Can big bear Michael Burry win this short? Scion Asset Management, led by Burry, disclosed its third quarter 13F report two weeks in advance, which is rare. Approximately 80% of Scion Fund's positions are concentrated in short selling Nvidia and Palantir, with a nominal value exceeding $1 billion, making it the most closely watched hot spot in today's market. I searched for short selling cases of Michael Burry in the past few years: Short selling Tesla on 20-21 resulted in failure and losses of hundreds of millions of dollars. Short selling ARKK and technology stock indices in 2021. Short selling too early resulted in partial failure, and by the time the market experienced an overall downturn in 2022, Burry had already significantly reduced his holdings, ultimately leading to a small loss; Short selling S&P 500 and Nasdaq 100 in Q2 of 2023, resulting in failure and a loss of one billion US dollars; Short selling semiconductor in Q3 of 2023, resulting in failure; In Q1 of the year 25, I worked on air concept stocks and Nvidia. The timing was actually right, but the final result was still a small loss. https://((x.com))/i/grok/share/3lG9jEMBH05h8avqHKIG7BUTm My observation is that Burry's short selling can win big if it is a market shock caused by macro risks, such as short selling subprime mortgages since 2006. However, if it is aimed at individual stocks and specific industrial targets, it often ignores the fundamentals and only makes bets based on the foam warning, and the results are not very good. For example, if we short the US stock market this time due to the squeeze on risky assets caused by tight liquidity, the logic is actually reasonable, because the government shutdown has not yet ended and it is unknown when liquidity will recover. During this period, the spread of liquidity tension will squeeze all risky assets, and the US stock market will kill valuations. https://((x.com))/qinbafrank/status/1985686986703978755? s=46&t=k6rimWsEbo2D2tXolYcM-A After reading Burry's arguments, we can actually discuss them in detail: 1) The capital expenditure situation 25 years ago is very different from the current capital expenditure situation, and the current big technology capital expenditure is not blindly building computing power. From the consumption of tokens, it can be seen that the consumption of tokens (tokens, units of measurement for large model computing power) is an important indicator for measuring the implementation of AI applications. Currently, the consumption of tokens is growing at a rate of over 100 times. Let me give you two examples: In October 2025, Google's daily average Tokes processing volume was 1.3 trillion Tokens/month, an increase of 134 times compared to 9.7 trillion Tokens/month in May 2024. For example, in May 2025, the daily average number of Tokens calls of ByteDance's bean bag model was 16.4 trillion, an increase of 137 times compared with 120 billion in May 2024. The monthly token consumption of OpenAI has also increased dozens of times in the past year. So the core of the giant's massive computing power is the soaring consumption of tokens. We should look at computing power from the perspective of electricity, not from the perspective of cables. Electricity has been invented for over a hundred years, electricity consumption in various countries is still rapidly increasing this year, and new power stations are still being constructed. And currently, AI applications have not yet exploded on a large scale. 2) Burry's comparison of the business growth rate of cloud computing vendors in 2018 with the current business growth rate actually overlooks one point. In 2018, the market gave Google and Amazon a PE that was much higher than the current level. In 2018, Google PE was 40-50 times higher, and Amazon was even higher. The growth rate of their cloud computing business has slowed down, but the valuation level is only half of what it was at that time. The valuation level given to them by the market at present has already taken into account the decline in growth rate. Another important point is that the business scale has multiplied several times compared to 8 years ago, and the larger the scale, the harder it is to maintain high growth rate. What I actually want to talk about is how to view the adjustment of the US stock market correctly? 1) The tight market liquidity is a high probability event that will drive the overall risk asset correction. We've talked about this before 2) The AI mainline in the US stock market has risen too much, and it is reasonable to use this liquidity shock to adjust and kill off a wave of valuation. 3) Without the impact of larger macro events, this wave of US stock expectations is a minor adjustment. 4) But the constant cry of foam collapse is actually slightly excessive. As we have been discussing before, different macro risk levels have different impacts on the market. If we really want to say that the AI main line foam has broken, it will be too late until most of the big technology companies' financial reports fall short of expectations, and the company's operating profit growth has been far behind the growth of capital. Simply put, it means looking at the valuation adjustment of the market at a small level, not at the overall significant collapse.