"The Big Short" goes crazy! Michael Burry targets AI faith from the grassroots level of technology: when the parameter trap collides with the production capacity peak, the AI bull market welcomes a "liquidation moment"

Michael Burry, known as the "Big Short," has been posting pessimistic comments on his Substack subscription platform, predicting that doomsday is near. As global funds continue to flock to the AI computing power infrastructure theme, he has been aggressively shorting popular AI technology stocks. The prototype character in the movie "The Big Short," Michael Burry, has recently been focusing more on shorting operations related to AI computing power infrastructure and AI semiconductor capital expenditure, such as expressing bearish positions on NVIDIA Corporation, Tesla, Inc., Micron, Applied Materials, and the iShares Semiconductor ETF through options or short positions. This globally renowned investor, who gained fame for accurately predicting the 2008 US real estate collapse and subprime crisis, has recently been actively shorting AI themes. Market participants interpret his actions as extremely crowded short positions on storage chip themes, semiconductor manufacturing equipment, and popular stocks related to AI GPUs and AI ASICs. He also has previously disclosed bearish options positions on NVIDIA Corporation and Palantir through his 13F filings. As the Korean stock market plunged to a halt on Monday and global stock markets with AI computing-related popular technology stocks experienced violent fluctuations, some investors are starting to agree with the narrative that the AI bubble is on the verge of bursting, and are joining Burry's bearish camp on the AI theme. Burry views SK Hynix and Samsung Electronics' unprecedented huge expansion plans for storage chips as a key signal that the AI investment cycle is "approaching its end." He sees the Korean projects as a signal of "peak capital expenditure" rather than "immediate capacity signals," and believes it marks the "beginning of the end" of AI prosperity. His judgment is not that these factories will immediately lead to oversupply next year, but is based on the classic spider web cycle in the semiconductor industry: the most optimistic demand forecasts, the highest chip prices, and the most relaxed financing conditions often lead all manufacturers to announce huge expansions at the peak of the cycle; as advanced wafer fabs require several years from planning, land, power, cleanroom construction, semiconductor equipment large-scale import to production, the real supply peak usually arrives when market demand has cooled. Burry's latest bearish positions on AI and his extremely pessimistic bearish stance on the AI super-bull market are in stark contrast to the bullish stance taken by Wall Street financial giants such as Bank of America Corp, Goldman Sachs Group, Inc., and Nomura. These top global financial giants unanimously believe that as enterprise IT budgets flow increasingly towards the AI intelligent body and AI reasoning power budgets, along with the launch of OpenAI GPT-5.6 and ChatGPT Work intelligent bodies, and the prospect of Anthropic transitioning from long-term losses to explosive overall profits, the global demand for AI computing power is nearly endless. Burry warns of an "AI parameter trap" and firmly targets the AI investment frenzy with his short positions. Burry, known as the "Big Short," quoted a classic line from the character Joker in Tim Burton's version of Batman in his recent tweet: "Dancing with the devil in the pale moonlight." He refers to the AI boom as an "addiction of the masses," predicting that it "may die a thousand cuts." On July 1, when the stock price of the US storage chip manufacturing giant Micron Technology, Inc. reached $1,051.87, he chose to short the stock, even though the stock had risen by nearly 700% in the past year. On July 10, he published a more specific and pessimistic view on the global AI investment frenzy on Substack, which goes beyond concerns about the bursting of the AI bubble and the expensive valuation of AI semiconductor themes. This is an attack on the underlying technology base of the entire artificial intelligence industry. Burry argues that the development path of artificial intelligence has taken the wrong direction from the beginning, and is now trapped in it. The argument of this article revolves around what Burry calls "going wrong at the beginning." He believes that the development of artificial intelligence chose language first instead of prioritizing reasoning; the industry has been paying the price for this choice ever since, but has not fully acknowledged it. He builds this argument around what he calls the "Ballard test." This is a philosophical case involving a character named Melville Ballard; who had profound reasoning ability before acquiring strong language skills. Burry uses this to present a specific viewpoint: true understanding does not necessarily require language. Language is the output form of intelligence, not the source of intelligence. "We mistook the output of artificial intelligence - language - as its engine - reasoning," Burry wrote on the Substack subscription platform. According to his interpretation of the technological route of the artificial intelligence industry, the industry found language to be a problem that could be addressed, and thus optimized around language; not because language leads to general artificial intelligence, but because language has significant scalability and is easier to obtain financing from the large tech industry. Burry believes that the ability of models to generate text is becoming stronger. Investors reward this. The industry continues to move forward along this path. However, in this process, the difference between generating language and actually reasoning has been gradually overlooked. What does Burry's concept of an "AI parameter trap" mean? The so-called "parameter trap" refers to an industry starting to believe that larger scale equals better performance for AI large models: more model parameters, larger scale AI computing infrastructure, more input data, and larger models. Each scale upgrade visibly improves benchmark test scores and performance. However, Burry's view is that enlarging language model scales does not solve the underlying logical reasoning problem, but rather makes the simulation seem more convincing to users in terms of language output. This has real financial consequences. Companies investing billions of dollars in AI computing infrastructure are betting on the notion that scale expansion will ultimately achieve the goal of general artificial intelligence. According to TheStreet, AI support expenditures for super-large-scale cloud computing companies in 2026 could reach $725 billion. The Philadelphia Semiconductor Index has risen by 88% since the beginning of the year. NVIDIA Corporation has a market capitalization of approximately $5.45 trillion, with a price-to-earnings ratio of 43 over the past 12 months. All these valuations and expenditures are based on the assumption that the current scale expansion path will continue to be effective in the long term. Burry believes that this path may not be feasible. If the entire industry is investing such a huge amount of money in scale, just to improve something that is not the true goal, then the economic logic of the entire transaction will be completely different. This Substack article did not appear out of nowhere. For months, Burry has been building short positions against popular AI technology stocks. His disclosed short targets include NVIDIA Corporation, Tesla, Inc., Micron Technology, Inc., Applied Materials, Carter's Incorporated, and the iShares Semiconductor ETF. As previously disclosed by the media, his short position on Micron on July 1 was established after the stock had already risen by nearly 700%. His analysis of the overvaluation of the AI semiconductor sector points out that the Philadelphia Semiconductor Index is approaching peak levels within the expected price-to-earnings range over the past 15 years. He argues that the rise in the prices of AI semiconductor stocks is because super-large-scale cloud computing companies are investing heavily in AI training/inference power clusters; and these cloud computing companies continue to invest heavily because they urgently need these AI semiconductors and core infrastructure hardware related to AI computing, thus driving the continuous rise of global chip stocks. This feedback loop may look like genuine demand, but according to Burry, some of the growth actually stems from market reflexivity rather than cash flows from AI already being fully realized. His recent shorts on NVIDIA Corporation, Micron, Applied Materials, ETFs on semiconductor exchanges, Tesla, Inc. and Carter's Incorporated actually cover multiple aspects of the AI capital cycle - from core accelerators of AI clusters (AI GPU/AI ASIC/TPU), storage chips, semiconductor manufacturing equipment systems, to the capital goods related to self-driving narratives and data center construction, rather than just targeting a specific AI-related tech company. --- The translation above seems quite long, so let me know if you would like a summary or if there are any parts you want me to focus on.