"The Big Short" Palantir's Short Position Surges Nearly 30% Overnight, But Its Bearish Thesis Remains Untested
August 5th, after releasing its financial report, the US stock Palantir surged nearly 30%, becoming one of the most dazzling AI software stocks of the day. The company's second-quarter revenue increased by 93% year-on-year to about $1.935 billion, with adjusted earnings per share of $0.41, while also raising its full-year revenue guidance to $8.15 billion to $8.158 billion. The US stock market also cooperated fully, with the S&P 500 and the Dow both hitting new closing highs, and the Nasdaq soaring by 2.6%, showing a significant return of risk appetite.
It is reported that Michael Burry, one of the prototypes in the movie "The Big Short," has been placing Palantir at the core of the AI bubble narrative in recent months. Scion Asset Management's third-quarter 13F filing last year showed that it held PLTR put options corresponding to 5 million shares of underlying stock, with a declared value of $912.1 million, accounting for approximately 66% of the disclosed portfolio. This figure was once interpreted by the market as "a $900 million short on Palantir."
However, there is a key distinction here: the $912.1 million in the 13F is the disclosed amount calculated based on the value of the underlying stock, not the actual option principal paid by Burry. Burry later stated that the actual premium was about $9.2 million, corresponding to about 50,000 contracts of $50 strike price puts expiring in 2027. That is, the maximum loss is the premium, and the return on the bet comes from the stock price significantly falling below the strike price.
In April of this year, Burry confirmed on Substack that he still holds long-term put options on Palantir, including $50 puts expiring in June 2027 and $100 puts expiring in December 2026. His judgment at the time was straightforward: Palantir's intrinsic value is far below $50 per share.
The issue is that last night's market temporarily stood on the other side. The data Palantir presented this time was strong enough, with US commercial business growing by 149% year-on-year and US government business growing by 90%. The logic of companies like OpenAI and Anthropic impacting the software layer that the market was previously concerned about was at least not realized in this earnings report. CEO Alex Karp also redirected the narrative to "AI sovereignty": enterprise and government clients need to control their own data, models, and workflows, and Palantir happens to be at the forefront of this.
However, this also explains why Burry's short logic has not completely disappeared due to a large bullish candlestick. Palantir's current risk lies not in poor short-term performance, but in the market having already given it a very high margin of error. As long as the growth continues to explode, the valuation can be sustained; once US commercial growth slows, overseas expansion is hindered, or enterprise AI budgets shift to cheaper models and cloud vendor solutions, the valuation compression will occur faster than profit downgrades.
Currently, the specific unrealized loss data for Burry is unable to be precisely calculated based on publicly available information. Scion has already deregistered, and there is no 13F filing available for their current positions. Burry has disclosed the direction of the holdings and some strike prices but has not revealed the latest quantity and cost. What can be confirmed is that the loss cannot be calculated by directly applying the stock price increase to the nominal amount of $912.1 million. The options premium and rolling cost are the actual areas under pressure.