Jeff Yass argues there is no such thing as a stock market bubble — and AI bears are making the same old mistake

Jeff Yass: The "Bubble" Diagnosis Misunderstands Markets Under Uncertainty

Jeff Yass, founder of Susquehanna, rejects the entire premise that AI stocks are in a bubble — and argues that calling bubbles is itself a fundamental misdiagnosis of how markets work.

The core claim: when share prices fall, it isn't a delusion correcting. It's uncertainty resolving. The word "bubble" assumes a true price exists that hype has obscured, and that market crashes reveal reality punishing irrationality. Yass says this misunderstands prices entirely.

Under uncertainty, there is no true price to lose sight of. A price is the market's best guess at value — every possible future weighted by its likelihood, made by millions of people with their own money on the line. The market isn't always right. It's just a much better instrument for predicting value than any pundit.

The 1999 logic applied to 2026

Yass walks through the math with a hypothetical 1999 internet stock trading at $100. If it becomes Amazon, it hits $1,000. If it becomes Pets.com, it goes to zero. Suppose the market assigns a 15% probability to the Amazon outcome. The expected value is $150 — meaning $100 is a sensible bet, not an overpriced gamble.

If the stock ultimately goes to zero because the 85% outcome came in, it wasn't irrational to buy at $100. Both the 15% chance and the most likely outcome of zero were true simultaneously. Most pundits can hold only one of those ideas at once.

Amazon itself fell over 90% from its 1990 peak. By the crash-based logic that equates losses with bubbles, Amazon would have been called a bubble. Yet investors who held the sector long enough came out ahead, because a handful of spectacular winners paid for all the failures.

Hendrik Bessemer of Arizona State University studied every US stock since 1926. Most returned less than treasury bills over their lifetimes. About 4% of companies produced all the market's net gains. If every losing bet counts as a bubble, the stock market has always been one.

The bubble-caller's free option

Calling a bubble is a bet with asymmetric payoff: if the market crashes, you're famous. If it doesn't, you shift to the next prediction. Yass notes that the people most famous for calling bubbles end up with roughly a 1% track record overall.

The challenge isn't whether bubble-callers eventually get one right. It's that timing, peak prices, and drawdown magnitude are nearly impossible to predict. When the fall comes — and prices will probably come down — it won't prove anyone was irrational. It will prove only that uncertainty has resolved differently than expected.

His rule: if you think AI is a bubble, buy puts. The premium caps your loss. If you're right, you'll get rich and will have earned it. If you won't buy them, then when the market falls, don't call the winners irrational. The market called it first.

The arrogance problem

Being rational isn't about being right. It's about being smart and humble about uncertainty. Being irrational is being dumb and arrogant about it. It's arrogant to believe you know more than the market and foolish to listen to a pundit who says he does.

AI is where the internet was in 1999. Nobody knows how big it will be, which firms will win, or whether winners will be 100 times or 10,000 times bigger. That uncertainty makes prices swing hard in both directions. A small chance of an enormous payoff is worth a lot today. To anyone not thinking in probabilities, the entire field looks too expensive.