeated in this election cycle, with a resounding vote against socialism. My contrarian belief is that we’ll see a dramatic rise in socialist movements in the United States in 2025.

We’re going to see an acceleration of progress and an unleashing of economic growth because of deregulation and AI. But some markets will also experience a downfall. We’ve talked about the decline of U.S. auto manufacturing and other industries. There’s going to be a significant shift in 2025. Some industries and companies will be huge winners, and some will be huge losers.

When you have change this fast, large contingents of people are often left behind. When that happens, socialist policies and movements gain steam.

When Juan Perón came to power in Argentina in the mid-1940s, the country was experiencing 8% GDP growth. Growth doesn’t mean it benefits everyone equally. Some people will see others go from being billionaires to $100 billionaires to the world’s first trillionaire, and that will fuel the rise of socialism.

We’re going to see an increase in the breadth and depth of socialist movements in the United States in 2025.

Government funding cuts through DOGE, reductions in federal employment, and cuts to federal contractors will create a lot of rapid change and upset a lot of people.

I totally agree. People are fond of saying that in a world of AGI or ASI, money will be meaningless. For a short period of time, money will matter more than it’s ever mattered before because the amount of money you can spend on AI and test-time compute will give you a massive advantage, whether you’re a company or an individual.

AI is going to amplify inequality for some period of time.

I hope I’m wrong. I hope AI leads to all sorts of opportunities for lower-income people, but there’s going to be a lot of employment and income disruption in 2025. It’s going to fuel socialist movements, and I think this will be a more difficult year than people expect.

Everyone in the tech industry and Silicon Valley thinks things are rosy, but life on the ground for most Americans could be much harsher than any of us anticipate. That could make for a very difficult political and social environment.

It’s a great contrarian prediction. I’d build on it by saying I don’t think this will just hit blue-collar workers.

I’m seeing people in the venture industry and entrepreneurs all over the place—people who are highly qualified and had six-figure or even mid-six-figure salaries—unable to find work at their previous compensation levels.

People are doing more with less. It’s better to invest in deep research and AI, automate or deprecate things, or delegate work to other regions than to hire Americans in some cases. That philosophy isn’t just going to hit truck drivers. It’s going to hit developers, designers, and writers.

What’s the contrarian part of what you’re saying?

Socialism is on the rise. The party line has been that socialism was knocked back this year, that there was a mandate against socialist policies. I think we’ve got that wrong.

Wokeism and progressivism will decouple from socialism. Wokeism and progressivism will be on a declining trend, but socialism in terms of government policy will rise.

Why don’t we have universal health care? Why don’t we have pre-K in every state? Americans have the right to ask why our government has failed at basic things such as after-school programs, universal health care, and universal child care.

You’re assuming those things are easy to do. That was also said about education when we gave everyone access to college through federal student loan programs.

The schools started charging more, tuition went up every year, and eventually the cost of education inflated away the benefit. The same thing has happened in health care, housing, and education. In every market where the government has stepped in to provide capital, the market no longer operates freely.

I appreciate the challenge. I think it’s easy if you put these things down to the states and make them more competitive.

If you introduce school vouchers and create competition, or take universal health care and run experiments where different states receive money from the federal government to run 50 different experiments, I think we could solve some of these problems.

You’re correct that anything the federal government does eventually becomes corrupt and inefficient.

My most contrarian belief was that OpenAI loses its lead, loses its nonprofit-to-for-profit transition, and becomes the fourth-largest player in AI. The total collapse of OpenAI is my most contrarian prediction for 2025.

That’s a good one.

I’m doubling and tripling down. It’s a spicy category.

Jeff Bezos could still run for president. I wanted him to. He bought The Washington Post, quit his Amazon job, bought a house in Washington, D.C., and went down to Mar-a-Lago.

Don’t be surprised if, after he gets through his midlife crisis, goes to Coachella, parties, and has a great time—which he deserves—he comes back and says, “I want to serve my country.”

Gavin, do you think he’ll hold on to The Washington Post?

Let’s stick with that one. Jeff Bezos will sell The Washington Post in 2025.

That’s a good Polymarket prediction. Sell it to Kara Swisher, have her run it into the ground, and double down.

Let’s move to best-performing asset. Last year, Chamath did a spread trade: long public tech stocks, short the private late-stage tech-stock index. The Nasdaq-100 tech-stock ETF was up 10% in 2024.

I chose a uranium ETF, which was down 1% in 2024. I looked at the components of that ETF last week, and it had absolute junk in it. It wasn’t the right way to trade uranium.

I went with the on-demand economy: Uber, Airbnb, and DoorDash. Uber was up 30%, Airbnb was basically flat, and DoorDash was up 74%. Shout-out to Stanley Tang and the team there.

Gavin, what would be the best-performing asset of 2025?

I think the companies that make high-bandwidth memory. We’re going to run out of compute.

It’s a shocking statistic: high-bandwidth memory is a bigger portion of the input costs on GPUs than TSMC is. Today, 2 companies can make it: SK Hynix and Micron. We’ll see if Samsung gets its act together.

HBM is in Nvidia GPUs, AMD GPUs, and Amazon’s Trainium chips. In a world where test-time compute and inference are so important, HBM is arguably more important than ever. It has been sold out for the last 2 years.

High-bandwidth memory would be my pick.

Let me preface this by saying that this pick goes to absolute zero 92 times out of 100. Six of the remaining 8 times, you make 10 times your money. The final 2 times, you make anywhere from 100 to 1,000 times your money.

This is a loser trade, but I would be long CDS—credit default swaps. I’m buying insurance, buying protection that there is no default event in 2025. I’m not going to tell you which companies or maturities, but the general idea is that I’d like an insurance policy in 2025 so the men and women we’ve elected have a chance to do their work in peace.

There’s a small chance of volatility next year. I hope this trade loses money, but if it hits, it will be the best-performing asset of 2025. It would be the equivalent of Bill Ackman buying CDS at the beginning of the COVID crisis.

How does someone buy those?

You need an ISDA. You talk to the big investment banks, and they price it out for you.

This isn’t something I think will happen, and it isn’t something I want to happen. But if you look at the amount of money you can make and the massive risk asymmetry, when you look at the concentration of the S&P 500, the total gross amount of debt, and rates spiking, having some insurance may not be a bad thing.

If Chamath’s prediction of a bank failure is true, you absolutely want to own CDS. You won’t make 100 times your money. You’ll make 1,000 or 10,000 times.

I went with Chinese tech stocks or Chinese technology ETFs. Everyone has dumped Chinese tech stocks over the last couple of years and taken an isolation stance, saying, “We can’t do business with China.”

I think the Trump administration, particularly with its recent request to halt the TikTok ban, is trying to line up what I would call the great deal with China. I don’t know what they’re actually trying to do, but I think the U.S. government wants to open the Chinese market to American companies in order to give Chinese companies access to the American market.

I think they’re going to get a deal done. There are 3 drivers.

The second is the cost of electricity production in China. They recently approved a $137 billion hydroelectric dam facility that will add another couple hundred gigawatts of electricity production, not to mention all the nuclear buildout we’ve discussed. The cost per kilowatt-hour is already lower, and the amount of electricity available is rising.

The third is that the Chinese Communist Party has an incredible ability to throttle free markets and entrepreneurship up and down. This may be a moment when it turns the throttle toward enabling more innovation and free-market activity.

When you put all of this together, there are a lot of Chinese tech companies that have been beaten up under the assumption that there will be a difficult conflict with the United States. I don’t think that may be the case going into 2025.

Alibaba trades at a reasonable multiple. These stocks could be poised for a strong run in 2025 if the macro environment works out.

This feels like a contrarian take. Trump isn’t going to find a way to balance the relationship with China very well, and I don’t trust any accounting statement coming out of China because they can fudge whatever they want.

Gavin, what do you think of this hot take?

I’ve had a no-China guideline ever since the Longtop Financial fraud more than 15 years ago. A lot of great investors owned it, and it had a Western auditor. Then it turned out the fraud was happening at the local post office. They were opening documents that local auditors had signed off on, changing them, and sending different documents.

It’s a hard place for someone who isn’t Chinese to make money, but I agree with a lot of what David said. Trump and Xi both want a deal. There’s a deal to be done that leaves Putin out in the cold.

If that happens, there are high-quality companies trading at mid-single-digit multiples. It’s a big market, and Chinese companies serve a global market—not just the United States, but Africa, South America, and the entire Southern Hemisphere.

BYD is all over Europe. Chinese companies have the best unit economics and the best cost of production. If you believe this is the year of the robot, there’s going to be massive global demand for automation and rebuilding manufacturing capacity. China could service those markets more efficiently than any other country of origin.

It’s a powerful set of macro drivers.

I think the Magnificent Seven will be the best-performing asset. I’m taking the other side, Chamath. What these companies have learned over the last couple of years is that they can generate incredible earnings expansion by not hiring people, outsourcing jobs, and automating.

The gains we’ll see from AI—which these companies are producing for other companies and consumers—are being applied internally first. That internal application will allow them to achieve earnings growth people won’t be able to comprehend over the next couple of years.

We’ve got a lot of different takes going in different directions.

Let’s do the worst-performing asset. Last year, Chamath said the late-stage tech-stock index, mostly SaaS. Friedberg put on a brilliant spread trade: short vertical SaaS and long AI cloud providers. SaaS had a rebound, while Google was up 36%, Microsoft 14%, and Amazon 46%.

I said LLM startups like OpenAI and Anthropic because too many players and open sourcing would kill pricing. The one valuation we can track is OpenAI, which doubled, but I think that will work long-term.

Gavin, what’s your worst-performing asset of 2025?

Enterprise application software.

2025 is going to be the year of agents, particularly in the second half. Agents are AI models that can take action on your behalf—anything online that a human can do online.

If the labs and big cloud providers dominate agents, which seems likely, enterprise application software will be in a lot of pain. Some of these companies are talking a big game about agents, but they don’t have their own models or compute, and I don’t see them being the ultimate winners in an agentic world.

I could be wrong. They have customer data, some data moat, and strong customer relationships. But most companies also have relationships with AWS, Google, or Microsoft. Enterprise application software will be in pain.

Gavin absolutely nailed it. I’m going to double down.

There’s a term we started using internally at 890 in 2024: the software-industrial complex. These are large, bloated enterprise software companies that have convinced enormous numbers of organizations to spend tremendous amounts of money wrapping heuristics and business rules around a CRUD database.

They’ve perfected a go-to-market and sales motion: golf trips, steak dinners, and everything Alex Karp has railed against. None of those things equate to product value.

In the world of agentic software and AI, you can rebuild many of these workflows in very efficient ways. The go-to-market will be driven by CEOs and CFOs exerting pressure on CIOs to manage spending.

These next-generation AI businesses are built an order of magnitude cheaper than the companies they compete against. Even if they offer the exact same features, their cost structure is meaningfully lower. You put your time and money where your mouth is on this one: 11 months ago, you founded 890, and getting 80% of the traction in less than a year is shocking. I think our 30-person engineering team does the work of 300 people; by next year it may do the work of 3,000, even if it only grows by 10% or 20%. The reason is that we use these tools and are productive to an order of magnitude I didn’t think was possible, so we’re able to price it differently.

The return-on-investment calculators and traditional go-to-market models are going to be blown up. In an RFP or sales environment, someone says it costs $100, and you show up and say, “$10.” At $10, it can still be highly profitable.

Fundamentally, these enterprise application software companies are based on making white-collar employees more efficient. The AI companies are going to say, “We’re just going to replace that worker.” It’s a fundamentally different paradigm shift.

I’m going to triple-underline vertical SaaS again. Per-seat pricing models will be challenged, and pricing will be compressed as companies explore in-house tools built with AI that replace traditional business practices.

It was an obvious one on my list, but I want to make a different prediction.

Consumers make 5 big decisions in their lives: college, spouse, kids, cars, and homes. The consumer is up against it with record debt. Since you can’t trade college, spouse, or kids, I think legacy car companies and real estate will face continued headwinds and be terrible assets.

We’ve overbuilt in some cases. There are tons of cars on lots, and people can’t afford homes with these mortgages. Those are going to be the 2 worst-performing assets.

If you look at Texas, housing values have gone down 2 years in a row, and rent has gone down 2 years in a row. The same thing is happening in other states where they allowed people to build, while people are leaving states where they don’t allow construction.

Let’s move to the most anticipated trend for 2025. Last year, Chamath predicted that Bitcoin would hit $100,000 for the first time. He nailed it with a half-court shot.

Friedberg, you said predictive models and AI-driven discovery in pharma and bioengineering.

There was a lot of funding, and Demis Hassabis won the Nobel Prize.

I picked efficiency in the form of AI advancements in labor and outsourcing.

Chamath, what’s your most anticipated trend for 2025?

There are a handful of canaries in the coal mine for the end of the deep state. I’d like to point to one obscure thing called the supplemental loss ratio.

It’s a mechanism banks can use to include or exclude Treasuries when calculating reserves. Why is this interesting? If we’re unable to manage the debt situation in 2025, what you’re going to see is maneuvering at the edges of these arcane regulations to kick the can down the road.

There’s a chance that could happen to help the banks because America may still need $10 trillion or $20 trillion of debt issuance to refinance the $10 trillion maturing this year and plug holes in the coming years.

We don’t need to understand all the details. If we don’t move the goalposts, it’s a great sign that the people running the show are the ones we elected.

My most anticipated trend is the announcement of a buildout of nuclear power in the United States in 2025, as a function of deregulation and new technologies.

I think the new government will be much more accommodating. It’s a necessity because of the rate at which we have to build power to meet competitive demand from China.

The United States will need to add more electricity-production capacity than we can scale with any other renewable source, so nuclear is inevitable. I think deregulation will happen in 2025.

I know a lot of smart people who are starting nuclear power companies or leaving very good jobs to do so in anticipation of this. I’m very bullish.

That’s a leading indicator. When smart people do something with their time, it’s a great indicator.

Gavin, what do you have?

AI will make more progress per quarter in 2025 than it did per year in 2023 and 2024.

With o1 and o3, we keep changing the goalposts for the Turing test and AGI, and we’re going to blow through them. We were scaling around pretraining on one axis, then we started scaling test-time compute. We’ve now added a third axis: reasoning.

The internet is composed of answers people give. Models benefit from the internal monologue of somebody getting to that answer. In AI terms, it’s called a reasoning trace.

You’re using models to generate synthetic data that contains those reasoning traces. You ask a model to solve a problem that has a verifiable answer and tell it to show its work. You have it do that many different ways and times, pick the best examples, feed them back into the model, and apply reinforcement learning.

Now you’re scaling along 3 axes that are multiplicative with each other.

Someone on the Google team tweeted that it’s going to be a straight shot to ASI—artificial superintelligence—and that might be right.

Ilya Sutskever gave a talk at the Naval Postgraduate School. He said these reasoning models are inherently unpredictable. The best reasoning models in the world today are the ones that play games, in the AlphaGo and AlphaZero style. They constantly make unpredictable moves that no human grandmaster could have come up with.

Now these models will make similarly unpredictable leaps in all sorts of domains. Hopefully that will be awesome, but it might not be.

They’re going to go around corners people might not have considered. To circle back, I asked Deep Research which banks had the greatest chance of becoming insolvent or experiencing a financial crisis.

It researched websites, found a list of U.S. banks, ranked the top banks by assets, found their latest financial statements, looked at capital-adequacy ratios, and examined loan losses. The research took about 10 minutes and pulled data from 162 websites.

It produced a report analyzing JPMorgan Chase, Bank of America, Citigroup, Wells Fargo, Goldman Sachs, and others. Its conclusion was that Citigroup and Wells Fargo appeared to be at the highest risk of insolvency or financial collapse compared with JPMorgan Chase, Bank of America, and Goldman Sachs, based on Citigroup’s recent net loss, lower CET1 ratio, and higher exposure.

I don’t know how much of it is correct, but it’s a pretty good starting point. Deep Research is an impressive product.

My most anticipated trend for 2025 was alluded to in an earlier prediction: exits and DPI showering down. With the end of Lina Khan’s tenure, I think M&A and IPOs will surge and we’ll have an incredible distribution year.

Let’s go to most anticipated media. I had 2 choices: James Gunn’s DC Universe, with Superman coming out this year, and Andor season 2. I think both could be amazing. I also predict they’re going to do a live-action Clone Wars series.

If you don’t know The Clone Wars, watch it with your kids. It’s an amazing animated series that takes place during the prequels.

But my most anticipated media is seeing what happens with legacy media outlets owned by billionaires or people who no longer want to pick sides. The Washington Post, CNN, and the Los Angeles Times are steering toward the middle and trying to get back to classic journalism.

The editors are revolting, Kara Swisher is upset, and they’re adding more right-wing voices. It’s going to be popcorn time for everybody.

Chamath, what do you have for most anticipated media?

It’s the enormity of the files that are going to be declassified and released by the Trump administration.

I think it’s going to be unbelievably interesting, salacious, useful, and earnest—all of the above. The JFK files, the Epstein files, the Diddy files, the moon landing, and who knows what they find across other fringe conspiracy theories that may turn out to have some shred of truth.

All of that content released in 2025 by the Trump administration will be incredibly interesting.

I’m interested in AI video games. The cost of production comes way down when you use AI, and you can have dynamic storylines and new gameplay concepts that don’t exist today.

The creative and technical talent you find at development houses can be unleashed with tools that have recently come to market. We’ve seen a lot of generative video, but video games can also be rebuilt structurally.

The video-game engine can run locally and generate parameters that use an existing rendering engine. You can have entirely new storylines and plot sequences.

I think there’s going to be a rewrite of video games and the video-game industry with the AI capabilities hitting the market. It’s going to be incredible entertainment. People who don’t play video games are going to find things they love.

My most anticipated media for 2025 is the second season of 1923.

I’m normally a science-fiction, fantasy, or spy kind of guy when it comes to television, but 1883 and 1923 were the first shows that hit me the way Game of Thrones did. I’m crazy excited for the new season.

Are you watching Landman?

I haven’t watched it yet, but I’m excited to try it.

It’s quite fun. Taylor Sheridan is in the zone.

I highly recommend The Day of the Jackal.

That was really good. Sicario, The Jackal—great suggestion.

Last year, Chamath predicted MrBeast. Friedberg predicted AI-generated news. I predicted Gladiator 2 and The Three-Body Problem.

Gladiator 2 was mediocre, and I didn’t finish The Three-Body Problem, so I guess it was mediocre too. MrBeast’s show on Amazon Prime was the top unscripted drama in 140 of 180 countries.

Let’s do our prediction markets. I’m going to put up a prediction market based on immigration and Trump’s promise to deport 15 million people from the country.

In the first year, I’ll set the over-under at 5% of that stated number, which is 750,000. After 1 year in office, will Trump have deported more or fewer than 750,000 people?

How are you going to measure that? There has to be a source of truth.

We’ll use the White House’s reporting of deportations.

Chamath, do you have a prediction market you want to put up?

The MAG 7 representation in the S&P 500 shrinks below 30%.

That’s a good one. Dispersion would happen in the other stocks, with less concentration.

I also want to see whether Microsoft, AWS, or Google Cloud has the fastest revenue growth in 2025. I get the sense that Google is accelerating ahead.

Google is smaller, which makes it easier to grow faster. The interesting question is who has the largest dollar gain in cloud revenue in 2025: Google, Amazon, or Microsoft. Azure versus AWS will be interesting.

The other one I want to do is the national debt. The national debt has grown about $2 trillion per year in each of the previous 2 presidential administrations. Over the last 8 years, we’ve gone up more than $16 trillion.

Trump said he would not increase the national debt, so I’ll set the prediction at an increase of $1 trillion over the next year.

It will take time for the policies to get going. Immigration will take time, too.

We’ll use the U.S. Treasury’s December 2025 report on federal debt. Will federal debt be above or below $38 trillion?

It will be above, but you need to set the line where people will take either side. If it’s obvious, everyone will take the same side.

Perfect. We’ll see whether they can control the spending or whether it will just be the same.

Gavin, at some point, if you’re back on All-In, I want to talk about UFOs and the drones.

Let’s talk about it now.

I don’t know what they are. It’s very clear, if you look at statements from the New Jersey mayors and governors who met with the police, FBI, and Defense Department, that they don’t know either.

Trump said someone in the government knows what they are. He also said he wasn’t going to go to Bedminster anytime soon.

Over the last 8 years, every 18 months there has been a major story in The New York Times, The Washington Post, The New Yorker, The Atlantic, or another credible media source, with dozens of interviews with fighter pilots and commercial pilots talking about seeing things with advanced sensors that made no sense to them.

There was an article in The New Yorker quoting people who had worked at the Skunk Works laboratories in the 1950s. They said they had seen extraterrestrial materials recovered from a crash, and that it was one reason the United States made such big leaps in materials science.

There was also the ʻOumuamua incident, where many astronomers—including the head of astronomy at Harvard—said it was clearly a UFO of some sort.

My conspiracy theory would be that, if these are UFOs and not government drones from the United States or China, the most likely explanation in New Jersey is some sort of drill. Once the hysteria starts, people misidentify commercial airplanes as drones. Why would UFOs have blinking green and red lights?

But it’s interesting that there was a major concentration of reports as we were scaling into nuclear technology from 1945 to 1960, and now that AI is getting started—the next technological phase shift for humanity—there’s another big wave of reports.

You can come back next week. You’ve won. Congratulations. We have our winner for All-In Idol. Gavin is now the fourth Bestie.

Friedberg, do you think there are documents that Trump could release about UFOs? There must be an entire spectrum of things that could be subject to a FOIA request.

It depends on how deeply classified they are. If the government doesn’t want to give something up, it won’t give it up.

There’s a broad spectrum of claims, but you have to distinguish between evidence of extraterrestrial life and reports of unexplained phenomena.

Let’s be honest. Do you think there’s documentation that the U.S. government has knowledge of UFOs or things it cannot explain?

I think the premise that extraterrestrial life would visit us in physical form is rooted in humanity’s current understanding of technology and biology.

There’s an extension of information gathering that doesn’t require moving physical biological life forms from one part of the galaxy to another. Eventually, every civilization reaches a sufficiently advanced level of technology that it no longer needs to move biological organisms around physically. It can gather and affect information.

Once you can convert any molecule into any other molecule and have access to sufficient energy, you can turn your local part of the universe into anything you want. You’re simply gathering information from the rest of the universe.

I don’t think it makes much basic sense that alien bodies would want to move around the galaxy. The whole premise of UFOs moving bodies around is rooted in the current state of human technology.

What percentage chance do you put on the U.S. government sitting on knowledge of extraterrestrial life?

At least 20%.

I’d say 25%. I’d take the over on 20%.

Friedberg, what chance do you think there is that the government is sitting on extraterrestrial life or proof of it?

I don’t generally align with the idea that our narrow understanding of technology and biology is what visits us. The more likely possibility is information gathering rather than physical biological life forms traveling across the galaxy.

If Trump releases information on extraterrestrial life, I’m voting for him to have a third term. I’m going full MAGA. He gets a third term from me, even if it requires changing the Constitution.

Do you think extraterrestrial life built the pyramids? Is that the most plausible explanation for their accuracy?

I don’t know. It is strange that, in so many cultures all over the world, there are depictions of what look like astronauts. There are also Renaissance paintings that appear to show what we would call UFOs in the skies over cities.

We’re closer in time to Cleopatra than Cleopatra was to the construction of the Great Pyramid. That was a very long time ago.

The most credible explanation I’ve seen is that the Egyptians used water to raise the stones in channels. They placed the stones on wooden slats, floated them upward through canals like a Panama Canal lock system, positioned them, and released the water.

That sounds plausible to me.

This has been another amazing episode of the All-In podcast. Thank you, everybody. Let’s have an amazing 2025. Let’s kick ass and take names. We’ll see you all next time.