Michael Burry has a track record of seeing financial illusions before they go away. Most people wrote him off as reckless when he predicted the 2008 housing market catastrophe. His attention has now turned to artificial intelligence, where billions of dollars are being invested by those drawn to the prospect of a technological revolution that does not yet have a clear economic roadmap. Though this time the story is presented in algorithms rather than domain names, the optimism feels remarkably similar to the dot-com euphoria.
AI has emerged as Wall Street’s and Silicon Valley’s new currency of aspiration. Building data centers and training models that use electricity at rates previously reserved for industrial plants is costing enormous sums of money for every major tech company, from Microsoft to Meta. The chip maker at the heart of this craze, Nvidia, has seen its value soar above $1 trillion. However, the economic underpinnings of this joy seem incredibly brittle.
Burry’s doubt breaks through the cacophony. He contends that predicted value, not actual profit, is the foundation of the entire AI investment system. It is said that he told a coworker, “It’s faith, not fundamentals.” Given that OpenAI, the firm behind ChatGPT, is reportedly losing billions of dollars a year while spending over $12 billion on inference—the process of producing AI outputs—his argument is especially compelling. In contrast, Amazon’s losses were significantly lower than its potential revenue, even in the early years of its unprofitability.
Bio & Background
| Category | Details |
|---|---|
| Name | Michael Burry |
| Profession | Investor, Hedge Fund Manager |
| Known For | Founder of Scion Capital and early predictor of the 2008 housing market crash |
| Education | M.D., Vanderbilt University School of Medicine; B.A., Economics, University of California, Los Angeles |
| Notable Investments | Bet against subprime mortgage bonds (2008), recent short positions against AI-related stocks such as Nvidia and Palantir |
| Current Focus | Founder of Scion Asset Management; critical of AI valuations and data center spending |
| Reference | https://www.bloomberg.com/ |
A peculiar equilibrium has resulted from this misalignment between story and data: investors are placing bets on growth that is impossible to measure. AI-focused data center capacity is expected to reach 117 gigawatts by 2030, according to venture capital firm Accel. This is a $4 trillion investment that will require over $3 trillion in revenue to be recouped. These predictions are based on the idea that AI would transform company productivity, but new research indicates otherwise. Researchers from MIT discovered that 95% of businesses that used AI tools did not get any discernible return on their investment.
However, the story continues to be too enticing for executives to ignore. In an interview, Lyft CEO David Risher acknowledged that “we’re definitely in a financial bubble.” “However, nobody desires to fall behind.” The emotional reasoning behind the boom is encapsulated in his comment. Similar to previous gold rushes, investors are more focused on making sure they don’t miss the strike than they are on the long-term yield. The ensuing zeal has caused valuations to rise well above what actual adoption can sustain.
Jarek Kutylowski, the CEO of DeepL, has voiced similar concerns, referring to many of the current AI valuations as “exaggerated.” Hovhannes Avoyan, the founder of Picsart, cautions against businesses that make “vibe revenue,” a term used to describe firms that are financed more by perception than by revenues. Their comments draw attention to an unsettling reality: the commercial promise of AI frequently feels more like spectacle than change. It appears that even CEOs who develop AI technologies are unsure if their organizations can keep up with the pace.
Burry’s wager is not the only one that casts doubt on this hope. Across the Atlantic, hedge funds are subtly preparing for what some refer to as the “AI unwind.” They cite businesses like Anthropic and OpenAI, whose revenue is falling behind infrastructure expenditures. According to reports, OpenAI’s operating expenses are many times higher than its revenue in an average quarter. Critics contend that such an imbalance cannot persist indefinitely. The illusion of endless profitability may come crashing down the moment investor patience wanes.
The actions of Big Tech exacerbate the conundrum. Companies with enormous cash reserves, like Google, Microsoft, and Amazon, are funding their AI operations mostly with internal capital rather than debt. However, their balance sheets show an incredible rate of spending. AI has been a major factor in Microsoft’s quarterly capital spending, which have topped $30 billion. Even then, only a small percentage of these companies provide income related to AI, which begs the question: why conceal the figures if AI is so lucrative?
The ambiguity is highlighted by the advertising conundrum. Subscriptions are still the only reliable business model for generative AI technologies. The introduction of advertisements has mostly failed. Before ending the initiative, Perplexity AI, one of the first companies to test AI-driven advertising, only produced $20,000. Because AI’s uncertain results cannot ensure the dependability of ad placement that platforms like Google and Meta have refined over decades, traditional marketers continue to be cautious.
The familiarity of this moment is what makes it so fascinating. The enthusiasm is reminiscent of the internet boom, when investors thought every website would transform the economy. Despite its flaws, the comparison is very instructive. Data traffic was the new oil in the early 2000s. It’s compute power now. Both influenced a generation of investors to ignore the economics of return, are expensive to generate, and are challenging to monetize.
However, the AI boom is quite modern in that it is driven by institutional giants rather than speculative consumers. As if it were inevitable, national governments, pension funds, and sovereign wealth funds are investing in AI infrastructure. The consequences could be far more widespread than the dot-com catastrophe if that faith falters. Millions of people’s stock portfolios now depend on whether AI’s potential is realized more quickly than its costs.
Nevertheless, there is yet hope. Every revolutionary technology, according to proponents, goes through a “hype-to-reality” curve where early losses are the cost of discovery. Before becoming essential sectors, the internet, electricity, and even the 19th-century railroads were all overhyped fantasies. According to such viewpoint, the volatility of AI may just be the turbulence preceding the rise. Skeptics point out that the distinction is that previous technologies yielded observable benefits long before they required this level of investment.
The unromantic foundation of Burry’s argument is that advancement does not equate to financial gain. His cautions are reminiscent of his 2008 position—unambiguous, counterintuitive, and unnervingly realistic. He contends that investor psychology, where fear of exclusion trumps financial reasoning, is reflected in the AI gold rush. He once remarked, “The market is driven by excitement, not evidence.” In a world where billion-dollar corporations are unable to define their products beyond “intelligence,” the remark is especially pertinent.
