OpenAI is essentially like a zombie child that was formed by all of the big tech companies that have become systemic to the stock market. And it's like Sam Altman is trying to make the company as systemic as possible such that eventually when things go wrong we have no choice but to bail them out. It seems like this is almost all part of the design that he wants to be too big to fail. He wants for the US government to own a 5% stake in it so that if anything ever goes wrong, why wouldn't the US government figure out a way to bail them out?
Today's number, 26.2 million. That's how much the citizens of Atlanta spent on OnlyFans last year. The most of any US city. Ed, why did the Scarecrow win an award?
Because he was outstanding in his field. So, in case you will need to be caught up, I'm going to tell dad jokes until Michael Semblat, the chief investment officer from JP Morgan, accepts our apology for offending him for telling a dirty joke on the episode he appeared.
I actually find the dad jokes funnier than the sex jokes.
I'm sorry. We need to fix that. Why can't you hear a pterodactyl in the bathroom?
Because the P is silent. Michael, accept our apology, or don't release the dick jokes.
Actually, that's my new position. I don't want him to accept it. I like this new direction.
I don't know about that because that was a pretty shocking performance. This will come out on Monday and we're recording before Sunday night. So, we'll see if England's still in the World Cup. But yeah, the Congo performance was not great.
Do they have to play in Mexico? They have to play is the next game Mexico.
Yeah, it's going to be a tough one. Did you see Team USA? Team USA looks good.
The scary part is the star of the US team, Feren, I think it's Balogun, received a controversial red card.
Did you see that? That was a, I think that's very anti-American. I don't... And by the way, do you... discrimination against the Yanks?
Do you know what Feren Balogun and Scott Galloway have in common other than exceptional athletic skill?
We're both anchor babies. His mother was 7 months pregnant and the airline wouldn't let her fly. And so he was born in the US. That's why he was on Team USA. And Sylvia Lavine and Tom Galloway who met in Toronto at a dance. My mom was 7 and a half months pregnant. Pretty sure they wed. I think they've been lying to me about that. And too late to check now. And they decided they couldn't endure another winter. This is a true story. And my mom read in the Toronto Globe and Mail that the city with the best weather in North America was this city called San Diego. So they loaded up my mom's Austin Mini Metro, which is essentially a lawnmower with doors. I don't know if you've ever seen Mr. Bean. It's the go-kart.
And seven and a half months pregnant, my mom and my dad traversed across the US with no air conditioning in their car and landed in San Diego. And five weeks later, I was born. So had this ridiculous attempt to turn back the 14th Amendment been in place now, yours truly would not be an American citizen. So what I am asking is that the IRS recognize that I'm not a citizen as I want to move to Dubai and I want the $100 million in taxes me and my company have paid over the last decade. I want it back. According to you, Stephen Miller and Donald Trump, I am not a citizen, meaning I'm not obligated to pay taxes in the United States. So, anchor baby here. Anchor baby. Revoke my citizenship and tax status, please.
Love it. I'm kind of a reverse anchor baby. Or I guess maybe I'm an anchor baby in Britain. American parents go to the UK. Born in the UK, come back here.
Oh my god. We're housing an undocumented worker.
Oh my god. It's almost bonus time. I'm going to be like that person that called ICE on the roofers when they were done roofing her house and she didn't want to pay them so she called ICE literally December 30th. Merry Christmas. Please meet Mr. Sanchez from ICE.
We were going to pay you a bonus, but it just wouldn't be right. It just wouldn't be right. Get him out of here. Why does that make me happy? Why does that make me happy? I'm calling my son.
That's good. That's good.
Yeah, but I'm a citizen, so I've got the dual passport, so it's not really going to work out, but you can try it.
Wait, so you weren't born here, so your parents applied for your citizenship?
I don't really know what they did, but they're American, so...
I'm sure you don't, you little illegal immigrant.
Everyone knows what's going on here. Everyone knows what's going on here. Okay.
All right. Well, now that we've figured out our citizenship status, you will have it revoked. And I apparently am illegal currently. Shall we get into our three big stories? We've got a lot to talk about today.
Let's get to the news. Stop bantering.
Let's do it. There were two strange updates in the AI world last week. First, OpenAI proposed giving the Trump administration a 5% stake in the company. That stake would be worth roughly $43 billion. Altman has reportedly argued that a government stake in the company would be the best way to share in the upside of AI with the public. And there's a vision for other AI companies to potentially join in, essentially creating a sovereign wealth fund. And then in the second piece of news, Meta is reportedly becoming a cloud provider. The company is preparing to sell its excess AI computing capacity. That is a notable shift after spending billions to build out infrastructure that it previously said would be necessary for its own AI ambitions. Investors did welcome the news though, sending Meta shares up nearly 9%, but the reaction wasn't nearly as positive elsewhere. Shares of Neocloud companies like CoreWeave and Nebius fell roughly 12% on the news. So Scott, lots to get into in the AI department here. First, I'm just going to take my own little personal victory lap because if you follow me on Twitter, you know that I actually bought Meta last week at $564. Here we are. The reason I bought is because I thought the stock was way undervalued, trading at 20 times earnings. S&P trading at 25 times. So, that's pretty cheap relative to the rest of the market. And this is a player that obviously is positioned for AI if we do see an AI boom. Plus, the last time it traded that low was in 2022. And since then, revenues have grown more than 70%. Gross margins have increased 200 basis points. So the whole thing just looked very attractive. The market was very worried that Zuckerberg was investing all of this money into these data centers, but hadn't laid out a plan for how he was going to monetize those data centers. Well, now he's given us the answer and he's basically just going to do what AWS did and what the cloud providers do. We just rent out the chips and sell it to enterprises. Now, there is a lot of debate over whether this is actually bullish or bearish because what we also know is that the plan was to use all of that compute to build their own AI products. And now they're saying, 'No, we're not going to do that anymore. We're going to sell it to someone else and someone else is going to build the AI products.' Which begs this question, who's going to build the AI products? And why isn't Meta down to do it if they are one of the most capitalized companies in the world and they are supposedly supposed to be a leader in AI? That's the bearish question. Lots to get into. Your reactions.
This sent chills down my spine or I had real deja vu of '99 again because what do we have here essentially? And I don't listen to markets unless nor enjoy it unless I'm on it. But I did find that guy you had on, Ed Zitron, he's very good and I was fascinated by what he said. He's basically saying this whole thing's crumbling is that's how that's my takeaway and I was fascinated by... I mean his basic thesis and tell me if I have this right is that all of the demand that trickles down to the trillions of dollars of capex and infrastructure and unbelievable valuations from construction firms to the chips guys is now basically down to demand being created by two players and that is Anthropic and OpenAI are still growing but Meta and xAI who were planning to create demand themselves with their own customer bases needs for AI products have basically said oh we overestimated the front-end demand we could create so we're going to lease out our infrastructure. So, all of a sudden, it looks like we've gone from another unbelievable pivot from a crisis of supply to a crisis of front-end demand. And probably the strongest evidence of that, of the kind of demand shock, if you will, is OpenAI delaying their IPO. In this market, you would think even if they'd lost some momentum, they would be jonesing to get out. And I mean it looks as if... so what did we have? We had Grok has 5% market share, Claude has 14, ChatGPT 49, Gemini 20 and Llama had 2.1. And I don't think it was the market saying, 'Okay, this is, we like the revenue that these guys are going to pick up.' It's actually a drop in the bucket for these guys. I think that what the market was saying is this is a capex race that is going to be a loser and we like that Meta is beginning to look more like Apple and they're basically getting out early of these capex wars and this all feels very '99 to me in that you had... so we knew the internet was going to be huge, people were confident of that, but the front-end guys, the application layer, the Pets.coms, the Amazon.coms, the eToys, it was clear that they weren't going to be able to create the kind of revenue in the short term that justified their valuations. So, everyone thought about, 'Wait, let's go to B2B.' So, we went from OpenAI to B2B, Anthropic, which was big in the enterprise, right? And then we found out there was companies, you know, remember this, like Internet Capital Group, which was supposed to be a marketplace so Pepsi could buy sugar. And then what everyone realized was it was just total nonsense, that it was much easier to pick up the phone or just send a fax and the B2B guys collapsed. And then they thought, 'Well, hold on. Okay, B2C and B2B, maybe we can't pick the winners, but the internet's huge. We need infrastructure investments.' So people threw in the hat on trying to pick which front end or which B2B application of the layer was going to be huge. And they invested in infrastructure and they invested in Cisco, right? They invested in fiber. And then Cisco from '99 to 2001 lost 92% of its market capitalization. And that sort of feels like what's happening here is the front end is stressed. It's not creating the demand originally anticipated. So we went from OpenAI to... wait, Anthropic is B2B. That's where it's at now. We're starting to see what is it, token nausea. People are saying you're spending too much in the enterprise on this. You need to scale it back. And I think the next shoe to drop is going to be in the infrastructure layer. And what people don't... and that's not to say the tech is not amazing. That's not to say these companies won't be great companies that'll be around in 20 or 30 years. Amazon lost 90% of its value in a 24-month period. It's obviously recovered then and some. The internet has lived up to the hype. But they went through incredible volatility because the initial... I mean it's gone from not what can AI do to will AI pay. And right now, it seems to me we're starting to see real cracks in the wall or in the narrative.
Just to go to what Meta and Mark Zuckerberg had previously said about these hundreds of billions of dollars that they were investing and spending on building all of these data centers, which again the market didn't like because they didn't have a clear answer to the question of how are you going to monetize them? But basically, here are some quotes that executives at Meta said that told us roughly what they were thinking. So Zuckerberg said in 2024, quote, 'Our goal with Meta AI is to build the world's leading AI service both in quality and usage.' So the idea was we're going to build our own AI stuff. In 2025, the CFO said when she was asked about the ROI question, she said, quote, 'Even with the capacity that we're bringing online in 2025, we are having a hard time meeting the demand that teams have for compute resources across the company.' I.e. we are building all of these AI products and we can barely keep up with the demand for our own teams within the company. Then this is the most important quote when asked if he would ever just start a cloud business which he's now doing as of last week. Zuckerberg said this year, quote, 'We haven't done that yet because we think that we have a use for the compute. Obviously, if we get to a point where we feel that we have overbuilt, then that is an option that we have and that is partially what gives us confidence in investing in building this out.' Well, they've now done the thing that they said that they would only do if they felt that they had, quote, 'overbuilt.' So, the question is, have they overbuilt? And I think that it would be a very, very reasonable answer to say yes, they have because look at what they're now doing. So I think that you're right in pointing out the difference between the front end of AI, i.e. using ChatGPT, using Claude, etc., and the back end which is building the infrastructure, building the data centers. Meta clearly tried to build a front-end AI business. They tried to do Llama, as you said, 2% market share. Clearly it didn't work. So now they're shifting to the back end. They say we're not going to build those products. We're going to sell stuff to other companies who are going to build the products. xAI, similar position. They tried to build a front-end business, mostly hasn't worked. Grok has like 5% market share. They're shifting to the back end. They're shifting more focus on building the data centers which they are now renting out to other companies that they should really be competing with. And so now most companies are basically deciding building the front end is a bad business. Better to build the back end, build the infrastructure and sell it to the front-end AI businesses. So then the question becomes, okay, who are these front-end AI businesses that they're selling to that supposedly is a good business they've invested hundreds of billions of dollars into? And the answer is there are two companies. It's Anthropic and it's OpenAI and that's basically it. And when we look at our estimates, those two companies alone account for between 60 to 80% of the AI revenues for Amazon, for Google, and for Microsoft. And according to The Information, those two companies alone make up half of the entire revenue backlog of the hyperscalers, i.e. the big tech companies. Meaning that the back-end infrastructure business only works if OpenAI and Anthropic continue to pay all this money and keep this whole thing afloat, which then begs the question, do you think they'll keep actually paying? And this is where the financials are so important, which were leaked by Ed Zitron, who we had on, and we learned that OpenAI made $13 billion in revenue last year. Okay, great. But they spent $34 billion. So their operating loss was $21 billion. Anthropic, we don't know the financials, but we at Prof G know that they made $4.5 billion in revenue last year. Based on our estimates, they probably spent around $15.5 billion. That's an $11 billion operating loss. Meaning the front-end AI business only works if the VCs continue to subsidize it to the tune of hundreds of billions of dollars. And by the way, if the hyperscalers continue to subsidize it, but if they stop doing that, then suddenly this business of building front-end AI doesn't work anymore, which means the business of selling the back end doesn't work anymore, which basically means the whole thing falls apart. Which brings us to OpenAI and their recent news, which is that they have decided that the best bet is to go to the government for what seems to be something like a bailout. And that is according to the Financial Times, they've discussed giving a 5% stake to the US government. And so it seems as though instead of Silicon Valley subsidizing those losses, maybe now just the taxpayers will. And maybe that's the plan. And maybe that's a good idea because that's what the banks did in 2008. And it didn't work for some of them, but it worked out for most of them. And so you have to think maybe they see this collapsing and that's why they go to the government. But either way, both of these new pieces of news in the same week, that's very, very bearish in my view and seems to indicate that this is a growing bubble that is nearing a point of maybe not collapse, but certainly massive course correction.
OpenAI. I predicted this six months ago that the biggest bailout in corporate history was about to happen and it was going to be the bailout of Nadella, Altman, Dario Amodei and it would be dressed up as investment or growth. It's not. It's a bailout. If the government were to take a 5% stake in OpenAI, great. They're going to favor OpenAI. They're going to overregulate their competitors and underregulate OpenAI. They're going to provide them with protection money and direct access to the White House. It's not even socialism. It's cronyism. It's like, you know, when things are really good, we want to capture all the gains ourselves, but when things are bad, we want to socialize the losses. That's not capitalism on the way up and socialism on the way down, it's cronyism. And that's what's being offered here. It's also evidence that there's something wrong in Mudville. And then just as the great flipping was the reversal of fortunes from the massive leakage of momentum from OpenAI to Anthropic, we've seen another incredible flipping, but this one is geographic and that is free Chinese models went from 30% of AI traffic to 60% in six months. The dominant AI models are now imports. I mean this is the thing that rocked the automobile industry was a product called the Honda Civic that was sort of a slow-moving train wreck over 10 years. This has been 10 weeks. GPU rates are collapsing. The hyperscalers built $300 billion of infrastructure for customers who are switching to DeepSeek for free. The way to describe this historically is this is the fiber overbuild of 1999.
The Chinese model problem is like another problem on top of all of their other problems because I mean as you pointed out a lot of these large companies are now switching. I can go through a list of some of them who have switched to Chinese models. Coinbase is now using Kimi. Cursor is using Kimi. Shopify is using Qwen. Airbnb is using Qwen. Siemens is using DeepSeek. Microsoft is now apparently testing DeepSeek. And the reason they're all doing this is because the Chinese models are way cheaper. Now, we've talked about why is that the case? Some say it's because Chinese have cheaper energy, and that might be part of it, but also it seems that they're kind of just stealing the models from the US companies through this process of distillation.
I don't think the Chinese would do that, Ed.
They hate stealing IP. It's not their game. Yeah, that's probably what's happening. And unless you can do something to stop that, then this train's going to keep running here. And the Chinese models are going to continue to take up market share. And the real problem is the fact that it has now sparked OpenAI to supposedly consider reducing their own prices. And if you're a company that's making $13 billion and spending $34 billion and you're going to decide that actually we're going to reduce the revenue number because we're being priced out by other companies that are offering not just comparable products but literally equal products. I mean studies have been done. These models are perfectly good on a variety of different metrics. That basically means that OpenAI's business is in big, big trouble. And again, they can say that they're giving the stake to the US government because they want the American people to all share in the upside, but what they're really saying is we want the American people to all share in the downside. Because right now, the business is currently all downside. You haven't figured out a way to make this product profitable. You haven't figured out a way to sell this product without spending literally like billions and billions of dollars on sales and marketing. Almost $6 billion. They could have bought Super Bowl ads for a decade. Every single Super Bowl ad for literally a decade. Like this is a business that clearly hasn't proven itself and they're already going to government and saying, 'You guys support us.' What I can't tell though is whether that... I mean clearly I'm bearish after hearing about the Meta news. But now I'm wondering, okay, if the government just bails them out then maybe it will all work out. Like you, I don't know if we should necessarily be betting against companies that are literally going to be subsidized by the full force and the balance sheet of the US government. I don't know if that's actually what's going to happen here. And I don't think it's fair. And I think it's honestly reprehensible. And I can't believe that we can't learn from our mistakes in the past and continue to do the same thing over and over and over again. And I know it's going to create hatred of these companies among the American people, i.e. the taxpayers who are going to pay for all of this. But part of me thinks maybe that works, just getting bailed out. And maybe that means that they can figure out a way to reach profitability if they can just continue to have their losses subsidized by someone who's way richer than them. It started off with Silicon Valley, then it went to big tech, like companies like Microsoft, and now it's going to literally the US government. I mean, maybe that's a winning strategy. I don't know.
This is a common theme in America. And that is my generation or the incumbents, if you will. It's not old versus young. It's not rich versus poor. It's incumbent versus entrants. But the incumbents who are sitting on top of a company recently valued at $875 billion, it's okay. If they go to $100 billion, they're still going to be worth twice what Ford is worth. They're still going to be one of the 100 most valuable companies in the world. Companies trading at $875 billion shouldn't get bailouts. Now, you could argue that if Ford goes from $10 billion to zero and has to lay off a couple million people in the middle class, jobs are lost in the Midwest. Okay, maybe that's at least a feasible argument for government intervention having societal implications or a positive justification. OpenAI and Anthropic, burn baby burn. If all of a sudden Nvidia goes from being worth more than the entire German and Spanish stock market to just being worth what the French stock market is worth, okay, there's absolutely no justification for me to whip out your credit card and run up our deficit such that the current owners, the private owners of Anthropic and OpenAI shares can maintain their gains. This isn't about an existential crisis where you think we would lose millions of jobs and the very existence of this category and losing this game. This is about keeping the incumbents rich. And that is basically a curse or a virus that infects America. I have two college degrees and a bunch of homes, going to make it harder for everybody else to get one such that the value of my current assets go up. No, that's not the way America is supposed to work. And the idea that we're even talking about this for companies that are... if Anthropic were in Europe, it was founded five years ago, it would be one of the four most valuable companies in Europe right now. What company that's one of the 20 most valuable companies in Europe has any discussion of a bailout? The plane is running at Mach 3 and doing just fine. Why do the pilots need to bail out? Why do we need to pay for these things? Say they lose 80% of their value. Amazon lost 92% of its value. We didn't bail them out. That's fine. That's part of capitalism. So the fact that this is even being discussed in my opinion is Sam Altman saying I have no fidelity to capitalism. I did this weird deal with TPG and other private equity firms that guarantee them a 16 or 17% pick. All that does is start to crush the previous rounds of investors because those numbers will start to add up pretty fast. But my god, if they lose 80% of their value. So you're saying it's only worth as much as the entire US auto industry. Sounds Tesla.
But it's almost like they want to be systemic. They want to be... it's like OpenAI is essentially like a zombie child that was formed by all of the big tech companies that have become systemic to the stock market. And it's like Sam Altman is trying to make the company as systemic as possible such that eventually when things go wrong we have no choice but to bail them out. It seems like this is almost all part of the design that he wants to be too big to fail. He wants for the US government to own a 5% stake in it so that if anything ever goes wrong, why wouldn't the US government figure out a way to bail them out? And it's almost like none of this growth is organic. Like give ChatGPT its credit on a user basis. That was organic growth. People loved the product and they started using it. But they haven't figured out how to monetize it in a realistic or sustainable way. And so all of the growth that we've seen from the valuations, from the circular deals, now going to the government, all of that is basically a setup where they are artificially growing this thing and they're trying to make it so big and so core and such a part of our lives such that by the time the business truly does flounder, we'll have no choice but to say, 'Hey, we need this thing. We're all in this together. Let's bail them out.' It feels like that is the direction that they are taking, that they want to become as systemic as humanly possible. And it's honestly gross. I just can't... seeing that headline and I bet Trump is down because he basically will do anything for anyone who shows up to the White House and kisses his ass. This is what we've seen time and time again.
It's more than just showing up to the White House. What is it? 93% of GDP growth, 75% of earnings growth come down to these firms' continued spend. So just as we have convinced ourselves it makes sense to spend $7 trillion on government spending with $5 trillion receipts just to keep the good times and the sugar high coming. Trump does not want the economy to slow down, especially going into the midterms. So he will potentially be open to the notion of doing whatever is required including pulling out your credit card to keep the music playing. But this is no different. Corporations, you know, Mitt Romney tried to tell us corporations are people. I don't think that's true. I think that corporations, you let your thoroughbreds run. The idea is they pay their taxes and then you try and fund a navy and parks and SNAP payments and unemployment insurance for actual people. But this really is no different than say someone is worth $100 million and they really like being worth $100 million and they're really good friends with Trump and they're really good for their community and they spend a lot of money everywhere. I'm like, 'Oh no, my business is [__] and I need the government to invest in my business and give me government contracts and also maybe put some of my competitors out of business, maybe give me government-backed loans such that I'm not worth just $25 million.' Is that a good use of the government's money? So if OpenAI and Anthropic lost 75% of their market cap, they'd only be worth as much as Uber or Adobe or Charles Schwab or Amgen. But because this run-up, this sugar high has been fueled by this new technology and the unbelievable expectations there. This would be the absolute worst thing for our economy that we could do is to decide that these thoroughbreds that are amazing companies and worth a ton of money need to be propped up. It would be absolutely no different than picking the wealthiest 100 Americans and saying, 'We're going to use the credit card and the debt capacity of future generations to make sure you continue to be the top 100 wealthiest people in the world.' And to a certain extent, that's what we've been doing by weaponizing housing, by weaponizing or sequestering degrees to college. Again, yet another transfer to the already super wealthy.
We'll be right back after the break. And if you're enjoying the show so far, send it to a friend. And please follow us on YouTube and Spotify or wherever you get your podcasts.
Support for the show comes from Framer. If your team wants a website that looks and feels handcrafted, but is still fast to ship, Framer is built for that. You design on a visual canvas with responsive layouts, hosting, and a CMS built in. So the work is production-ready from day one. Agents work alongside you to draft pages and polish sections. Then you review and publish what goes live. Framer is the pro builder for creators, teams, and businesses that want a professional site and care enough to get every detail right. Agents solve the gap between AI-generated ideas and production-ready website work. The agent works in the same place where the real site is designed, managed, reviewed, and published. It lands on the canvas, stays editable, and can be published when the team is ready. Agents and Framer work alongside teams to streamline collaboration on the same canvas. Build custom code components, create and manage CMS content, optimize SEO settings, and ship everything all in one place. Learn how you can get more out of your site from a Framer specialist or get started building for free today at framer.com/markets for 30% off a Framer Pro annual plan. That's framer.com/markets for 30% off. framer.com/markets.
