Buffett Warned About a 'Gambling' Market. His Company Just Bought $19.8 Billion.

You've seen the headline. Buffett gave a "stark warning," valuations are at records, and history says he's right to be scared. The conclusion the story quietly invites is simple: the market is overpriced, the top is near, so watch out — maybe sell.

That's a reasonable-sounding read. It's also, almost certainly, wrong — not about Buffett's numbers, which are real, but about what they're for. And the fact that exposes the misreading is the least flattering thing the "be careful" crowd can point at: in the same quarter he gave the warning, the company he built stopped selling and turned a net buyer of roughly $19.8 billion of stocks.

So what did the warning actually mean, and why does the difference change how you should handle your own money?

The Warning Is Real. The "Sell" Is Invented.

At his annual meeting on May 2, 2026, Buffett told CNBC he had "never had people in a more gambling mood than now". He pictured the market as "a church with a casino attached" — the church being slow-and-steady, patient ownership, the casino being short-term, high-risk betting — and said the casino "has gotten very attractive to people... that's not investing, it's not speculating, it's gambling." Over the following months he added that "prices for an awful lot of things will look very silly" and that it's "tough to find value when everybody is preferring gambling".

None of it is a rumor. But look at what is not in those words: no date, no "sell your portfolio," no "the crash is coming." What's in them is a statement about mood and about how hard it has become to find a fair price right now. Headlines converted a comment about temperament into a market call. That conversion is the whole error.

The sharper tell is in the money. Buffett stepped back as CEO at the end of 2025; Greg Abel now runs Berkshire. In the second quarter of 2026 — the same months as the "gambling mood" warning — Berkshire broke a 14-quarter run of net selling. It then turned a net buyer of roughly $19.8 billion of stocks, buying about $23.5 billion and selling about $3.7 billion. The signature new bet, a $10 billion stake in Alphabet, was one Buffett personally said he "initiated." The cash pile fell from a record near $380 billion to about $365 billion, and the company kept buying its own shares in July. If the warning meant "get out of the market," the most credible buyer on Wall Street would have been selling. It wasn't.

An All-Time High Is Not a Date

Here's where "history says he's right" needs a hard look, because history is being asked to say something it does not say.

Two gauges do the heavy lifting in these stories. The first is the Buffett indicator: the total value of all U.S. stocks divided by the country's GDP. It sits near 238%, an all-time high, well past the 200% level Buffett once called "playing with fire." The second is the Shiller CAPE ratio — a price-to-earnings multiple smoothed over ten years — which has held above 40 since May and is the second-highest reading in roughly 150 years, behind only the top of the dot-com bubble. Both numbers are accurate, and both are genuinely high.

But both are regime gauges, not timing gauges. They describe how expensive the whole market is relative to its own history. The honest historical lesson from a CAPE this high is not "the top is in the next few months." It is the much duller, much harder-to-trade statement that the average return over the next decade is likely to be modest. That's a description of the next ten years, not of next Tuesday. It has no date on it, and it has never had one.

The embarrassing part is that the dot-com "top" is exactly the proof. The CAPE sat at high levels for several years before the 1999–2000 peak, and the market still had more to run before it finally broke. A gauge at its highest reading is a condition, not a countdown. When a thermometer reads 103 for years, the useful conclusion is that you're in a fever, not that today is the day it breaks.

There's a second, deeper misread — and it's the one that actually governs your returns. A high price multiple mechanically implies a low forward return, and the crowd keeps optimizing the number instead of the denominator. Buy a market at 40 years of earnings and the return you can expect over the next decade is mathematically thinner than if you had bought it at 20 years of earnings — even if every company in it turns out to be a wonderful business. The metric the crowd celebrates is the level. The number that should be carrying your decision is the forward return it implies, and the two are pulling in opposite directions.

The Casino Is the Real Threat — and It's About Who's at the Table

So if the warning isn't a sell date, what is there to fear? This is the part the "all-time high" framing misses, and it's the part with teeth.

When Buffett says the mood has turned gambling, he's describing a change in who is buying and why. A market run by patient, long-horizon owners — the "church" — tends to grind up, because nobody is in a hurry to be wrong. A market run by short-horizon speculators — one-day options, prediction markets, the betting-app instinct — tends to spike and then drop, because the same people who bought the spike are the first to sell it. The casino changes less the price than the texture of the ride: bigger moves, faster, with thinner cushion between you and the downside.

That's a risk-management warning, not a valuation verdict. Its practical content is about your own behavior, not about the calendar: don't leverage up into a casino crowd, don't pile your whole position into one bet the crowd is also making, and don't hold a horizon shorter than the thing you're buying. None of that requires you to sell the market. All of it follows directly from "never had people in a more gambling mood than now."

The bulls will answer, fairly, that Buffett may be overstating the froth — that if earnings keep compounding fast enough, the CAPE can stay flat or even fall while prices rise, and the "expensive" label quietly dissolves. He's right that's possible, and it's the strongest defense of the rally. But it's a hope about earnings that has to be re-proven every quarter; it's not a date either.

Here's the version of the warning worth keeping. The man whose name is on the "stark warning" just paid about $19.8 billion to get in. The thing that would change that read is not the index level — it's whether Berkshire's next filing is back to net selling, and whether the money around the table is getting more patient or less. The crowd is watching the thermometer. The signal is in who's holding it, and how long they're willing to keep it warm.