Why Michael Burry's Nvidia and Palantir Put Exit Isn't the Signal You Think It Is

Michael Burry just spent a month closing his bearish option bets on the two most-hyped AI names, and the first instinct is to read it as a verdict on Nvidia and Palantir. It isn't. The man who made his name shorting the 2008 housing market told his readers he "pulled in risk" during September, raised his cash, and is now "happy to sit on" it while the market does whatever it does this fall. In his words: "This will be an interesting market this fall."

The details matter more than the headline. Burry exited his December 2026 put options on Nvidia (NVDA) and Palantir (PLTR), and he was explicit about why: he sold them to avoid the relentless time-based decay that eats an option's value as it nears expiry. He did not roll them into later-dated contracts. And critically, he did not abandon his bearish view — he still holds 2027 puts on Palantir and on the Invesco QQQ Trust, and he still lists Nvidia, Palantir, Oracle, Nebius and the iShares Semiconductor ETF among his largest shorts.

That is the first lesson worth pulling out, because it reframes everything. What Burry changed was the timing of his exposure, not his thesis.

What "exiting a put to avoid time decay" actually means

A put option gives the buyer the right to sell a stock at a fixed price by a set date. Burry used them to bet stocks fall. Such a position has three distinct costs: how wrong the direction has to be, how far the price has to move, and how much time it has to get there. That third cost is the quiet one. As an option approaches expiration, its value steadily bleeds away all on its own — a force options traders call theta. Even if a bearish call is correct, a near-dated put that doesn't reach its target fast enough still loses money.

That mechanism, not a change of heart, is what drove Burry's September move. Observing that the short position and its economics have come apart is not the same as concluding the stock is going up. He kept the longer-dated puts precisely because they have the room to work. Closing December puts while holding 2027 puts is the definition of a timing adjustment, not a reversal. Headlines that frame it as Burry "turning bullish on AI" would be wrong; so would the mirror-image claim that he knows the top is in.

A quarterly, lagged, notional headline

There is a second reason not to over-weight any single reading of Burry's positions: the disclosure itself is stale and smaller than it looks. His bets reach the public through 13F filings, the form big money managers must submit each quarter. Those filings arrive weeks after the trades happen, so by the time you hear about a position it is already history — the market may have repriced it fully in the gap.

The size is also routinely overstated. At the peak of the noise, coverage described a "$1.1 billion" bet against Nvidia and Palantir, figures that reflect the notional value of the puts — the face amount of stock they reference. The actual cash Burry put at risk, the premium he paid for the contracts, was far smaller, as options math quickly showed. A famous bear with a big notional book is still one fund in a market where institutional money and aggregate liquidity dwarf anything a single manager does. What moves Nvidia and Palantir is not the order flow of one hedge fund.

Where the AI trade's direction really comes from

So if not Burry's order book, what? The two stocks live on very different engines. Nvidia's is real earnings and a capex supercycle: hyperscalers are pouring hundreds of billions into AI data centers, and that spending is the demand that fills Nvidia's order book. Palantir's is a valuation story: for stretches its forward price-to-earnings ratio has exceeded 275, a multiple that embeds years of near-perfect growth, not just this quarter's results. One is a business cycle question, the other a sentiment and expectations question.

My own frame for both is the same map I draw for every asset: one global cycle driven by aggregate liquidity and financial conditions, overlaid on the longer secular forces of debt and technology. Apply it here and the takeaway is humbling for anyone looking for a celebrity signal. The direction of these names will be set by the liquidity cycle and by whether AI capex keeps compounding even as the market debates a bubble — not by a quarterly snapshot of one manager's puts. When a narrative as loud as "AI is a bubble" collides with a capex cycle that big, the honest position is to check the lead indicators rather than to take sides on conviction alone.

And note the flip side, which is easy to miss when everyone fixates on a famous bear. Burry holding cash and trimming risk is itself a sentiment data point — a prominent skeptic choosing to watch rather than press his case. But sentiment, as extremes go, cuts both ways. A much-discussed short that gets partially closed, on insiders also selling their own AI winners, and yet the stocks still hold — that is not a market screaming "top." It is a market absorbing sellers, which is closer to the signature of a durable trend than a crumbling one.

The usable lesson

Strip away the drama and Burry's September letter leaves a retail investor with two genuinely useful things. First, options are instruments with multiple costs, and a manager can be right about direction and still exit a position for mechanical reasons — so never read a disclosed trade as a pure opinion about the company. Second, the value of a celebrity bear's positioning is mostly entertainment. The question that actually decides whether you should own Nvidia or Palantir is where the liquidity cycle and the AI earnings cycle point next, and that is a question the data — not any single investor's puts — has to answer.