Google Surges Past OpenAI and AI Faces a Bigger Crisis That Could Break the Boom | The Weekly Wrap

Meet the computer you can talk to. With Copilot on Windows, working, creating, and collaborating is as easy as talking. Got writer's block? Share your screen with Copilot Vision to help spark inspiration. And use Copilot Voice to have a conversation and brainstorm ideas. Or maybe you need some tech help. With Copilot Vision, Copilot sees what you see. Let Copilot talk you through step-by-step guidance so you can master new apps, games, and skills faster. Try now at windows.com slash copilot.

You're about to make a trade. Which you do you listen to? Is it get optioning those options? Or let's do a little research. Learn more at finra.org slash trade smart. The S&P 500 last week was down 2% and Nasdaq was down almost 3%. But this Thanksgiving week, the market staged a very strong rally, completely making up for last week's losses. Dell says it expects about 31 and a half...

billion in sales in the fourth quarter versus only $27.6 billion estimated by analysts. So a big jump in revenue guidance. Google launched its latest AI version, Gemini 3. Until now, the perception has been at OpenAI through ChatGPT has been way ahead. Well,

no longer. And when you compare the financials of Google versus OpenAI, it's not even a competition. Chips are no longer the real binding constraint. The binding constraint is power. Data centers account for five to seven percent of electrical demand. By 2028, it will be well north of 10 percent. But we are already having problems.

Hi, this is Steve Eisman, and welcome to another edition of The Weekly Wrap. This is for the week ending November 28, 2025. Hope everyone had a great Thanksgiving. Last week, the market suffered a decline.

as fears about an AI bubble surfaced. The S&P 500 last week was down 2%, and NASDAQ was down almost 3%. But this Thanksgiving week, the market staged a very strong rally, completely making up. for last week's losses. This week's rally might have been sparked by dovish statements by a few Fed governors renewing speculation that the Fed will cut in December. The market is not at all-time highs, but...

Despite all the volatility, it's pretty close. Post-NVIDIA earnings, it was a slow holiday week for earnings. I'll go through one company that reported and then focus on some themes, especially AI themes, that I think... are front and center these days. So Dell reported earnings. Now, Dell has a January fiscal year, which is a bit odd for a tech company. It's probably because it sells many consumer products. So like a retailer...

Christmas is important. It has a consumer retail style fiscal year. Anyway, Dell is an AI related play, but with some important caveats. It's a great company. and sells mostly personal computers and servers. The server business is a big beneficiary of the AI database build out as more and more servers are needed. The caveat is that Dell does not really sell a proprietary product.

and has to compete with the likes of Supermicro on price and service. Now, Dell reported fiscal third quarter earnings on Tuesday that missed Wall Street expectations for revenue, but beat slightly on earnings. Earnings per share were up 17% versus last year at a beat. Revenue was up 11% versus last year, but a slight miss. Dell says it expects about $31.5 billion in sales in the fourth quarter.

Versus only $27.6 billion estimated by analysts. So a big jump in revenue guidance. And it also raised fourth quarter earnings per share guidance. So very good guidance from Dell. But again, because of the lack of proprietary products. Dell's one year out P.E. is a measly 11 times, despite EPS growth expectations of mid to high teens. Supermicro wields a similar low multiple. These low P.E. multiples.

are akin to a bank. And that's not a compliment. Not a high-flying tech company. And again, it's a low multiple because Dell and Supermicro sell few proprietary products and must compete. on price and service. Last week on the Weekly Wrap, I explored some of the risks involved with the AI boom. What I focused on was analogizing the AI revolution to the railroad boom of 1870 to 1895. Now back then, the railroads transformed the U.S. economy from a regional agrarian one

to a national industrial economy. It was quite a revolution. But it was not always profitable. There were two market crashes, 1873 and 1893, when it became questionable. how profitable the railroads would become. Also, there were multiple railroads built and several went bankrupt. Some of those railroads ran side by side.

I want to return to issues surrounding AI in two ways. First, I want to explore a bit more this analogy of two railroads running side by side, thereby creating overcapacity. And then... I want to explore what I believe is the binding constraint in AI. With respect to the two railroads analogy, last week, a memo was leaked from Sam Altman, the CEO of OpenAI, to all the employees at OpenAI.

that really presses this two railroad analogy home. Recently, Google launched its latest AI version, Gemini 3. Until now, the perception has been that OpenAI through ChatGPT has been way ahead. Well, no longer. In a memo to employees, Sam Altman, the CEO of OpenAI, admitted that Gemini 3 has leapfrogged chat GPT in some ways and that, quote, we know.

We have some work to do, but we are catching up fast, unquote. And he cautioned that, quote, I expect the vibes out there to be roughed for a bit, unquote. What OpenAI through ChatGPT has achieved is nothing short of amazing. But it looks like the second railroad has caught up and perhaps surpassed its upstart. This is a big deal. We have been taught...

Time and time and time again over the last 20 years that upstarts defeat incumbents because the incumbents are asleep. That's why Amazon obliterated old retail. It's why Netflix killed old media. But Google, Amazon, and Microsoft are not Macy's. They all woke up fast. And when you compare the financials of Google versus OpenAI, it's not even a competition. In 3Q25 alone, Google generated $102 billion in revenue, up 16% versus last year, and $35 billion in net income, up 33% versus last year.

More importantly, despite all the incredible sums Google is spending on AI, which is probably north of $90 billion in 2025 alone, Google can fund. that entire amount from its own cash flow, with very little additional debt. Now, how do I know that? Because long-term debt is only up $10 billion versus year-end 2024, and cash and cash equivalents are at $98.

billion, which is slightly higher than year-end 2024. The second railroad has caught up. Now, OpenAI will generate around $13 billion in revenue in 2025. and plans on burning hundreds of billions of dollars in pursuit of AI supremacy. Now, that pursuit is completely dependent on Sam Altman being able to continue to raise hundreds of billions of dollars. He could do it.

But the second railroad that is run by an international powerhouse has finally caught up. Now let's talk about binding constraints for AI. A few weeks ago, Michael Burry, who was portrayed in the big short... put out a table on X arguing that hyperscalers are playing accounting games by lengthening depreciation schedules. It's an interesting critique and argues that big tech companies are inflating their earnings. In some cases,

By a lot. And if he is right, many big tech companies are not just inflating earnings, but may have to take asset impairment write-offs at some unknown point in the future. On November 18th, Baidu. The large China tech company did just that. It took a large impairment charge because some of its older technology that it was using was obsolete. So I take Burry's criticism very seriously. But for now.

My view is that either all this AI spending will create incredible new products and productivity savings, or it won't. That will be true regardless of the depreciation schedules. And if the bulls are correct... and these AR products are in fact that revolutionary, then that will dwarf the depreciation argument. We will just have to see. And by the way, Burry also thinks that all this AI spending is just premature.

that it is similar to all the telecom spending that took place during the internet bubble years. He points out that when the dot-com bubble burst, only 5% of all that telecom fiber was lit. So in the short and long term, the big question is, will there be great applications and cost savings that will make all this gargantuan spending worth it? Burry thinks not. He could be right. My view is that it is too early to tell.

Sometimes it's just okay to admit that you don't know. But even assuming all this spending will be worth it. I want to focus on a concept that is crucial to the current state of AI that I am borrowing from the world of large banks, the binding constraint in large banks. The binding constraint is the annual stress test given by the Federal Reserve. That test is all important. In that test, the Fed provides the banks with a very draconian economic scenario, and the large banks have to model.

how much money they will lose and how much capital they will then have left. To pass the test, each bank must have a minimum amount of capital left plus some. That is the binding constraint on banks. What do I mean by that? Well, depending upon how much excess capital a bank has after the draconian analysis determines how much capital in the form of dividends and buybacks.

the bank will be allowed to deploy for the year, fail the test, and severe penalties are imposed. The annual stress test is the binding constraint. This episode is brought to you by Amazon Prime. Black Friday game day on Prime is an epic day of live sports. It all starts at 9 a.m. Eastern with the Capital One Skins game. Then, Black Friday football returns when the Bears take on the Eagles at 3 p.m. And it culminates...

with the final night of Emirates NBA Cup group play with Bucks-Knicks at 7 p.m. and Mavs-Lakers at 10 p.m. Black Friday game day, only on Prime. Extra Value Meals are back. That means 10 Tender Juicy McNuggets and medium fries and a drink are just $8. Only at McDonald's. For limited time only. Prices and participation may vary. Prices may be higher in Hawaii, Alaska, and California and for delivery.

Now, there is a current binding constraint in AI, and it's not chips. NVIDIA, AMD, Broadcom, etc. are now manufacturing enough chips that the hyperscalers want to buy. Sure. The demand is insatiable, but chip companies are meeting this demand. And that wasn't always true. Maybe a year ago, there was a real struggle to get chips. But no longer. Chips are no longer the real binding constraint.

The binding constraint is power. Now, before we discuss the power issue, let's get our definition straight. A typical light bulb might be 100 watts. A kilowatt is 1,000 watts. A megawatt... is a thousand kilowatts. A gigawatt is a thousand megawatts. And the big one, the terawatt, is a thousand gigawatts. The United States will produce roughly 4,200 terawatts of electricity in 2025. Now, prior to 2021, terawatt usage in the U.S. was growing at most

1% per year. But now it's growing 3% per year. Now, that may not sound like a lot, going from 1% to 3% growth. Big deal, you say. But 3%... of 4,200 terawatts equals 125 terawatts of growth per year at least, which is equivalent to providing electricity to multiple big cities per year. New York, for example, uses about 60 terawatts of electricity per year. Now you get the point. All this growth is being fueled by AI. Today...

Data centers account for 5% to 7% of electrical demand. By 2028, it will be well north of 10%. But we are already having problems. And never forget, these projects take time to build. In 2019, six years ago, Digital Realty got permission to build a 48 megawatt facility in Santa Clara, California. It's now built. A similar size facility has been built.

by a digital realty private competitor called Slack. And they are both ready to go. One small problem. No power. Silicon Valley Power promised the power. but it is still working on its $450 million electrical grid upgrade. So two facilities are not working because they have no electricity and may not get power until 2028.

And these two facilities are not anywhere close to the size of facilities that are being proposed and built now for AI. Those new facilities utilize many multiples that amount of power. Many hyperscalers are looking at Texas for power. But even Texas is not without problems. In 2021, a huge winter storm caused rolling blackouts. If you add up all the projects proposed...

For Texas, one report out there argues that if all these projects are built, it would be the equivalent of electrical consumption of over 150 million homes. Right now, ERCOT. The Texas grid operator services only 30 million homes. This is why hyperscalers are looking for novel solutions. Nuclear is back. Building a mini utility and laying it right next to a hyperscale project is another potential solution. This is why a company like Oclo, whose stock price is up 300% this year, is so popular.

The company has a novel nuclear solution, but it has no revenue yet. Nothing has been approved yet by the federal government. My point is that everyone is buying the tools of AI. Forget for a second about how efficacious AI will be. Assume it's going to be great, but without sufficient power, some data centers will just sit idle. Power is the binding constraint, and it takes time to build power. Planning.

regulatory approvals, and building take years. Let's hope for many mild winters. I also want to focus today for a bit on CEOs. There is a tendency in the U.S. to worship at the altar of CEOs. They are treated like superstars. But sometimes they develop feet of clay. Take Jack Welch, CEO of GE in the 1990s. He was treated like a superstar. But when he retired in 2001...

the company began a long-term decline. Now, much of that can be laid at the feet of Jeff Immelt. But looking back, there is no question that Welch ended his tenure with GE leaving a bunch of hidden problems. not least of which was a sprawling financial services division that was a ticking time bomb. CEOs can be good for a time and then not. Let's look for a second at Brian Nickel, the CEO of Starbucks.

Mr. Nickel was the CEO of Chipotle, where he was lauded for his incredible marketing acumen. He was CEO from 2018 until September 2024 when he left for Starbucks. Under his regime at Chipotle, The stock went up around nine times, and then he left to go to Starbucks. And two things have since happened. Since he left, Chipotle, the stock, has been cut in half. Now, things don't fall apart so quickly.

The signs must have been there when he left, or at least some signs. And Nickel has not managed to engineer a turnaround at Starbucks whose problems run deep, at least not yet. My point is that nobody is perfect. And even good CEOs can run out of steam. And now for the mailbag. Now, normally, I just answer some mailbag questions. But once in a while, I get a comment that is so interesting that I just read it. In April of this year...

I interviewed Wolfgang Wundschau, the author of Kaput, the end of the German economic miracle. And we focused mostly on why Germany's economic growth has deteriorated so much. But we also discussed why European growth... is also so sluggish. I received this email from a viewer and I thought I would read an edited version of it. The email says, Hello, Steve. I am an avid viewer of your podcast. I am an entrepreneur from...

And he names the country. I'm not going to name it. When you said in one of your podcasts that you do not know what is behind the U.S. exceptionalism as regards to why Europe is always falling behind. I thought I would also provide you my own viewpoint. Perhaps you would understand why this is the case. I have lived and worked in, and he names a city in Europe, for 14 years before the matter I talk about in this email takes place.

In 2024, I tried to invest in a project in any names of European country that I carefully thought out. With a business plan of 30 pages and a rather okay capital, I have sent a permit application based on this idea. to a government agency that is supposed to give okay on the idea before I can come to this particular country and set up the company. This was my first business idea that came to my mind.

He then describes the product, but I will not describe it other than to say it is a consumer product. My first application was rejected. And then I made another application with even more documents and collateral, considering that the original proposed capital might have been viewed as insufficient. I have done the maximum extent of providing the facts that I was capable of setting up the company.

find a place to start the production, which I thought of around a million of euros in revenue that it would generate in the beginning as being feasible. Anyway, this second application was also rejected, and I had to take this matter to the local courts.

By this time, I was always wondering why this country, like most European countries, which has a stagnant economy and has no light at the end of the tunnel, has the audacity to so swiftly disregard any investment, even though it looks boring on paper. Sometimes boring ideas become successful, especially in tough economic times like this one we are going through. If I did not want to contribute to this economy, trust me, I had alternatives to choose. But this European country...

seemed like the best option. Going to court will make me lose a year on this business idea. I am just thinking this phenomenon is not limited to this experience and it's all over the place. Coming to the point in question, the reason Europe is always falling behind despite having the resources is, quote, the mentality. This also means the mentality in the U.S. is years ahead. In other words, Europe has the shovels.

Buzz does not know how to dig for gold. And that's the end of his email. I'd just point out that for this man's product in the U.S., there would be some regulatory issues, but no one would just say no. I think this email goes a long way to explaining why the European economy is so sclerotic. And that's the wrap.

This last Monday, I posted an interview with Amy Butte, the CFO of a new public company called Navon, a business travel company. Amy was also the CFO of the New York Stock Exchange when it went public in the early 2000s. We explore what it means and how difficult it is to take a private company public. So check it out. This Monday, I will post an event I attended and recorded on October 29.

I was invited to speak at Case Western University in Cleveland by Ivy Zellman, the famed housing analyst. I spoke in front of about 100 students about the financial crisis and the current market. and I took multiple questions from the students. It was a great event, and I think you will all enjoy watching it. So please tune in. If you haven't already, please consider subscribing to our YouTube channel. Also. Thank you.

be sure to check out our website, RealizeManPlaybook.com. There you can easily access all our episodes as well as the Financial Literacy Masterclasses, our new blog, and some other goodies as well. Check it out. And see you soon. Please do your own due diligence and consult a licensed financial advisor before making any investment decisions.

Now's the time to get up to 50% off Black Friday deals. With savings this big, you can get gifts for everyone on your list. Plus, if Black Friday lines aren't your thing, skip them. At Kay, you can buy online and pick up in store or get free shipping right to your home. This holiday season, unwrap love and savings with Kay. Exclusions apply. See Kay.com slash exclusions for details.

Hey, Ryan Reynolds here, wishing you a very happy half-off holiday because right now Mint Mobile is offering you the gift of 50% off unlimited. To be clear, that's half... price, not half the service. Mint is still premium, unlimited wireless for a great price. So, that means a half day, yeah? Give it a try at mintmobile.com slash switch.

Upfront payment of $45 for three-month plan equivalent to $15 per month required. New customer offer for first three months only. Speed slow after 35 gigabytes of networks busy. Taxes and fees extra. See mintmobile.com.