Transcription
[Music] Right now, as I speak to you, in late November 2025, we're witnessing one of the most dangerous moments in modern financial history. The stock market just lost $450 billion in Nvidia's value over three trading days. And that's after the company reported record-breaking earnings of $57 billion in quarterly revenue. Think about that paradox for a moment.
For the past two years, Nvidia has been the Atlas holding up the entire technology sector and by extension the broader market. When the S&P 500 climbed to new heights, Nvidia was responsible for a disproportionate share of those gains. But here's what keeps me up at night. A Bank of America survey just revealed that 45% of fund managers now identify an AI bubble as the single biggest threat to markets. That number was 33% just one month earlier. The S&P 500's price-to-earnings ratio sits at 29.5, roughly 90% above its historical average. And even after spectacular earnings, Nvidia's stock sits 15% below its October peak of $212 per share, trading around $178. The rescue party is over. Nvidia can no longer save this market. Let me explain why.
You know, back in Omaha, we've got a saying, when you're in a hole, stop digging. But Wall Street never learned that lesson. They just keep digging faster and calling it innovation. Let me walk you through some mathematics that should terrify every investor paying attention.
Nvidia's market capitalization peaked at around $3 trillion in October 2025. To put that in perspective, that's larger than the entire GDP of the United Kingdom. Now, the company just reported $57 billion in quarterly revenue with spectacular profit margins. That's genuinely impressive business performance. But here's where the arithmetic gets uncomfortable. At a $3 trillion market cap, Nvidia was trading at roughly 13 times annual revenue. Think about that ratio for a moment. Apple, one of the most profitable companies in human history, trades at about eight times revenue. Amazon, during its highest growth periods, rarely exceeded five times revenue. Even at today's reduced price of $178 per share, Nvidia still carries a market value that requires absolutely perfect execution for years to come with zero margin for error.
The S&P 500 index itself has been carried disproportionately by just seven companies, the so-called Magnificent 7, with Nvidia being the crown jewel. According to JP Morgan's analysis, AI-related stocks have accounted for 75% of S&P 500 returns, 80% of earnings growth, and 90% of capital spending growth since ChatGPT launched in November 2022. That's not a diversified rally, folks. That's concentration risk on steroids.
Here's what makes this mathematically unsustainable. For Nvidia to continue propping up the entire market, it would need to keep growing at rates that become increasingly impossible as the base gets larger. The company grew revenue 62% year-over-year in its most recent quarter. That's remarkable for a company of this size. But sustaining 60% growth on a $3 trillion market cap means adding $1.8 trillion in value annually. That's not going to happen. And everyone with a calculator knows it. I've been studying markets for over seven decades and I can tell you that no company in history has been able to carry an entire market index on its shoulders indefinitely. It's not a comment on Nvidia's quality as a business. They make excellent products that genuinely solve real problems. It's simple mathematics. The weight becomes too heavy.
Consider what's happening right now. In November 2025, Nvidia reported earnings that beat analyst expectations across every metric. Data center revenue hit $51.2 billion, up 66% year-over-year. Guidance for next quarter called for around $65 billion in revenue, well above Wall Street estimates. In any rational market, that would have sent the stock soaring. Instead, it dropped 5.5% that week. The NASDAQ fell 3% in the same period. Why? Because the market has finally started asking the right question. It's not how good is Nvidia's business. The question is how much future growth is already baked into this price? And the answer is too much. We're seeing peak valuations collide with peak expectations. The forward PE ratio for Nvidia, even after its recent pullback, remains in the stratospheric range. Meanwhile, the S&P 500's Schiller PE ratio, which smooths earnings over 10 years to avoid temporary distortions, stands at 39.2. That's the highest level since October 2000, right before the dot-com crash took the NASDAQ down 78%.
Some of you might remember my famous quote, "What the wise do in the beginning, fools do in the end." We're now firmly in the fools phase of this AI investment cycle. When I see venture capital firms pouring 52.5% of all their investment capital into AI and machine learning startups, I don't see wisdom. I see panic. The panic of missing out. The panic of being left behind. The panic of watching your competitors make money while you sit on the sidelines. But here's what history teaches us. The companies that get in early on technological revolutions can do extraordinarily well. The companies that pile in at peak valuations usually get slaughtered. And right now, we're at peak valuations.
The options market is telling us something important, too. Nvidia's implied volatility stands above 50%. With options volume running at 200% of normal daily levels, that's not the behavior of a stock that's going to smoothly support the broader market. That's the behavior of a stock where traders are hedging enormous risks and betting on massive price swings in both directions.
Here's another mathematical reality. Nvidia's success requires continued massive spending by its customers. Microsoft alone accounts for nearly 20% of Nvidia's revenue on an annualized basis. Meta expects its capital expenditures to exceed $100 billion next year. Alphabet projects $91 to $93 billion in spending for 2025. Amazon, Oracle, and others are committing similar amounts. We're talking about total AI infrastructure spending approaching $1.6 trillion across the industry. That spending has to generate returns. Companies can't indefinitely pour hundreds of billions into data centers without showing their shareholders that those investments create value. And here's where the arithmetic gets really uncomfortable. So far, the payoff from AI investments has been modest compared to the capital deployed. OpenAI, despite raising billions and achieving a $500 billion valuation, is still losing money. Most AI applications remain in the experimental phase, generating more hype than revenue. I learned a long time ago that price is what you pay and value is what you get. Right now, the market is paying unprecedented prices for AI infrastructure, but the value being generated remains largely speculative. That gap has to close eventually, and it rarely closes in favor of those who paid top dollar.
The real danger is that Nvidia has become too important to the market narrative. When a single company's quarterly results can swing the entire S&P 500 by percentage points, you don't have a healthy market. You have a house of cards. And the mathematics of supporting that house become more impossible with each passing quarter, regardless of how good the underlying business performs.
Now, let me tell you about something that should make every serious investor deeply uncomfortable. It's what I call the great AI financing circle. And it bears disturbing similarities to schemes I've seen blow up twice before in my career. Once in the late 1960s and again during the dot-com bubble. Here's how it works. And pay close attention because this is important.
Nvidia invests $100 billion in OpenAI. OpenAI then commits to buying Nvidia chips worth hundreds of billions over the next few years. Microsoft invests in OpenAI and also buys massive amounts of Nvidia chips for its data centers. But Microsoft also invests in Coreweave, an AI infrastructure company, which then rents computing power back to OpenAI using Nvidia chips. And Nvidia has an equity stake in Coreweave, too. You see the circle? Money flows from Nvidia to OpenAI, which flows back to Nvidia through chip purchases, while Microsoft funds multiple players who all buy from Nvidia. And Nvidia owns pieces of the companies buying its products.
Michael Burry, you might remember him from The Big Short, recently wrote on X, "True end demand is ridiculously small. Almost all customers are funded by their dealers." When I started in this business, we called this vendor financing, and it was always a red flag. The last time I saw it practiced at this scale was during the dot-com bubble when companies would loan money to customers to buy their own products, then book those purchases as revenue. It created an illusion of demand that evaporated the moment the financing stopped.
Let me be crystal clear. I'm not suggesting anyone is doing anything illegal or even unusual by business standards. Vendor financing happens all the time in various industries, especially during infrastructure buildouts, but the scale we're seeing now is unprecedented. We're talking about tens and hundreds of billions of dollars circulating within a relatively closed ecosystem of about a dozen major players. The Yale School of Management recently published an analysis of this phenomenon. They found that AI-related capital expenditures surpassed consumer spending as the primary driver of US economic growth in the first half of 2025, accounting for 1.1% of GDP growth. When a single industry's capital spending becomes a measurable percentage of national GDP, you need to ask some hard questions about sustainability.
Here's what happens when this circle breaks. And trust me, it always breaks eventually. One company in the chain starts questioning the return on their AI investment. Maybe they realize their massive spending on data centers isn't generating the revenue growth they promised shareholders. So they scale back their next capital expenditure budget. That impacts Nvidia's next quarter. Nvidia's stock drops. The companies holding Nvidia stock as part of their AI investments see their balance sheets deteriorate. They cut spending further. The circle becomes a spiral, but now it's spiraling downward instead of up.
We're already seeing the first cracks. OpenAI, despite its $500 billion valuation and the hype surrounding ChatGPT, is losing billions of dollars annually. Its projected revenues for 2025 are around $13 billion, but it's committed to spending $60 billion per year on computing power with Oracle alone over the next five years. That's not a sustainable business model. That's a company burning through cash while hoping that someday, somehow, the economics will work out. The CNDC report about Oracle is particularly telling. The company expects to lose considerable sums of money renting data centers primarily to OpenAI, having already incurred a $100 million loss in its most recent quarter. Think about that. Oracle announced a massive deal with OpenAI. Its stock soared 40% in a single day, adding nearly a third of a trillion dollars to its market value. But the actual economics of the deal are unprofitable for Oracle. That's not investing, folks. That's speculation masquerading as business strategy.
Charlie Munger, my late partner, used to say, "Show me the incentive and I'll show you the outcome." Right now, every major tech company faces intense pressure to demonstrate they're not being left behind in the AI race. CEOs go on earnings calls and tout their AI spending as proof they're forward-thinking. Wall Street analysts reward companies that increase AI capital expenditures with higher stock prices, but nobody's asking the uncomfortable question, when does all this spending generate actual returns?
I've lived through enough cycles to recognize when markets lose touch with business fundamentals. In the late 1960s, the Nifty Fifty stocks traded at any price because growth was assumed to continue forever. In the late 1990s, internet companies with zero profits commanded multi-billion dollar valuations because they had eyeballs and mind share. The specifics change, but the pattern remains the same. When price becomes disconnected from underlying business value. When complex circular financial arrangements replace straightforward profitability. When everyone justifies extreme valuations by pointing to everyone else doing the same thing. That's when the wise investor starts looking for the exits.
Now I want to be fair to the companies involved. AI is genuinely transformative technology. In 20 or 30 years, we'll look back and see that artificial intelligence changed the world in meaningful ways, just as the internet ultimately did. But that doesn't mean current prices are justified. And it certainly doesn't mean every company pouring billions into AI infrastructure will survive the transition. The internet bubble taught us a crucial lesson. Being right about the technology doesn't mean you'll make money as an investor. Pets.com was right that people would eventually buy pet supplies online. They were just 15 years too early and priced as if the future had already arrived. Amazon survived because Jeff Bezos raised enough capital to weather the storm and had a sustainable business model underneath the growth story. Most others didn't.
Today's AI investments feel eerily similar. The technology is real. The potential is enormous. But the economics remain murky, and the valuations assume perfect execution of a future that may take much longer to materialize than current prices suggest.
Here's what keeps me up at night about this circular financing scheme. It makes Nvidia's earnings reports less reliable as indicators of genuine end-user demand. When your customers are largely funded by entities you've invested in or partnered with, and those same customers are paying you with money that partly came from you in the first place, you're not seeing organic market demand. You're seeing a carefully orchestrated dance that works beautifully as long as everyone keeps dancing. But the music has to stop eventually. The question isn't if, but when. And what happens to stock prices when investors realize that billions in revenue was actually just money moving in circles?
Let me tell you about a pattern I've observed over seven decades of investing. Markets always need a hero: a company, a sector, a story that justifies rising prices and soothes nervous investors. In the 1960s, it was the Nifty Fifty blue-chip stocks that supposedly couldn't fail. In the 1990s, it was internet companies that would change everything. In the 2000s, it was housing and mortgage-backed securities. Each time investors believed they'd found an asset that defied gravity. Nvidia has been that hero for the past two years. Whenever the market wobbled, investors could point to Nvidia's extraordinary growth and say, "See, the AI revolution is real, and as long as Nvidia keeps executing, the market has a foundation."
But here's what history teaches us. When markets become dependent on a single company or narrow group of companies to maintain their momentum, they're already in trouble. The numbers tell a stark story. On November 20th, 2025, Nvidia reported those spectacular earnings I mentioned earlier. $57 billion in quarterly revenue, up 62% year-over-year with guidance suggesting continued strength. The company did everything right. And yet, its stock fell 3.2% that day, contributing to a broader market sell-off that saw the NASDAQ drop 2%, the S&P 500 decline 1.56%, and the Dow lose nearly 390 points. That disconnect should tell you everything you need to know about where we are in this cycle. Good news is no longer good enough. Spectacular news barely moves the needle because the market has finally realized what value investors like me have been saying all along. At some point, price matters more than growth.
The VIX volatility index, what some call the fear gauge, jumped 14.2% to hit the 20 level on that same day, marking its biggest single-day increase in more than a month. CNN's fear and greed index has plunged into extreme fear territory. These aren't the indicators of a healthy market being supported by a strong foundation. These are the indicators of a market that knows it's overextended but hasn't figured out how to walk back from the edge.
Here's what I learned from watching markets for more than 70 years. When one company becomes responsible for holding up an entire index, that's not strength. It's fragility disguised as resilience. According to various analyses, the Magnificent 7 tech stocks have accounted for the vast majority of S&P 500 gains over the past two years. Take away their performance and the rest of the market has barely moved. The equal-weighted S&P 500 tells a very different story than the market-cap-weighted version because a handful of enormous companies have pulled the average up while hundreds of others stagnated. This concentration of returns creates a dangerous illusion. Investors look at the S&P 500 hitting new highs and think the market is healthy. But beneath the surface, you've got a market where most stocks aren't participating in the rally. It's like saying the average temperature in a room is comfortable when half the room is freezing and the other half is on fire. The average might be fine, but the reality is dangerous.
Now, let me tell you what happens when markets run out of saviors. I've seen it twice before in my career and it's never pretty. First, you get denial. Investors see warning signs but convince themselves that this time is different. They point to genuine technological advances, and AI genuinely is revolutionary technology, to justify valuations that make no mathematical sense. They tell themselves the traditional valuation metrics don't apply to transformative companies. I heard the exact same arguments about internet stocks in 1999.
Then you get the crack in confidence. It usually starts with one bad quarter or one piece of unexpected news. Maybe a major tech company reports that its AI investments aren't generating the returns they expected. Maybe one of the circular financing arrangements I described earlier unravels. Maybe interest rates move in a direction that makes current valuations look even more absurd. The specific trigger doesn't matter much. What matters is that once confidence cracks, it spreads quickly.
Finally, you get the repricing. This is where it gets ugly. When investors simultaneously realize that they've been paying unsustainable prices, they all try to get out through the same narrow door. The stocks that were market darlings become market pariahs. Companies that couldn't possibly fail suddenly trade at fractions of their peak valuations. And the broader market, which had become dependent on those few high-flyers to maintain the illusion of health, suddenly has nowhere to hide.
We're seeing early signs of this pattern right now. Last week, the tech-heavy NASDAQ plummeted 3%, its largest weekly decline since Trump's tariffs rattled markets in April, even after bouncing back on Friday. The week ended with massive volatility and uncertainty. Declining stocks outpaced advancing ones by more than 3 to 1 on the New York Stock Exchange. That's not the action of a healthy, broadly supported market. That's the action of a market where the weight has become too concentrated and the foundation is starting to crack.
The truly concerning part is what happens to portfolio construction when your hero fails. For the past two years, the smart money strategy was simple: own the Magnificent 7, especially Nvidia, and you'd outperform. Diversification looked foolish when a handful of AI-related stocks generated all the returns. But that strategy only works as long as the hero keeps rising. When Nvidia can no longer save the market, suddenly those concentrated portfolios become extremely vulnerable.
I've always believed in Charlie's principle. The first rule of compounding is to never interrupt it unnecessarily. But there's a corollary that's equally important. Know when you're not actually compounding value, but rather riding a wave of speculation. The mathematical reality is that Nvidia, no matter how excellent its business, cannot continue carrying the entire market on its shoulders. The weight has become too great, the valuations too extended, the expectations too high.
Bank of America's survey showing 45% of fund managers identifying an AI bubble as the top risk represents a fundamental shift in sentiment. When nearly half of professional money managers publicly acknowledge that they're worried about a bubble, you can bet the private concerns are even more dire. These are the people managing trillions of dollars, and they're telling you they see danger ahead. The real question isn't whether Nvidia can continue saving the market. The math says it cannot. The real question is what happens to your portfolio when the market accepts that reality? Because here's what seven decades of experience has taught me. When markets finally acknowledge an uncomfortable truth, the repricing happens fast, not gradually, not with gentle warnings, but suddenly and severely.
Right now, you've got a stock market trading at the second highest valuation in history, supported largely by a handful of AI-related companies, with most investors knowing the valuations are stretched but hoping they can exit before everyone else does. That's not a market being saved by Nvidia. That's a market waiting for the rescue party to end, and hoping they're not the last ones standing when the music stops.
Let me bring this home with some straight talk. The kind I've given Berkshire shareholders for over six decades. The current situation with Nvidia and the broader market reminds me of a quote I've used many times. A pin lies in wait for every bubble. And when the two eventually meet, a new wave of investors learn some very old lessons. We're at that meeting point right now.
Nvidia is a phenomenal company with outstanding products and excellent management. I have tremendous respect for what Jensen Huang has built. But being a great company doesn't make you immune to market mathematics. At current valuations, even continued strong performance may not be enough to justify the price, much less continue carrying an entire market higher. The S&P 500's PE ratio of 29.5 means you're paying nearly $30 for every dollar of current earnings. The Schiller PE of 39.2 means you're paying nearly $40 for every dollar of inflation-adjusted 10-year average earnings. Both metrics are screaming that stocks are expensive by historical standards. Not slightly expensive, dramatically expensive. The only other time we've seen valuations this stretched was right before the dot-com crash.
Now, I know what some of you are thinking. Warren, you're missing out on the AI revolution. Technology is different now. These companies have real earnings, real cash flow, real business models. And you're partially right. Today's tech giants are far more profitable than the dot-com companies I avoided in the late 1990s. But profitability doesn't insulate you from paying too much. As I've said before, price is what you pay. Value is what you get. And right now, the market is paying record prices while the value proposition becomes increasingly questionable. The circular financing schemes, the concentrated returns in just a handful of stocks, the massive capital expenditures with uncertain returns, the valuations that require perfect execution for years to come. These aren't signs of a market with a solid foundation. They're signs of a market that's become dependent on continued expansion of speculation rather than fundamental business value.
My advice, the same advice I've been giving for 70 years. Understand what you own. Pay attention to price. Maintain a margin of safety. And don't assume that just because everyone else is doing something, it must be right. As I wrote in my 1989 shareholder letter, "What the wise do in the beginning, fools do in the end." We're firmly in the end phase of the AI speculation cycle. Nvidia cannot save this market because no single company, no matter how excellent, can justify valuations this extended across an entire index. The mathematics don't work. The historical precedents are clear, and the warning signs are everywhere for those willing to see them. The most dangerous words in investing are, "This time is different." It's never different. The details change, but human psychology and mathematical reality remain constant. And right now, both are telling us the same thing: we're paying too much, hoping for too much, and depending on too few companies to deliver miracles.
I'll leave you with this final thought. Back in 2008, when financial markets were collapsing and fear gripped Wall Street, I wrote an op-ed for the New York Times titled "Buy American, I Am." That wasn't market timing or heroism. It was a simple recognition that prices had fallen to levels where value exceeded price. Today we face the opposite situation. Prices have risen to levels where they significantly exceed any reasonable estimation of value. Being right about timing the market is impossible. But being right about value is achievable if you're patient and disciplined. Right now, the market is pricing in perfection from Nvidia and the handful of companies surrounding it. When markets price in perfection, they set themselves up for disappointment. And when that disappointment comes, not if, but when, those who paid premium prices for speculative growth will learn expensive lessons. My recommendation remains what it's always been. Focus on businesses you understand. Never pay prices that require miracles. Maintain adequate diversification. And remember that your goal as an investor is to compound wealth over decades, not chase momentum over quarters. The market might continue rising for a while. Speculation can persist longer than logic suggests, but gravity always wins eventually. Stay safe out there and remember, it's better to miss out on gains than to participate in losses.