Burry Is Still Short Nvidia. The Calls He Bought Explain Why

Days before Nvidia's earnings report, Michael Burry bought December call options on the stock — while adding to his short position in it. "I am not playing for gains here," the investor made famous by The Big Short wrote of the calls, calling them a hedge, not a bet. Then Nvidia reported the largest quarterly revenue in its history — $96.2 billion, beating the $92.3 billion consensus — and the shares jumped about 7% in premarket trading the next morning.

So the man who called the 2008 housing crash is shorting the best-performing company of the AI boom, losing money on that one position, and buying upside options on the same stock as insurance. It looks contradictory. It isn't. The structure of the trade is the story — and it says more about how to think about Nvidia than the latest earnings beat does. He closed his hedge fund in late 2025 and now runs a paid Substack where these moves go public.

Why buy calls on a stock you're shorting?

Start with how the short is built. Burry doesn't short Nvidia outright. He buys put options, which rise in value when the stock falls, and his maximum loss is capped at what he paid for them. Explaining an increase to the position earlier this year, he said he had considered shorting the stock outright but chose options "because the maximum loss is limited."

The December calls sit on top of that. Strikes in the mid-to-high $200s, bought for single-digit premiums. As he described it, the premium is "entirely offset" by the short and put positions behind it — which is to say the insurance is nearly free because the bearish wager already covers its cost. He put the combined short-and-put exposure at 3.5% to 4% of his portfolio. Asked essentially why, he gave the honest answer: he would not have bought the calls without a large short already in place, and his track record with this exact hedge-around-earnings maneuver is mixed.

That last admission is the part worth underlining. Burry is not posing as an oracle. He is paying a small, defined price to cap the event where Nvidia's earnings rip the short open — what he called "but a coin flip" heading in. Insurance, not prediction.

The actual bear case, in his words

The short itself is not a claim that Nvidia sells bad products. It is a claim about how the AI buildout is being financed.

Over the past year Burry has built the argument in public. He calls a new Nvidia framework — a partnership with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to mobilize more than $500 billion in financing for chip buyers — "a sign of desperation," adding it has "shades of Enron." His point: when a chipmaker helps customers borrow to buy its own chips, then stands ready to support the residual value of those chips, it has effectively turned compute into a financed asset class. Compute becomes a product that is sold, debt-funded, and warehoused — not something customers are paying for out of the returns it generates. Enron, in his telling, did the same trick with wholesale power.

The underlying research his former firm assembled leans on a spending-to-revenue gap: hyperscalers roughly quadrupled capex to near $400 billion a year, expected AI infrastructure spend runs to about $3 trillion over five years, while the revenue the end applications actually generate is still small — OpenAI is projected past $20 billion in annualized revenue, against all that. His analogy is the telecom bust — fiber built ahead of use, wholesale prices collapsing. He also argues Nvidia's near-monopoly is "short-lived" and that the company will plow its cash into expansion instead of returning it, investing "into and through the top of the bubble."

That is a coherent thesis. The problem for Burry is what the data has done since.

The scorecard cuts against him — on Nvidia

As of early August, six of his seven short bets were profitable, and Nvidia — his largest-name short — was the only loser. The other six: Palantir, Micron, Tesla, Caterpillar, Applied Materials, and the iShares Semiconductor ETF. His Palantir puts were up 35% by February; Oracle fell about 51% from its late-2025 peak. The AI complex is cracking in places. Just not the flagship.

And the flagship's numbers are extraordinary by any standard. Revenue has climbed from $57 billion, to $68.1 billion, to $81.6 billion, to $96.2 billion across the last four reported quarters, with data center revenue of $89 billion and a next-quarter guide of $108 billion. Gross margin sits near 75%. Market capitalization is around $5.2 trillion.

Here is where the trade gets instructive rather than merely loud. Nvidia had beaten expectations in five straight quarters — and heading into this report the stock had fallen after four of them. A company can print the best numbers in markets and still disappoint, because the price already assumed the growth. That is Burry's point in reverse: he says the stock looks cheap on a trailing P/E near 33 and will still fall, because the narrative has run ahead of what the financing structure can sustain. The market's answer, this week, was to jump 7% on an "inflection point" quote from Jensen Huang.

Both can't be fully right, but both are making legitimate observations about different things. Burry is betting on the financing tail. The market is betting on the revenue stream. Nvidia's own customers — Microsoft reportedly walking away from data center projects, Alibaba's chairman warning of an AI infrastructure bubble — are the tiebreaker data, and it is genuinely mixed.

Read the numbers before you read the headlines

One more thing to check before treating the famous short as a signal: the headline size is not the risk. The "$1.1 billion position" reported against Nvidia and Palantir is the notional value — the shares the puts control — not what Burry paid or the most he can lose. He reported the same way in 2023 on puts against the S&P 500 and Nasdaq marked at $886 million and $739 million of notional; regulators never disclosed what the contracts actually cost, which was almost certainly a small fraction of the face value. Same pattern again. A bought put's maximum loss is the premium, and the calls he layered on shrink even that tail.

This is the durable lesson for anyone who sees the headline and feels the pull to act on it, in either direction.

A great company is not automatically a great stock at any price — Nvidia's four straight post-earnings declines while beating expectations is the cleanest demonstration of that in markets right now. A defined-risk structure is how a contrarian stays alive while being early — Burry is losing on Nvidia, and the loss is contained, because he sized and hedged the position like insurance, not like a gamble. And when a famous bear and a record market both make sense in their own terms, the disagreement is real and neither is a mandate. The honest move is to know your own downside, because that part you control.

He spent the week astride the biggest earnings report of the cycle with a losing short, cheap insurance, and a public admission that his own hedging track record is mixed. That is not a man telling you where Nvidia is going. It is a man demonstrating how to be wrong gracefully — and it is worth more than the headline.