Michael Burry: Artificial intelligence is heading in the wrong direction

Famous investor Michael Burry criticized the core technical approach of the artificial intelligence sector, arguing that its development has been on the wrong path from the very beginning. Burry, who shares his posts on the Substack platform, stated that AI development should have been started with a focus on reasoning rather than being language-focused. He suggests that the industry chose to solve language because it is scalable and investable, but claims this choice does not create true intelligence.

According to Burry, the artificial intelligence industry has fallen into a "parameter trap" that assumes larger models yield better results. He notes that even though companies scale up models by using more parameters, computing power, and data, this does not solve the underlying reasoning problem; it only makes language generation more convincing. Because hundreds of billions of dollars in infrastructure spending relies on the assumption that the current growth strategy will work, this situation could lead to serious financial consequences.

The investor also stated that he has been building short positions against AI trades for months. In addition to these positions opened in assets such as Nvidia, Tesla, Micron, and the iShares Semiconductor ETF, he shorted Micron stock on July 1, which had risen approximately 700 percent the previous year. Burry argues that the rally in semiconductor stocks stems from heavy AI spending by hyperscalers and claims this cycle is partly merely a reflexivity effect.

This technically grounded criticism by Burry presents a stronger stance compared to the arguments of other AI bears that focus solely on price valuation. In an environment where hyperscaler AI spending could reach 725 billion dollars in 2026 and Nvidia's market value is around 5.45 trillion dollars, Burry warns that the current scaling strategy is missing the mark.