DOGE 对阵 USAID、加密框架、Google 的 750 亿美元 AI 支出、美国主权财富基金、GLP-1

How’s the feedback been, Dave?

I got so much great feedback on your Ray Dalio episode. Is it the end of the empire, or are we coming back?

What’s awesome is that a lot of the recommendations he shared are becoming policy, it seems. Trump, Elon, Bessent, and others in the administration have echoed the idea of trying to get the government deficit below 3% of GDP. That seems to be the economic magic number, and if you can do that, rates drop.

It was great to have him publish that a couple of weeks ago and talk with us about it. Today, we’re super excited to have our friend Antonio Gracias joining the show. Antonio is the CEO of Valor Equity Partners, and he’s made some solid investments. He was one of the first investors in Tesla, SpaceX, and Athena. He was the second investor in Athena after me. Welcome to the program, Antonio.

Thank you, Jason. I was also in Uber with you, too, by the way.

Yes, yes. I did the Series A, I believe. Or the B? What did you do?

Right behind you. We were in the B behind.

Yes, you were in the Shervin round. Well, welcome to the program, Antonio Gracias. We’ve got a full docket today, and we might even have a special caller from the White House. No promises, but you never know.

We are 17 days into the Trump 2.0 presidency, and it seems like the main character at the moment is DOGE, the Department of Government Efficiency. They seem to have found a little-known agency, USAID. Let’s unpack it and talk about USAID and DOGE, maybe in 3 acts.

First, let me educate the audience on what USAID is and then get your general reaction to it. Most Americans probably haven’t heard of USAID. It stands for the United States Agency for International Development. It was established by JFK through an executive order back in 1961. The Wall Street Journal summed up its purpose as, quote, “make friends and influence countries in the American interest.”

According to the U.S. government website, the purpose of USAID is to extend assistance to countries recovering from disaster. It had a budget of about $45 billion a year under Biden, which is about $150 per American per year. It had at least 10,000 employees, or did, and as of 2023 had programs in 130 countries. Obviously, there are 195 countries in the world. The budget doubled under Biden, from $26 billion to $45 billion. The budget was between $15 billion and $20 billion during Trump’s first term.

Then the White House and, I guess, Elon and the DOGE team found out about this. DOGE went in there and found all kinds of interesting spending: $2.5 million to fund an EV charging station in Vietnam; $2 million for sex changes and LGBTQ activism in Guatemala; $1.5 million for a Serbian LGBTQ group; $70,000 for a DEI musical; and $47,000 for a transgender opera in Colombia. The DEI musical, by the way, is in Ireland, so if you make it to Galway, you can see that DEI musical. This went on and on, and it has become quite a story.

Antonio, you’re our guest here this week. You, I, Sacks, and Elon spent a little time at Twitter during the takeover, where I think a lot of these techniques were first put into action. Your thoughts on what’s happening with DOGE?

I want to thank you guys for having me. It’s great to see everyone.

Jason, have you guys seen the musical? I like to see musicals. No, I booked it. We’re all going. What is it called? Is there actually a name for the musical?

Yeah, it’s DEI: The Musical.

Oh, man. It’s a heartwarming story: “I apologize for getting a job. Sorry, guys, you’re not qualified for it. I apologize.”

I think the DOGE story maybe starts with the Twitter takeover. Twitter was spewing the woke mind virus into the world, which is why you did it to begin with. When we got there, as you know, it was basically breaking even, and there was $12.5-ish billion at 10%, so a billion and a half dollars in interest costs that there was no money to pay.

That led to the turnaround, which I think was the biggest turnaround of all time—literally the biggest turnaround, I think, ever in history. It was the second-biggest tech deal ever done. Eighty percent of the people were gone, and that doesn’t include all the contractors who were there.

The company is now servicing its debt, and they just priced one of their bank debt deals at 97 cents. This is a huge win. It means the company is doing extremely well. Other people bought the bank debt, and it then traded up to a little over 98 cents—98.58. It’s a giant business success, a rousing business success. Jason, you were there for part of it. You saw it.

It was a disaster. There were tampons in the bathrooms and every woke thing you could possibly imagine, and no one in the office. It was so bad that the pens had gone dry in the conference rooms, which is incredible.

Do you mean the pens had gone dry because they were never being used, or because they were being used so much that they were never replaced?

I think it was both. We got there on Halloween weekend, and we started whiteboarding. There was no ink. Everything was incredible. Flowers were impeccable, food was being made fresh every day for thousands of people and thrown away 3 times a day. It was tossed in the garbage; we didn’t give it to the homeless. It was shockingly bad—bureaucracy gone mad. Bad incentives mean people don’t care.

I think this turned into 2 things. One was just to stop the woke mind virus from going into the world. The interesting thing is, they do external brand-safety checks there now, and they have a 99% rating in brand safety. So, stop the woke mind virus from going into the world, number 1. Number 2, fix the company. Both have happened. It’s been 2 and a half years, really, and both have happened.

I think you take that as the warm-up act for what’s going on at DOGE, which is that the president of the United States, President Trump, wants to make America great again. That begins with making the government great, and you have to fix it. This is the biggest turnaround of all time. President Trump has the courage to do it, and he’s got a great ally in Elon to make it happen at DOGE.

The numbers are pretty easy. We’re spending $6.5 trillion, we bring in $4.5 trillion, and we’ve got to find $2 trillion somewhere. Interest costs are $1 trillion a year. The bond markets were going up—the long bond was going up a lot—because people believed that we couldn’t stop spending and creating inflation. You see those trading down now as DOGE is starting to take effect. People see that it’s real.

Elon said this publicly: fraud, waste, and abuse. He thinks it’s kind of 10% of the budget is probably fraud. I think it might be low, actually. You’re talking about $650 billion to $1 trillion in waste. I think that’s probably about right. That alone fixes the problem.

Jason, for me, what scared me when I first started thinking about this was that we had a democracy that turned into a bureaucracy. What I’m afraid of now is that we have a bureaucracy that is about to turn into a kleptocracy—a Latin American-style kleptocracy.

The stuff you’re talking about is pure fraud, and you’re making some jokes out of it, the DEI musical. But if you go into the data that’s coming out of USAID, what you find is that there are a lot of political contributions going on. Politico itself is being funded by USAID. That is pure corruption. That’s a Latin American-style autocracy, and we cannot let it go there.

I think this happened at a good time, and I’m super grateful to all the people there. There are 80-plus people there, all patriots who have gone full-time.

Let me try to give you a little bit of historical context, because I think that’s important. Whenever I hear so many people breathlessly saying some version of “WTF,” as if this is totally new, it’s not new. I’ll give you 2 examples, but I really want to double-click on 1 of them.

In 1941, the Truman Committee was formed because there was a fear that spending by the Defense Department was completely out of whack. Over the next 6 or 7 years—and this is really what gave Truman the credibility to then become Roosevelt’s vice president—they found incredible levels of waste, fraud, and abuse before the war effort and then after Pearl Harbor during the war process.

Over 7 years, that committee—which was a Senate committee—was tasked and budgeted with only $20,000 or $30,000. Over the next 6 years, they spent less than $1 million. Inflation-adjusted to 2023, that was about $65 million. Do you know how much they saved? It’s estimated they saved somewhere between $1 billion and $15 billion in 1941. That’s a quarter of a trillion dollars in 2023.

The second example is that we did this under President Clinton, and that was called the National Partnership for Reinventing Government. The first thing that’s important to acknowledge is that this is not new. We’ve done DOGE twice before, and both have been successful.

The Harry Truman one was incredibly important because it really set guardrails for how this could be done. Everything in that committee—it was a Senate committee—was unanimous. Republicans and Democrats found waste, fraud, and abuse everywhere, and they saved an enormous amount of money.

The interesting thing is that these last 2 versions of DOGE were driven and led by Democrats—the same people today who are basically saying, “Hold on a second. We have congressional committees or inspectors general.” They lack the awareness to know that their own party was the driving force for this 2 times before.

I think it’s important to acknowledge that those methods aren’t good enough anymore, because what we’re finding in the early days is that the rot is pervasive. There’s no accountability. So, I think you need people to look at it with fresh eyes.

What is DOGE at the end of the day? They are read-only auditors of the truth. The press secretary in the White House made this very clear. It’s incredible what read-only access gives you. All it allows you to do is take the data and present the data. They can’t manipulate it, and they’re able to publish all of this stuff in real time. That’s why this is so important, because if it went into a congressional committee, to be honest with you, it would sit there and stew for 6, 7, 8, or 9 months. You might get small little tidbits of it, but now you’re getting the full thrust of it.

The biggest thing that I find concerning is what Antonio said: The media, which are supposed to be this intermediary layer that is totally objective between the government and the people, weren’t independent at all. They had all kinds of hidden incentives—$8 million to Politico, several million dollars to the BBC. I think it’s important to ask what’s going on.

That also has historical context. This is exactly what happened in the 1960s and 1970s, when it turned out that record companies were paying DJs to play songs. There was a huge set of lawsuits and trial cases, and the result was a change in the law. We call that payola now: You cannot take this money without disclosing it.

Had this money been absorbed by these entities and actually disclosed, maybe we’d be okay. Nico, I sent you a link. Maybe you can throw it on the screen for these guys. We talked about this, and we didn’t realize how connected it all was. During the election cycle last year, we were asking ourselves why all these articles were buried and why we weren’t really getting the truth.

It turns out that the people who were responsible for telling the truth somewhere along the chain were co-opted or just told not to tell the truth, influenced by all of this back-channel money going back and forth from the government to these folks.

Let me just give some numbers to what you were referencing there. The USAID organization has been giving money, as have other agencies, to journals, databases, and subscriptions. There’s probably some amount of that that makes sense. However, when we look at this, during Trump’s first term, spending on Politico was averaging around $1.3 million a year, but it suddenly ballooned up to $8 million a year under Biden. You can see the quarterly payments here on this chart.

There’s some normal amount to spend on publications or for a library at an organization. What we’re looking at here is all federal agencies, and suddenly during Biden there’s a very suspicious ramp-up in spending on Politico. A lot of this is breaking news, and it hasn’t been verified yet, so we’ll put that caveat on it.

There’s a number of $34 million that’s been floating around. That’s all the years back to 2008, not just 2024. A lot of people have that number, I suppose, misattributed. When Politico was acquired back in 2021, it was doing about $200 million in revenue, so this would be about 4% of its revenue. It’s pretty significant.

The BBC also received $2.7 million in funding from USAID in 2023. That was 8% of its annual income, which is a little suspicious. Thompson Reuters, the consulting arm of Reuters, received $120 million from the federal government since 2011. That has to be looked at and double-clicked on, and half of it came during the Biden administration.

The New York Times hasn’t actually received all that much—$370,000 from the federal government last year under Biden. It went from $100,000 a year to around $300,000. The New York Times data has been cleaned up a little bit.

All of this is to say that there’s spending going on with the press that certainly doesn’t look good and should obviously be verified and challenged. We’re in a breaking-news environment, so we’ll see how that information shakes out over time. One of the great things about DOGE is that it’s getting this information out there, and citizen journalists are looking at public databases.

Friedberg, we were having a conversation about this 2 years ago when you kept harping on every episode about our debt and the interest payments. You were way ahead of the curve, and now here we are. We didn’t think anybody would ever take this cause up, and now it’s the cause of the moment. What are your thoughts on the first 2 weeks of DOGE?

Magnificent. What else is there to say? This is the eggplant emoji. It’s what we needed.

If you zoom out on what DOGE is doing, I think the best way to describe it is zero-based budgeting. In organizations that go through zero-based budgeting, you do a cycle, typically annually, where you take all of your opex—all of your expenses in running a company or an organization—and you take it down to the studs. You take it down to zero and rebuild it up.

You say, “What are we trying to achieve this year from first principles?” Based on that set of objectives and goals, what do we need to do? What’s the minimal expense we need to run? You don’t start with last year’s numbers and ask what else you need to do, adding on top of that. You do a hard, high degree of scrutiny on every dollar that moves out.

That’s what DOGE is effectively doing. They’re doing zero-based budgeting on the federal government. They’re looking at every line item and asking the fundamental question that I don’t think we talk about in the public discourse enough: What is the essential role of government?

There’s a great debate to be had around that point. Should the government be providing humanitarian aid in international markets? That’s a good debate to be had. Should the government be providing security to nations that can’t provide security for themselves? Does the U.S. government have a role in that? Should the government be providing loans for people to go to universities? Should the government be providing loans for people to buy homes that are overpriced?

As we start to ask questions about how we’re spending money, I think it leads to the most important question: What is the essential role of government? That’s the debate that needs to be had in order for the democracy to last. I’m very happy to see the effort of DOGE, and I think it’s the beginning of what I hope will be a long-term process of asking the fundamental questions about essential government.

Antonio, zero-based budgeting was the first thing you introduced when the Twitter takeover happened. You asked, “What do we need in terms of design? What does the sales team need to hit these sales numbers? How many servers do we need?” The waste, fraud, and abuse at that company were shocking. They had buildings that were $100 a square foot, if I remember correctly, being used to store furniture from the previous building that had been upgraded. They were storing it in Class A office space.

You referenced the commissary. Twenty people were having lunch a day, but they had never ratcheted down the amount of food they were making, so each meal was about $800 on average when you did the math for those 20 meals in the San Francisco office.

Could you talk a little bit about how the shock-and-awe campaign of just freezing spending, and then seeing what was actually necessary to accomplish the task, worked out at Twitter—and how you see that being executed inside our government?

The Twitter experience was pretty extraordinary. The first thing you do in a turnaround like that is try to get the checkbook and just turn payments off, then see what happens. There are a lot of people who shouldn’t be getting paid. Sometimes they complain, and sometimes you find that the people who complain the loudest are the ones committing the worst fraud in the whole game. The worst grifters are the ones crying foul the most.

At least at Twitter, there were financial statements. There was an audited financial statement, and they were basically correct in some ways. There were some issues with user growth, DAUs, and so on, as well as the incentive plans, but the actual cash-flow numbers were pretty much correct.

Here, the problem is much, much deeper, and it makes this zero-based-budgeting question much harder. The way the government works is that a department basically asks Treasury for money, and Treasury sends it out.

We all run businesses. There’s a reconciliation process. You have a contract, you issue a purchase order against it, something comes in, you check that it came in, the service is rendered, and then you issue a payable. A month later, you pay it. That doesn’t happen in the U.S. government. That process is broken. It used to happen, but it’s broken now.

I used to ask myself, “Why are the numbers always revised? Why are they always wrong? How can the government know how much money it’s spending? Just hit the button in the computer and figure it out.” The problem is, that button doesn’t exist. We spent time early on trying to track how the money actually flows. No one could tell us how it actually flows or where it’s going out. People didn’t know, and it’s totally crazy.

Does it all go through Treasury, Antonio? Can you just explain that to us?

I’ll try to explain it to you. I’m not sure I have full command of it, and I’m not sure anyone does quite yet.

It goes to Treasury ultimately, but it was supposed to flow through a different process. It changed in the 1970s. In 1971, the Nixon administration came off the gold standard, which allowed deficits. In 1973, near the end of the Nixon presidency, Congress took away from the presidency the executive power called impoundment, which was the power of the executive to stop spending.

Nixon was abusing it by stopping spending he didn’t want, so Congress took it away from him. What that means today is that when the executive reaches in, it’s very hard to just stop payments.

The government is put in a process where an authorized executive would simply stamp a bill that got paid. That broke, and I don’t know when it broke. At some point, it broke. The money flow now works like this: A department gets a budget authorized by Congress, it goes to OMB, OMB allocates the budget, and that department then sends a money request to Treasury to pay it. It isn’t reconciled against what happened. That’s it.

There’s no controller function like there is in a normal company. There’s no control, there’s no controller, and there’s no reconciliation. The reason they can’t pass audits is that you can’t audit something you haven’t reconciled.

The only audit I’ve seen is from the Social Security Administration. When you read it, it’s riddled with material weaknesses. I’ve got one of our partners who has read the thing, and it makes you think there are individuals who have made millions or billions of dollars from mispayments overseas.

Is this what’s gone on with the missing money in Ukraine? Do you think it’s found its way to the wrong places?

I don’t know what happened in Ukraine. It’s a crazy story.

I do have business experience trying to work in businesses that had Medicaid and Medicare payments. I wanted to make that better. We wish to make the world better, and we just stopped because we found so much fraud. I’m certain of it.

We literally have a rule here: If a government payer in those areas is more than a third of the business, we don’t do it. In the services space, this is why we stopped. We found continuous fraud in the companies we were looking at investing in.

Just threading these things together, the really interesting thing with the Twitter pausing of payments was that, at some point, we were in a meeting at 1:00 a.m.

On a Saturday at 1:00 a.m., we were in a meeting and someone said, “Let’s turn the credit cards off to see what bounces.” Of course, we started getting calls. People began routing through obvious connections because they knew Sacks was there, you were there, and I was there. Someone would say, “A company that I share an investor with—or another company—said they’re not getting paid for this,” and it would turn out to be some SaaS software that nobody was using.

You start asking, “Is this software even being used?” There was so much software and so many services that had been paid for that nobody had ever logged in to set up. It was just being paid for as pure graft. Now, you said, Antonio, correctly, that the people who come first are probably the ones who are in on the biggest grift, because they figured out how to grift the money.

How USAID got to the top of the DOGE list is one of the most interesting aspects of this story. On January 21, Trump decided he would issue a bunch of executive orders, and one of them was to pause foreign aid for 90 days. That seemed reasonable: We have to take care of our country. That was part of his mandate for becoming the 47th president of the United States.

A couple of days later, the White House said, “USAID leadership is trying to circumvent the executive order.” In other words, they were going to keep paying people even though the executive order had come out. That alerted the DOGE team. Elon confirmed this on X. He said, “All DOGE did was check to see which federal organizations were violating the president’s executive orders the most. It turned out to be USAID.”

That became their focus. According to NBC, security officials at USAID tried to prevent DOGE from getting into the building or accessing the systems. To your point, Antonio, the person who probably has the most to hide is the one who’s going to fight the most.

Over the weekend, DOGE gained access to USAID, and then people started tweeting out all of this crazy spending. Any American who looks at it says, “Wait a second. If we haven’t fixed the water in Flint, Michigan, why are we sending money to Galway, Ireland, to do a DEI musical?” It makes no sense.

I think that’s the key piece of this, Antonio: Can you win the hearts and minds of Americans? Some of this—I don’t know if it’s legal or illegal, or against protocol or consistent with protocol—we’re going to find out. There have been tons of legal cases and lawsuits filed, but this is how Elon found out about USAID.

Jason, let me ask you a question. Where does that place you in your political philosophy around the importance of human rights abroad?

Great question. I’ve always felt that the West should act in unison, and if the United States is the greatest economy or the strongest country, we should lead. There is something called the Universal Declaration of Human Rights. The United States, with Eleanor Roosevelt, wrote this at the United Nations, and everybody tries to hit those notes.

I think it’s great that we try to reduce suffering in the world, but I think we should be doing it multilaterally, not unilaterally, and not to try to manipulate governments. That’s the piece of this. USAID is obviously a grift, where people are trying to steal money for their pet projects, and I don’t think they’re acting in unison or above board to ask, “Where’s the most suffering in the world, and how can we help?”

That seems like a noble thing to do. If the United States has the budget to help fight AIDS in Africa or reduce poverty, there’s a chance that some of these human rights issues were embellished to try to get more money.

I do think the answer is yes. I recently tweeted about this. Amnesty International, where I worked, was one of my first jobs. They were working on people who were imprisoned, tortured, raped, and murdered—systematic murder of dissidents—for freedom of speech, freedom of assembly, religion, and so on.

Somewhere along the line, Amnesty International started tweeting about trans rights, and that became the big focus. This is a very small number of people on the planet, and I don’t think those human rights violations in any way relate to the tragedy we’re seeing of people being murdered, tortured, raped, caned, and beheaded in certain places in the world.

There is a way to look at suffering and say, “We should handle this first.” Somebody who feels they were misgendered may be way down the priority list compared with the systematic rape of women in war zones. Those are the high-order bits.

The other fascinating thing about USAID, which I’d like to get your thoughts on too, Chamath, is at what point did this switch? The left’s position in the 1980s and ’90s, when we were growing up, was that we shouldn’t intervene. We should let democracy flow and let these countries figure it out for themselves. We shouldn’t be doing empire-building or imperialism.

Then, all of a sudden, it became such a grift that everybody had their hands in the pie. Lindsey Graham is involved in some nonprofit that gets money from this—I don’t know if that’s above board or not. A lot of people who were formerly in government seem to be part of this NGO train. The whole thing turned out to be a grift, and it’s obviously a bipartisan grift.

I think the biggest thing we’ll have to confront is that many of the things we thought were issues or problems may have actually been somewhat embellished because of this money cycle. That’s going to cause a lot of people to feel foolish about some of the decisions they made. All the attempts at cancellation are going to look really dumb in hindsight.

All right, nice segue to our friend David Sacks, who’s coming online. Okay, here he is.

David, the one thing I want to point out is that this goes way back, maybe more than a decade ago. David tried to explain to me that neocons had taken over American foreign policy. To your question about why the Democrats changed and why the Republicans changed, the neocons took over foreign policy on both sides—Democrats and Republicans—and populated it with this very activist, interventionist stance, which is bad for America. Jason, I think that answers your question as to why the Democrats shifted.

With us is David Sacks. How are you doing, brother? Are you literally in the White House?

I’m in the EOB.

There’s a podcast studio in the EOB?

There’s actually a podcast studio in the EOB.

Fantastic. Are you wearing a suit every day?

You’re right. I have to wear a suit and tie every day.

I was just listening to your conversation about DOGE. Jason, I’m surprised that you never figured out a way to get involved in USAID. Everybody’s on the take except you. What’s going on?

If I had known, I would have started an NGO. Where’s my NGO? You had everything except the money laundering. You had the grift and the virtue signaling. You had it all except the money.

Let me level this up for a second. We knew the U.S. government runs a $2 trillion deficit every year. We’re in debt almost $40 trillion, and we also knew that anytime anyone tries to cut anything in Washington, the whole city screams bloody murder.

Why? Now we know: The money is all going to them. It’s round-tripping to them. The New York Times is getting paid. Politico is getting paid. Bill Kristol, the perennial warmonger, is getting paid. Ukraine is getting paid. Eleven out of twelve publications are getting paid by USAID.

Viktor Orbán, who is the prime minister of Hungary, was saying that although he’s very popular in Hungary, his political opposition is funded by USAID. In Poland, the left-wing political opposition is funded by USAID. It goes on and on.

The BBC is getting paid. You wonder why everyone in the U.K. believes the BBC. Every left-wing organization in the world seems to be getting paid by this slush fund at USAID, which disperses about $50 billion a year. That’s more than $1 billion a week. That’s a lot of money.

It makes you wonder. The left in general tries to portray itself as a movement of the people, as grassroots. This is the exact opposite. This is astroturf. It’s basically money coming from the top down, out of Washington, to fund all these groups—not even necessarily in the United States, but all over the world.

It makes you wonder what the real level of local support is for these left-wing policies all over the world.

Crazy. You made a big announcement this week on crypto and creating a framework. I caught some of it. Maybe you could tell us, from the bottom up, what is the mandate from the president, and what is your advice to him on how to move crypto out of the shadows, offshore ICOs, and all that craziness into legitimacy? What’s the plan here to legitimize and regulate crypto?

The plan was really spelled out by President Trump in his week-one executive order on crypto. The principles are all spelled out there. The president said he wants to support the responsible use and growth of digital assets and blockchains across every sector of the economy.

Yesterday, I was invited up to Capitol Hill to meet with the chairmen of the important committees that will govern crypto. We had a conference there to announce the legislative plan. There’s Chairman Tim Scott, who is the chair of the Senate Banking Committee. To my right is French Hill, who is the chairman of the House Financial Services Committee. To his right is John Boozman, who is the chair of the Senate Agriculture Committee. To the left of Tim Scott is G.T. Thompson, who is the chair of the House Agriculture Committee. He’s out of frame right now.

Those are the 4 committees that govern crypto. You may ask why the Agriculture Committee is involved in crypto. The reason is that the Agriculture Committee supervises the CFTC, the Commodity Futures Trading Commission. Commodities all came out of agriculture.

It’s interesting: You need 4 committees across the House and Senate to get legislation done on crypto. It’s not just the House and Senate; it’s actually 2 committees in each chamber. This is the first time we’ve had all 4 chairmen of the 4 key committees come together and say that they’re ready to support crypto legislation.

There were a lot of people online who felt like this wasn’t a mind-blowing announcement. They wanted something they could trade on right away. That’s not what this was. This was a statement of commitment from the chairmen of the 4 committees that we’re going to get legislation done this year, maybe in the next 6 months. That’s really the goal, and we’ve never had that before. That’s pretty monumental.

I used to work in this area. When I first launched the Climate Corporation, we were actually selling commodity contracts online. We set ourselves up as an exempt commodity trading platform, so I remember the old legislation.

There was the Commodity Futures Modernization Act, if I remember correctly, when they deregulated the energy market. One of the features of that act was that they created the concept of an exempt commodity contract, where you’re not delivering a physical good. That’s basically what weather derivatives were, along with energy derivatives and other indices that didn’t have a tangible physical supply.

It was still shuffled into the commodities world. That’s why the legacy of all this stuff sits with Agriculture. The way this is likely to move forward is that it will look like a new extension of exempt commodities and be treated that way, rather than being treated as securities.

The question you’re describing is called market structure. What are the definitions going to be? Digital assets can be many things. Some digital assets are cryptocurrencies; they’re actually currencies. Then there are crypto securities. There are commodities. Bitcoin is regulated as a commodity right now.

Then you’ve got things that aren’t securities or commodities. They’re collectibles, like NFTs. There are all these different categories, and one of the things the market needs is clarity around the definitions so founders know the rules of the road and can comply with them.

The legislation would give them those definitions and describe how a crypto project could start, for example, as a security, and eventually the protocol could become decentralized enough that it becomes a commodity. That whole idea is called market structure.

There was a bill in the last Congress by French Hill, who is now the chair of the House Financial Services Committee. It passed the House with 71 Democratic votes, so it was fairly bipartisan. Then it went nowhere in the Senate because the Banking Committee at that time was run by Sherrod Brown, who was anti-crypto. It got stopped in the Senate right away.

Now we have Republican control of the Senate, and Tim Scott is the new chair of the Senate Banking Committee. He’s expressed support, so I think we could get a bill on market structure, like FIT21. That was French Hill’s bill last Congress. I think we could do a revised and updated version in this Congress, and that was one of the things all the chairmen expressed support for.

I think there’s a pretty good chance we could get this done in the next 6 months.

What’s the opposition, Sacks? It feels to me like, with the market-structure question being addressed and answered, you would also have more protection for consumers. Businesses would know the rules of the road, follow them, and have a structure that protects consumers. Why would people be opposed to moving this legislation forward and getting it all behind us?

I think this is an area where there’s a really good chance of having bipartisan support. We had it in the last Congress. The House bill got more than 70 Democratic votes, as I mentioned. In the Senate, we’re going to need 7 Democratic votes to get to 60, which is the number you need if you don’t go through the reconciliation process.

I think there’s a good chance this passes with significant Democratic and Republican support. It’s not going to be unanimous, because there are still forces that are hostile to crypto in Washington.

Do you think it’s going to be a discrete bill? It seems like you’re going to have to get a border-security, energy, and budget bill passed. Everything seems to be moving toward reconciliation. Would this be an add-on?

The question is what you can get through reconciliation. In order for a bill to go through the reconciliation process, where you only need basically 50 votes, it has to have a budgetary impact—or predominantly a budgetary impact. I think it’s called the Byrd Rule.

That rule was pushed pretty hard in the last administration. You remember that the Biden administration got the Inflation Reduction Act passed through reconciliation, along with all those subsidies for clean energy. They opened the window pretty wide on what can go through reconciliation.

There’s one other bill that I think is going to move pretty quickly here, too. I just mentioned the market-structure bill. The other area is stablecoins. Senator Hagerty, who is on the Banking Committee, just released a stablecoin bill. There are counterparts in the House.

What the 4 chairmen indicated is that they’re going to take up stablecoins first, and then market structure will follow very quickly. I think we could see a stablecoin bill pass Congress in the next several months.

The SEC and Gary Gensler were the blockers previously with crypto. They said, “There’s an existing set of rules here. Just follow those rules.” Obviously, those rules don’t exactly apply to the innovation happening in crypto.

Stablecoins feel like a layup and a great place to start. That would be an early win, and it would reinforce the dollar’s supremacy if stablecoins are tied to the dollar. That’s good for America.

Maybe you could talk about protecting consumers. We all saw, in the first couple of generations of crypto, all kinds of grifts and ICOs and things that were never delivered. How do you balance protecting consumers who may get really enthusiastic about this with preventing people from preying upon them?

The first thing you want to do if you’re going to protect consumers is bring the activity onshore. When all the activity gets driven offshore, it’s hard for regulators to supervise it. Moreover, it’s hard for the market to know who’s a good actor, who’s a bad actor, what’s a good project, and what’s a bad project.

The first thing you want to do is have the innovation happen onshore in the United States. It’s probably not a coincidence that the biggest fraud in the history of crypto, FTX, was based in the Bahamas.

That’s probably a little bit of a tell.

It will be an even stronger tell when the good projects feel like they can come back into the United States. Then you’ve got the shady ones in the Bahamas or other countries, and they’re going to stick out like a sore thumb. Everyone will understand that those guys are too shady to operate in the United States.

The number-one thing we need to do is bring the innovation onshore. In terms of the framework, I think the market-structure bill is going to define what’s a security, what’s a commodity, what you have to do, and the disclosure requirements around creating a crypto project. All of that will be in the bill.

In the meantime, the SEC has created a new task force under Hester Peirce, who is an SEC commissioner. She’s already starting to work on defining a better regime at the SEC for crypto projects.

You mentioned that Gensler said the SEC’s doors were open to crypto companies and that they should come in, talk to us, and work with us. That was very disingenuous. Crypto companies would tell me they would go see the SEC, and the SEC would tell them nothing about what the rules were. Instead, they would have enforcement people in those meetings just writing down everything they said. The next day, they would get sent a Wells notice.

The truth is that the SEC was not cooperating. They were not providing any clarity. They were just honeypotting founders to come in, and then they would immediately investigate them. It was really terrible.

We expect founders to play by the rules, abide by the law, and be compliant. But when you won’t tell them what the rules are and then you prosecute them, there’s no fair way for them to comply.

The most important thing is to give them a framework. I think the SEC is starting to do that already, and legislatively we’re going to have a bill moving through Congress over the next few months.

This is absolutely awesome. We’re super encouraged that you’re doing this, and we really appreciate you coming on the pod. We wish you could participate in the other 3 or 4 crazy discussions we’re about to have, but we understand that you’ve got to stay focused on the mission.

Anything on the AI front that we can look forward to in the coming weeks? I know that’s the other part of your mission.

The big thing is, as I talked about last time, that the president rescinded the Biden executive order. It was this 100-page monstrosity of burdensome regulation on our AI companies. I think that decision has been proven even more right in the wake of DeepSeek, because we know that China has basically caught up, or is very close to catching up.

It felt like the Biden executive order was written in a vacuum in which the United States was the only player in AI, and that if we imposed a bunch of burdensome rules on our companies, somehow that wouldn’t allow China to catch up.

It’s pretty clear that China is very competitive. If we burden AI companies, it’s going to benefit China. I think that was a very good decision. What the president said in his executive order is that we should devise a new AI action plan to replace the Biden executive order, and we’re working on that right now.

Any fun anecdotes from DOGE you want to share?

I’ll give you one anecdote. I’ve been working late here a number of nights at the EOB. I won’t tell you where the DOGE guys are based, but I know where their office is, so I went by there to say hi.

The whole room was full of young coders. I think they were engineers, but they were wearing suits and ties, so that was a little different. They were all working really late. They were working late on Friday night, and the facilities people didn’t know what to do because they’d never had people ask to stay late before.

They had to create new facilities access for these guys. They were saying, “You’re coming to the office and doing work? We don’t have a protocol for that. How does that work? We’re going to need to get you a badge to come into the building.”

That’s absolutely awesome. It’s great to see the progress you’re making in the first 2 weeks. Continued success. We’re so proud of the effort.

Thanks, guys.

All right.

Do you guys think the Democrats are going to lose people over their opposition to DOGE? Is DOGE really viewed as oppositional to Democratic Party interests? For the average person, it’s a war.

I think there was a coalition that the Democrats had, and there was a coalition that the Republicans had. The Republicans did a better job of reforming their coalition.

I actually got this totally wrong. I don’t know if you remember, but a couple of years ago my thought at the time was that if the Republicans didn’t figure out how to fix themselves, they were going to lose for the next 10 or 15 years. The reason I said that was that they would walk into every midterm and get their asses handed to them.

I think they figured it out. This is something I’ve been thinking about a lot: There’s a fight in Western societies, and it’s a pendulum between labor and capital. The conventional wisdom used to be that Republicans were pro-capital and Democrats were pro-labor.

The brilliance of Trump is that he took over the Republican Party and made it totally populist, which is to say pro-labor. The crazy thing about the Democrats is that they are the most sophisticated liars, because if you look at what happened under Biden, you had record-high stock markets, which was purely in favor of asset owners; record-high deficits; record-high illegal immigration; and record-high wage suppression.

All of these things are massively pro-capital, but they tried to present themselves as pro-labor. That entire ruse is now being undone.

All of this data will consolidate the Democrats to a shell of their former selves. It will take a year or 18 months, but unless they figure out how to totally hard-reset, they’re going to be in a really difficult struggle to find a cohort of people beyond 15% or 20% of the population for a long time.

It’s so dumb to come out against waste, fraud, and abuse. The best argument the Democrats had, it seems, was that people’s Social Security numbers or privacy were being violated because DOGE went in and looked at the data.

That is the height of not getting the point and not reading the room. One hundred percent of Americans don’t want their tax payments stolen. They don’t care if you looked at their Social Security number. This isn’t a privacy issue because DOGE is looking at some database.

The issue is how much money is being stolen from the American public. Anyone could have picked up this issue at any point in time. This has been going on for years. I think you pointed out last time that the last person who really addressed this in earnest was Clinton.

This has been going on under Obama, Trump 1.0, and Biden. Everybody has been raising the debt. All of this grift has been going on, and it’s only this time around that somebody picked up the free money and said, “Here’s an issue. Stop wasting money.”

We now see what happens when somebody picks up the issue of stopping wasteful spending: It’s a popular issue. This is only going to make Trump more popular.

Jason, I would add to what you and Chamath said. I think you’re both right. I’ll give you a very concrete example. Rahm Emanuel is now back from Japan. He was chief of staff in both Democratic White Houses after Clinton and Obama.

He wrote an op-ed this past weekend that basically said the Democrats have lost their way because they’ve forgotten what he calls kitchen-table issues—the things regular people care about.

Chamath is right: They forgot about inflation. Inflation is terrible for the average person. It’s okay for people who own productive assets because those assets go up in value, but it’s terrible for wage earners and people who have savings. It’s terrible for older people living off their savings.

Rahm makes the point that if the Democrats are going to reform themselves because of all these other issues, they have no home and no base from which to build right now unless they go through a great reset.

Part of that is understanding that they have not actually been pro-labor. They have been pro-capital. But that requires such a schism from the deepest believers in the Democratic Party, who thought, “Eat the rich”—they even wore it on their dresses. It was so important to them. In fact, they were feeding the rich, and they didn’t even know it. That’s pathetic.

It’s funny that Margaret Thatcher famously said the problem with socialism is that you eventually run out of other people’s money and run deficits. You destroy the country. This happened in Venezuela. This is always the end of socialism. It’s how it finishes. It’s how the movie ends. We’ve seen it around the world. Look at South America.

We were heading in that direction. I said earlier that I was afraid we might become a kleptocracy if this didn’t stop. I’m grateful to all the great patriots of DOGE, to the government, and to the president for making it happen, because we were heading in that direction.

The Democratic Party is lost. They’ll continue to be lost.

An interesting thing came up this week. On Monday, President Trump signed an executive order laying out a plan to establish the first sovereign wealth fund for the United States.

For those who don’t know, a sovereign wealth fund is essentially an investment fund for a country. It’s almost universally based on natural resources. Norway, Saudi Arabia, and the UAE all have them. Australia’s sovereign wealth funds are based on minerals or, typically, oil.

The United States isn’t known for having the oil reserves of Saudi Arabia, the UAE, or Norway. This public investment fund would apparently be anchored potentially by the TikTok shares that Trump said he wanted to get 50% of by giving it a license. A lot of that is unclear, since this license has never existed, but that’s the concept.

The Treasury secretary and Commerce secretary have been tasked with developing a plan over the next 90 days. The plan should include recommendations for funding mechanisms, investment strategies, fund structure, and a governance model.

Chamath, you were tweeting about this. What’s the point of having a sovereign wealth fund in the United States if we’re $36 trillion in debt? Shouldn’t we just pay the debt down? Where is this money going to come from?

It’s not an either-or thing. If there are assets that are effectively minted overnight, which I think a 50% share of TikTok would be—call it $150 billion—the question is what you should do with it and who should govern it.

I like the idea of having a group of 5 elder statesmen. I’m just going to throw some names out there: David Tepper, Stan Druckenmiller, Ken Griffin, John Doerr, Mike Moritz, Bill Gross, or some other bond guy. My point is that you get 5 people who are very sophisticated across all market categories.

One of them could be the rotating CEO for some number of years. People should rotate in and out. These would be unpaid jobs, because everybody who has these slots should be mega-billionaires. They shouldn’t be doing it for their own personal advancement.

They should deploy that capital so that, as you sell down the TikTok shares, or as you sell federal lands and generate more oil revenue, you take it all and invest it on behalf of America into American companies. I think that’s an incredible idea.

Antonio, do you think the government should be in this business?

I think it should, and I agree with Chamath that governance is very important. He has a good idea there.

I think about it for a different reason: We don’t have an industrial policy in America. Many of our competitors around the world, particularly China, have a long-term industrial policy, and they put enormous amounts of capital behind it.

A sovereign wealth fund would be a stealthy way to create an industrial policy in America.

What do you mean by industrial policy in this context?

In China, they want to build chip fabs and catch up to TSMC. What do they do? They take the dollars from the trade surplus they get from the United States every year and pour them back into making that stuff inside their country.

For decades, part of the problem we’ve had with China is that capital is free because the banking system just pushes money out to manufacturers. Manufacturing moved from the United States to China, in part because of the WTO.

When people say the Chinese have a long-term, 100-year vision, the way it manifests is through this industrial policy. We don’t have that here at all. We try to do it with things like the CHIPS Act, which goes through Commerce, and we have government bureaucrats deciding how to spend $200 billion to modernize Intel. That needs to happen, but I would much rather have Ken Griffin, Bill Gross, or any of the people Chamath mentioned deciding, “We have 5 industries in America that we want to invest in. Let’s make great investments for America.”

They have to be economic. They have to make money for us, but they also have to be good for the country. I think that’s what most of these sovereign wealth funds do.

Some of them make portfolio investments, but many of them—like the sovereign wealth fund in Saudi Arabia, the Public Investment Fund—are making enormous domestic investments to remake the economy toward tourism, for example.

They’re literally building cities in NEOM. They have a dozen cities being constructed and are trying to take an oil economy and shift it to a tourism economy, a technology economy, and a private-equity economy. I think this is a great idea.

Antonio is totally right. Part of Scott Bessent’s congressional testimony—you guys probably saw this—was that he laid out what he calls the 3-3-3 plan.

The 3-3-3 plan says we want GDP growth of 3%, deficits no greater than 3%, and 3 million barrels of oil produced domestically in the United States.

If you double-click on that and look at the total energy reserves in the United States, they’re 3 times greater than the total energy reserves of Saudi Arabia—3 times, across all forms. Not just oil: oil, gas, and coal.

If we actually move, as Antonio said, toward an industrial policy that’s pro-energy, where the incremental cost of energy is effectively zero, where we want a gross abundance of electrons flowing through America for all the great ideas that could pop up, it will by definition generate an enormous amount of revenue for the federal government.

Having the sovereign wealth fund be the rainy-day fund that can bank a percentage of all of that starts to do a lot of good for the long-term strategic guidance of the United States.

Friedberg, what do you think? Is this a situation where we’re making the government too big and now we have the government competing with BlackRock, Sequoia, and Andreessen Horowitz? They’re going to be on the boards of technology and energy companies and investing in them.

What happens when Obama, Biden, Trump, or Bush puts different people in charge of these things and they want to do their pet projects? It seems to me like this could get awfully conflicted, awfully quickly. Should this be a business, or should the government be in it?

I think one of the things the government is bad at is capitalism, so I wouldn’t make capitalism the mandate of a sovereign wealth fund—certainly not when we have $4 trillion in federal debt. It doesn’t make economic sense.

Our cost of capital is 5%. That’s how much the interest is on this 10- to 30-year debt right now. It’s very hard to make real, risk-adjusted returns when our cost of capital is so high.

I don’t think the mandate should be, “Put a bunch of capital in a pool, go invest it, and try to make money for the United States.” That seems silly.

My point about the government being really bad at capital is that the United States government owns, has access to, and will acquire through other means significant assets and resources that should be monetized in a smarter way.

I would think about the sovereign wealth fund as more of a strategic vehicle for monetizing high-value government assets. For example, if Trump actually negotiates a 50% equity position in TikTok U.S., that needs to sit somewhere. It should not sit in the Department of Commerce or some other department. It should sit with a capitalist manager who ultimately makes the decision about when and how to monetize that asset, return the cash to the Treasury, and pay down the debt.

Similarly, the United States has large amounts of land and access to other large assets that get transferred through seizures. There have been Bitcoin seizures over the years as the government has cracked down on criminal enterprises. The government owns this Bitcoin.

Do you think the smartest people are making the decisions about where and how to sell down that Bitcoin? I guarantee you they’re not. I would much rather have a capitalist making that decision.

I would view the sovereign wealth fund as less about raising capital through other means—borrowing money through Treasuries, given the debt level we have—and trying to invest it. It should be much more about what strategic assets the U.S. government can monetize and using this as the mechanism for doing so. Ultimately, I think the objective should be to return that capital.

I also think there’s an opportunity to manage Social Security in a smarter way. Social Security is functionally going to be bankrupt in 8 years, given the way the trust is set up, the cash and assets that are there, and the demands on Social Security from an aging population and rising payouts every year.

Another way to think about this is: What are the long-term debt obligations, which is ultimately the point of these sovereign wealth funds, and can they be invested in a smarter way? Why is the Social Security entity ultimately owning 3%-yielding bonds when it could own interests in equities?

We seized 144,000 Bitcoin, and that’s sitting somewhere in the Department of Justice.

They sold it.

I think we still have it. I thought the idea was that it was going to become the Bitcoin Strategic Reserve.

You could also ask: The DOJ sells these assets, so who do you think at the DOJ is making the Bitcoin market decisions? Is that the right person to be monetizing these assets?

David made an interesting point about Social Security that I want to follow up on. When you think about Social Security, it’s $6 trillion of our $36 trillion debt, and it’s actually a fake Treasury bill.

One of the things we figured out in the early days before the inauguration was that the $6 trillion sits on the ledger as just a paper ledger. If it gets paid out, it will also be very inflationary. It’s essentially a fake Treasury that gets rolled over and pays a very low rate.

It would be much better to have that invested in an economically productive way for people with real money. By the way, the Social Security Administration is the only thing we’ve found in the government that’s audited. It has an audit, and when you go through the audit, it’s crazy—the material weaknesses you find.

It would be better to have that invested in a way that was economically productive for people with real money.

As we’re speaking, the federal judge just put a temporary restraining order on DOGE and barred Elon and his team from accessing U.S. Treasury payment data.

So now we’re going to have a real grudge match between the public and the federal government.

Which judge was it?

The federal judge.

Do you know why, Chamath? Does it say why?

No, but Elizabeth Warren is doing a victory lap, so I’m sure she’s part of it.

The government agency set up by the president can’t look at the Treasury database. To double-click here, DOGE is not a new agency. It’s the renaming of an existing agency—I can’t remember the name of it—that was set up under Obama to accredit an audit function to the government. Elon tweeted about this several weeks ago.

What’s crazy about this ruling to me is that Congress delegates to the executive branch the ability to spend money. Congress appropriates the money, and the executive branch spends it. How can you spend money if you don’t know where it went? How can you be responsible? You have the authority to spend it, but not the authority to examine it.

Guys, this is flying fast and furious. President Trump just unveiled his framework for his tax plan: no tax on tips, no tax on senior Social Security, no tax on overtime pay, renewal of the middle-class tax cuts, and an adjustment to the SALT cap.

Again, it’s very pro-labor and populist. He also wants to eliminate special tax breaks for billionaire sports-team owners.

I’m no longer supportive. I already sold my team. I already sold my fees.

Close the carried-interest tax loophole, which allowed you to claim carried interest as—

Sorry, Antonio, I’d better run.

I’ll be back in a few hours.

I have to sell some things.

Isn’t that incredible? He’s really going for the jugular.

I support this 100%.

The SALT deduction is coming back, is that right?

Marginally. I don’t think he’s going to give it back the way it was before.

Isn’t this incredible? Who’s going to stand up and lie down on the railroad tracks for being able to amortize a multibillion-dollar sports-team purchase, or for getting long-term capital-gains treatment when you make a fund investment? Who’s going to be that person in this administration? Nobody’s going to stand up for these things.

I think we should stop doing venture capital and start NGOs. We should start a whole rat’s nest of NGOs to ship money around. You could have an incubator.

Absolutely. If you’ve got a great idea for an NGO, why didn’t you launch one in every random developing country in the world?

If I had known, you would have had $8 billion of assets under management. Just launch one in Vietnam. We could have had a launch accelerator in Vietnam for transgender people.

The Jason Calacanis Launch Fund of Equatorial Guinea.

It’s so crazy to watch the Democrats self-own by coming out and saying we have to stop people from stopping wasteful spending. They don’t seem to understand how unpopular these are as kitchen-table issues. Everyone cares about that. If you care about fringe issues and not the things people care about at the kitchen table, you’re missing the point.

If you think about the transgender sports issue—biological males playing in female sports leagues—that issue Trump just issued an executive order on yesterday was such an obvious issue of fairness. It has nothing to do with transgender people; it has to do with biology.

If a biological male plays basketball on a biological female team, somebody is going to get hurt, and that person is going to score 100 points. It’s obvious. The fact that the Democrats couldn’t see that issue as being 100% or 95% popular makes no sense. It’s such an obvious litmus test of logic.

The Constitution is a document about fairness. The people who founded this country did it because they were being treated unfairly at home. All of us here are 1 or 2 generations away from immigration, and the reality is that’s why people come here.

It’s un-American.

We have 2 people here who are immigrants.

I’m reading the temporary restraining order, and it looks like it narrows who from DOGE can have read access down to 2 people: Tom Krause and Marko Elez.

They don’t want the whole team to be able to see what’s at Treasury. They want a process and some sign-off.

That doesn’t sound unreasonable. I just want to point out that no matter how good they are, I’m sure they’re working 24/7. Looking at all the payments in the U.S. government is not a small task.

It does seem like it needs to be more than 2 people, but it doesn’t seem like they’re saying you can’t look at it. They just want to have some controls in place.

It feels like an administrative block. It’s the narrowest thing you could do: “You can come in, but you can send only 2 people, and only these 2 people.” If they get sick or something happens, no one else can come in. That seems to be a mystery block designed to slow it down.

We’re interpreting this in real time. The facts will come out over time.

Google dropped 7% after reporting earnings on Tuesday. Revenue was up 12% year over year to $96.5 billion. Cloud revenue was up 30% year over year to $12 billion. YouTube ads surged 14% to $10.5 billion. Net profit was around $26.5 billion, up 28% year over year.

They’re clearly focusing on profitability. Full-year numbers are extraordinary: total revenue of $350 billion with $100 billion in net profit. Cloud and YouTube finished 2024 with a combined run rate of $110 billion. YouTube is basically Netflix inside Google, and Google Cloud is essentially AWS inside Google.

The thing that concerned investors, Chamath, was that Google said it would invest $75 billion in capital expenditures in 2025—a 42% jump over 2024 and 29% more than analysts expected. Obviously, this has to do with data-center servers and the AI buildout.

What do you think about that number? Is this an absolute waste of money, or is it a gargantuan number that Google can easily absorb and use in the future because of its profitability and cash flow?

I think I should stipulate that Google’s models are probably the best of all the models across a broad base of capabilities. If you test them, that’s probably where you land. Let’s start there: Whatever they’re doing is working.

The thing they need to do is translate those models into better products. I think that will happen slowly. If you look at Deep Research, most people online who are evaluating it would now say that OpenAI is both faster and better on the margins.

All of these things can be improved. I don’t think that’s a comment on the base model. I think it’s a comment on post-training and how they’re attempting to productize these things.

The other thing Google has is a money machine that directly benefits from these AI-driven optimizations on ad targeting. The only other company that has anywhere close to the same credibility is Meta.

What both of these companies need to do is a better job of explaining how that $75 billion is segregated. How much goes to the AI-enabled models that actually improve ad optimization?

There was a really interesting discussion by a former Meta machine-learning engineer on X about how they did it. It’s pretty amazing and staggering. If Google could say half the money goes to that and the other half goes to more speculative pre-training and post-training, I think the market would have eaten it up.

It’s probably more of a disclosure issue for Google. I would say that right now, their model quality is the best.

Do you have thoughts on this buildout, Antonio? You’ve obviously been involved with xAI and with Twitter and X previously, as well as with Colossus’s buildout, which was extraordinary to watch over a period of time.

Do you have concerns, as some people do, that this buildout is too expensive and would be too hard to monetize? Or is the opportunity in AI so obviously huge that you have to take the leap of faith and say, “If you build it, the revenue will come”?

I think Chamath has the right framework here, which is return on invested capital. What he’s saying is that if Google had said half the money is for ads and half the money is for much-needed post-training, the market would have seen that the first half has a higher return on invested capital.

Google’s return on invested capital is going up after the implementation of AI models into the company. That sort of abstracts the entire market. People are waking up to the fact that return on invested capital in data centers will matter.

The models are basically commodities. They’re super-competitive. In the best case, it’s a land war in Asia; in the worst case, it’s a melee and total commoditization.

What matters is the return on capital in the data center, which is influenced by how good the data centers are. You mentioned the xAI data center. It’s a 100,000-GPU cluster, the densest coherent cluster in the world. It will train faster and better than other clusters, and it was built more cheaply and quickly.

xAI will have the highest return on training capital, and I also think the best training data, so it will win. I think Google will also win because it has Tensor Processing Units, makes some of its own chips, focuses on R&D, and has a great monopoly to fund it all.

I don’t think it’s overblown. I think this is going to be, as you guys have said before on the podcast, bigger than the Industrial Revolution. But you also need to have good ROIC, and if you’re not transparent about it, you can see what happens.

I would argue that Google has probably been the most frugal, thoughtful, and well-managed computing-infrastructure investor of all time. In the 1998-to-2005 era, Google was all about cheap, throwaway racks. That was its big advantage: It wasn’t using expensive Oracle servers. It had a 2- to 3-year depreciation timeline on those, but they were super cheap, so the ROIC was quite good.

Then Google got into energy efficiency and realized that energy was a major cost driver. It started building systems that were more energy-efficient. As a result, they lasted longer, and the depreciation schedule moved to 3 to 4 years, meaning Google could write down the value of the servers over 3 to 4 years.

From 2010 to 2015, its hardware system for cloud allowed it to extend the life of the infrastructure through repurposing and increase the depreciation schedule from 4 to 5 years. In 2021, Google made a big change to its depreciation schedule on data-center infrastructure to 6 years.

When it invests capital expenditures in a data center, it writes down the servers over a 6-year cycle because of AI optimization on maintenance. It started using AI for the internal management of the infrastructure.

I would view the $75 billion in capital expenditures as a very positive signal for the company. It means they have a strong line of sight on how they’re going to get full utilization and a great return on this investment.

If you do the ROIC math on $75 billion and assume a 20% ROIC, you’re generating roughly $15 billion of incremental profit a year, plus the amortization of the $75 billion. Take $75 billion divided by 6, which is $12 billion, plus $15 billion. They would need to make approximately $27 billion of incremental operating profit a year on the $75 billion for this to meet their ROIC performance.

That doesn’t seem crazy. It’s just under 20% of their annual operating profit. Google doesn’t make these investments only to build out AI in the future. It has a strong line of sight on how this can increment, and the hurdle for it to pay back isn’t huge.

I would take Antonio’s point and view this as a positive. If you use Google’s historical ability to manage infrastructure and make predictions on investments as an indicator of the future, this is a strong and positive indicator.

For all the naysayers who think search is going to evolve into chat, you could look at this as an important proof point that Google has confidence it will be able to move from search to chat. As Chamath points out, it has great, performant models. I would view this as more positive than negative. If Google were underinvesting, I would be worried that it didn’t know where to invest. Seeing this degree of investment highlights its confidence in the strategy.

Antonio, I don’t know if you’ve been watching the agent space or Deep Research from Google and OpenAI. OpenAI launched its copycat version of the product.

I’m curious about your thoughts on job displacement. We’re looking at self-driving. Waymo has cars on the road, and Tesla—which you were previously on the board of, and where you were the first institutional investor—is making progress. Self-driving is pretty good. I only get 1 or 2 disengagements per hour, and they tend to be the ones where I want to take a turn a little more sharply.

That’s getting pretty close. BYD is very close. You have a lot of potential job displacement. Millions and millions of drivers could lose their jobs in 10 years. Researchers working at Gartner or Boston Consulting Group could be displaced. It feels like there could be massive job displacement.

Do you have concerns about that in the American economy and globally?

There’s a lot of hand-wringing about this, and it’s real. In prior moments of labor disruption, there have been job displacements, because generally people get retrained for something else. But you can’t always retrain people.

There were studies that Monfre did of the steel industry in Pittsburgh in the 1970s. The cost of each steel-worker job loss was about $1 million to the economy because the people couldn’t be retrained. That might happen here.

I have a more benign outlook personally. I think there will be job loss, but the amount of productivity released in the U.S. economy is going to be extraordinary.

GDP growth is a function of the number of people working times productivity. It’s very simple. Economists want to make it more complex than that, but it’s the number of people working times productivity.

If productivity goes up to 5%, 6%, 7%, or 8%, you get a massive boost in GDP growth. Then what happens? People can get retrained. They can get different jobs. Different services emerge. People start companies.

The application layer of large language models is just starting. The barrier to starting a company is quite low. Things like Launch probably explode because there are all these people who no longer have jobs—accountants, for example—who want to start something.

It’s hard to know what happens. I believe in American agility. I believe in this country, and I believe we’ll figure out a way to make this all work. If there’s enough productivity and money flowing through the economy, people will find new jobs, start new businesses, and find new things to do.

We have to get out of the way, take down regulation, let Americans be creative, and unleash the American productivity machine. Let’s make that happen.

That’s really strong. That’s the game we’re seeing on the field. How many companies are we seeing hit $1 million in revenue with 5 employees, where that used to take 25? The efficiency is there.

Chamath, what’s your outlook on this issue? It keeps coming up, specifically with truck drivers, Uber drivers, and all kinds of research jobs that seem to be done pretty well by AI. Where do you stand on this issue today?

I think it’s a process. Buffett wrote about this in an annual letter. He described the changing nature of jobs during the Agrarian Revolution.

You saw a large cohort of people who supported themselves through farming, and then the total number of those jobs shrank by 90% when you had industrialized farming, tractors, and so on. But the economy grew around that business and added other kinds of businesses that didn’t make sense in that moment.

When economies get more evolved, you see the growth of service businesses that can only happen when you have excess. The person you pay for closet organization would not have had a job at the turn of the Agrarian Revolution or the Industrial Revolution, but they can exist in 2025. Frankly, they can make a good living.

There are life coaches, podcasters, influencers, venture capitalists, venture capital allocators, and mutant strawberry creators. These jobs come out of nowhere.

I think we’re waiting for this next turn of creativity. The big problem we have—maybe I’ll take the more glass-half-empty version of what Antonio said—is that we really haven’t been unlocking people’s creativity over the last 15 years.

Aside from a few companies that we all know and can repeat endlessly, the ones that are truly innovating and operating at the edge are rare. Everybody else is just diddling around.

The real problem is that we haven’t had many repetitions of being creative. I’ll give you a simple example. Nick, can you find this? It’s in extremely state industries. Did you see the BYD clip of the car that parks itself with a little swipe?

Yes, with a little swipe.

How tragic is it that you look at this and you’re blown away? You’re blown away for 2 reasons. First, I didn’t think it was possible. Second, why doesn’t it exist in America?

The reality underneath the hood is extremely benign. That’s how it’s implemented. I think the problem is that we’ve spent so much time losing the script.

Antonio is right: When you unshackle people so they don’t have to focus on the mental load of getting the pronouns right or this or that, you won’t be so overwhelmed by things like this because you’ll already have been pushing the boundaries of human creativity.

We need to get back to that. We need to let creative people cook. I hope that happens, and I think it will happen.

Friedberg, can you do a science corner on the FDA approval for the non-opioid painkiller?

Not today.

No, I know. Could you do it in a couple of weeks?

The one I wanted to do is the new macro study on GLP-1s, which I think is super interesting.

Tell us.

You guys remember I talked a while ago about how they were able to mine VA data. The VA takes care of veterans, and it has all their medical records. On an anonymized basis, it can make that data available to researchers.

This is how they identified that the Epstein-Barr virus—the virus that causes mono—is statistically certain to be the trigger for multiple sclerosis in a cohort of hundreds of thousands of patients in the data set. No one got MS who didn’t get Epstein-Barr virus. If you didn’t get Epstein-Barr virus, you didn’t get MS.

The data set you can mine at the VA is incredible. Researchers pulled all the data from everyone who had been on GLP-1 agonists and identified all the health effects across multiple indications and the statistical differences between the cohorts.

The research team out of St. Louis pulled all the data from the VA database. They looked at 1.2 million people with diabetes who didn’t take anything, compared with 215,000 who took GLP-1 receptor agonists and another 600,000 who took other diabetes drugs.

This cohort segmentation allows them to isolate the effect of the GLP-1 drugs. As you can see here, this shows, across hundreds of thousands of patients, the effect of GLP-1 on a hazard ratio—which means how likely you are to have the following health condition versus the population that isn’t taking GLP-1s.

On the right are the increases in risk, and on the left are the things that go down. The only things that increased were nausea and vomiting, by around 8% or 10%; musculoskeletal complications; GERD, which is gastric reflux; and sleep disturbances related to indigestion. It’s all abdominal stuff.

On the benefit side, you see a decrease in shock, hepatic failure, respiratory failure, and cardiac arrest. For cardiac arrest, you see a 30% decrease in the probability of having cardiac arrest in the cohort taking GLP-1s. There was also a decrease in schizophrenia.

This goes to the point that the CEO of Eli Lilly made when I interviewed him a couple of months ago: They have clinical trials going on for different indications for GLP-1 receptor agonists, and they’re seeing health benefits beyond weight loss, including reductions in kidney disease, liver problems, mental-health problems, and so on.

Do we know why? If we don’t know why, do you think it’s because the drug suppresses food intake, and the lack of food or the change in food consumption creates this? Do you know what I’m asking? Do you think the drug itself is actually causing the effect?

You should watch the interview I did with RS. I think he highlights that this class of drugs turns genes on and off. There’s a gene-expression cascade that occurs with certain compounds.

We know that the GLP-1 receptor agonist binds to GLP-1 receptors, and there’s a cascading effect of genes that then get expressed. What that seems to do is turn off things like inflammatory markers and turn on things like C2 genes, which can actually increase cellular repair.

There seem to be other benefits from these drugs beyond appetite control. It’s not the appetite control itself; there seem to be other effects.

Let me ask you a question. Given these receptors being activated, would you put your kids on this?

No.

Would you put your wife on this?

I would consider it. I would consider it for myself, too, just for the anti-inflammatory effects.

How would you make that decision?

For me personally, the things I weigh against it are muscle loss and bone-density loss. The biggest downside effect is that you should increase the protein in your diet and do weightlifting. There are things you can do.

Frankly, if you do those things anyway—if you increase protein in your diet and do more weightlifting—you’re going to see very good health benefits from those actions alone. They may actually outweigh the benefits you get from the drug.

I have a question, Dave. When you look at that data and talk to the CEOs, how much do you think, in the long term, when the long-term studies are out, it will show that it was the drug, versus the fact that being obese is very bad for you?

When you take your body fat down dramatically, all these other gene expressions happen anyway. Which one do you think it will be?

This is what they’re starting to isolate. They’re starting to see that there are other expressions that aren’t related to obesity, including in people who aren’t obese and are using these drugs.

That cohort data is emerging now. Phase 2 clinical studies were published, and I think we’re going to see Phase 3 studies and some of these indications soon. It’s looking very positive that it’s not just the loss of obesity.

To your point, being obese, not exercising, and eating poorly destroys your health. If you stop doing that, everything gets better.

Right. If you do a GLP-1 receptor agonist, let us know.

I will. Right now, I feel like I want to go through a process of increasing my weightlifting routine first. I’ve been trying to create a more rigorous schedule, but my schedule is terrible. That’s been the hardest thing for me.

I want to go through that first before making the decision. I don’t want to confound the 2 factors.

You know what I did that made it super easy for me? I got egg whites in a carton, and I have this incredible crunchy, spicy garlic thing that everybody makes and is crazy about. In the mornings, I’ll eat 10 or 12 ounces of egg whites with that spicy stuff. It’s delicious.

I try to get 30 or 40 grams of protein first thing in the morning, and then I do the walking.

This is easy. Just wear a 35-pound weight vest, Friedberg, and walk a mile or 2.

My problem, Jason, with all of this is that every time I see something, I see a new protocol. I saw Gary Brecka on the Shawn Ryan podcast recently, and Shawn asked him for a handful of things he recommends for everybody.

He recommends mineral salt, a methylated vitamin, amino acids, and a whole protocol. Then, if you happen to catch a clip of Andrew Huberman on X, he has a protocol. Bryan Johnson has a protocol.

The problem is that all these protocols are slightly the same but different enough to create a huge cognitive load for a normal person like me. You’re busy, you have a job, and you have kids. How do you decide?

I’d really love something that’s not necessarily a gold standard, but that tells you what the real bang for your buck is. Are you better off just losing 50 pounds? This is why these products are successful, and I think they’ll continue to grow dramatically. It’s a pill, and it solves all those problems. It doesn’t require a cognitive load.

It will be in pill format soon, right? The pills are almost here.

It would be very cool, actually, Dave, if you did this. To Chamath’s point about people being confused, you could do an everyman’s story around this journey and document it.

You could say, “This is my weight. I did weightlifting first, and then I did the GLP-1,” and do a weekly check-in. Even if it were 10 minutes on X, you could give people the journey in a way that wasn’t so complicated.

I think people are confused.

I have very good doctors. You guys are good doctors, but if you don’t have good doctors, you don’t know what to do.

I had a doctor in Los Angeles and a doctor in San Francisco. I would have them do their own versions of things, and then I would have somebody else help me compare them. It cost me way too much money, and all that complexity actually made the quality of my healthcare go down.

What I really wanted was a simple protocol that said, “Take metformin because it’s good for you. Take vitamin D. Take omega-3 fatty acids. Otherwise, just eat this meal plan.”

That would help me a lot more than having to cobble it together myself. When I see someone like Gary Brecka, who is very articulate and very smart, the first thing I do is populate an Amazon cart with everything he mentioned. My instinct is, “I should do the right thing for myself. This is a couple hundred bucks, so it’s worth the investment.”

Then the next day somebody else says something else.

You have a little OCD, though, Chamath. I’ve known you for a long time. You get very obsessive.

My father died because of poor health, and my best friend died of poor health. I feel like you should at least do the preventable things.

Have you guys seen that post Chamath did? He was half naked in the mirror.

I look great. What are you talking about?

He looks great. If you could look like that, you’d do it too.

No, no, I’m talking about the glucose monitor. I was sitting with him at the poker table. He had the glucose monitor, took a sip of wine, checked his glucose, had a piece of beef, and checked the glucose monitor again.

It literally becomes obsessive.

Come on.

What’s so funny about this picture is that all these clowns on the internet don’t understand that when you’re 6 feet 2 inches, those are big legs.

They’re not big legs.

When you’re a short king, when you’re 5 feet 7 or 5 feet 8, I get it. You guys are all stubby and short.

This is the problem with generative AI. You can tell it’s a fake photo. Nobody has legs that thin. How could you have biceps like that and then legs that thin? It doesn’t make sense.

Unbelievable. What a thirst trap. Now you guys are going to find photos of Antonio and me.