“Is the AI Bubble Becoming Reality?” A Market Driven by Leverage and Frenzy, with Repricing Imminent Amid Investment Euphoria

Accumulating AI investment excess distorting capital structures
Growth illusion created by supply leading demand
An imminent value repricing phase for the AI industry

Concerns over an “AI bubble” driven by overheated investment in the artificial intelligence sector continue to mount. Classic hallmarks of historical bubbles—overvaluation, excessive investment, and expanding leverage—are now compounded by the unique cost structure of generative AI and uncertainty over monetization, rapidly amplifying market unease. In particular, a structure in which share prices race ahead without verification of underlying demand, even as massive infrastructure spending and cash burn persist, is underscoring the fragility of the current exuberant phase.

Persistent Warning Signals on an AI Bubble

On the 15th (local time), global investment strategist Ruchir Sharma, chairman of Rockefeller International, outlined four key indicators for identifying an AI bubble in an op-ed published by the Financial Times: over-valuation, over-ownership, over-investment, and over-leverage. He noted that U.S. technology stocks have risen more than tenfold in real terms over the past 10 to 15 years, while AI-related equities have outperformed the broader market by more than 100 percent over the past two years. “Historical experience suggests that such surges are unmistakable warning signs of a bubble,” he wrote.

Sharma also pointed out that Big Tech firms, which for years operated with cash-rich balance sheets, are now issuing massive amounts of debt to finance the AI arms race. In recent months, Meta, Amazon, and Microsoft have emerged as the largest debt issuers, a development he described as a textbook signal of the late stage of a bubble cycle. He further warned that U.S. households are excessively concentrated in equities, with stocks now accounting for 52 percent of total household assets—an even higher level than during the 2000 dot-com bubble.

Another red flag lies in deteriorating corporate profitability and cash flow. According to Sharma, among the so-called “Magnificent Seven”—Apple, Microsoft, Amazon, Nvidia, Meta, Alphabet, and Tesla—Amazon, Meta, and Microsoft have now entered net debt positions, while free cash flow has plunged amid surging AI capital expenditures. Financial leverage across the broader market is also a concern. Assets under management in leveraged exchange-traded funds, which are easily accessible to retail investors, have increased sevenfold over the past decade to approximately 140 billion dollars.

Earlier, U.S. financial journalist Andrew Ross Sorkin also sounded the alarm on an AI bubble in his recently published book. His diagnosis of the stock market surge closely mirrors Sharma’s indicators. Sorkin identifies excessive leverage, abundant liquidity, and investor lunacy—the so-called “three Ls”—arguing that the AI boom is rooted in market distortions created by easy money. Microsoft founder Bill Gates has likewise joined the chorus. Speaking on the 8th at Abu Dhabi Finance Week, Gates warned that some highly valued AI companies could ultimately lose value. While acknowledging that AI is arguably the most important technology to date, he stressed that not every company can justify lofty valuations.

Rising Risk Aversion Toward AI Investments

The persistent AI bubble debate is increasingly reflected in equity market performance. On the 15th, the Dow Jones Industrial Average closed down 41.49 points, or 0.09 percent, at 48,416.56. The S&P 500 slipped 10.9 points, or 0.16 percent, to 6,816.51, while the tech-heavy Nasdaq Composite fell 137.76 points, or 0.59 percent, to 23,057.41. Although indices opened higher, selling pressure quickly emerged, reversing gains.

Analysts attribute part of the downturn to AI risk-off sentiment triggered by Broadcom last week. Broadcom slid 5.59 percent and has now fallen nearly 20 percent over three consecutive sessions. With AI sentiment failing to recover, the Philadelphia Semiconductor Index also declined 0.61 percent, marking a third straight day of losses. Palantir, another focal point of the AI bubble debate, dropped 3.97 dollars, or 2.12 percent, over the same period to close at 183.57 dollars. The stock, which “The Big Short” protagonist Michael Burry has disclosed he is shorting, had been trading at more than 230 times forward earnings estimates until last month—a level difficult to justify even under aggressive valuation models, despite surging profits.

Burry, who famously foresaw the 2008 global financial crisis, argues that pinpointing the timing of an AI bubble burst is virtually impossible. He believes that even after a collapse, it may take one to two years before it is recognized, and that the AI bubble could expand further from current levels. In a post on his paid Substack, he warned that “the current stock market is in a phase that could move toward an extremely large ‘blow-off top,’” adding that a peak could occur “at any time, even today or tomorrow.” He also highlighted what he sees as excess supply in the AI sector, arguing that technology firms are committing to massive data center construction, GPU orders, and multi-billion-dollar investment pledges without confirmed end demand—mistaken by investors as genuine consumption.

Economists, in particular, view the current AI surge as historically unprecedented. Capital flowing into AI today is estimated to be 17 times larger than the investment that poured into internet companies just before the dot-com bubble burst. Nvidia, with a market capitalization of 4.6 trillion dollars, now exceeds the economic size of every country except the United States, China, and Germany. There is little doubt that the AI investment frenzy has fueled not only U.S. equity market gains but also broader economic growth. However, experts broadly agree that uncertainties surrounding the medium- to long-term impact of such massive investment remain unusually high. At the macro level, it is unclear whether AI spending will meaningfully boost labor productivity and economic growth rates. At the micro level, it is difficult to predict when AI infrastructure investments—centered on data centers—will translate into sustainable revenue and profits. Even OpenAI CEO Sam Altman has acknowledged that “parts of the industry are feeling pretty bubbly right now,” underscoring why predictions of an eventual AI industry correction persist.

OpenAI Put to the Test Amid the Bubble Phase

As a result, expert attention is increasingly shifting toward which players will survive once the bubble phase passes. With the current overheating likely to approach its peak within one to two years at most, the decisive factor thereafter will be whether growth quality can be proven in tangible terms. From this perspective, OpenAI stands out as one of the most closely watched companies. At the Cerebral Valley AI Summit held in San Francisco on the 16th, an informal survey of roughly 300 founders and investors ranked OpenAI second—after Perplexity—among AI companies most likely to collapse first. While OpenAI has firmly established leadership in the chatbot AI market with ChatGPT, mounting losses driven by aggressive investment are fueling growing anxiety.

According to the Wall Street Journal, OpenAI is projected to generate 13 billion dollars in revenue this year while incurring 9 billion dollars in costs. Its cash burn rate is expected to remain at 57 percent in 2026 and 2027, before ballooning into a projected loss of 74 billion dollars in 2028. Market observers also point to long-term infrastructure commitments totaling between 1 trillion and 1.4 trillion dollars, as well as a circular transaction structure centered on OpenAI—linking Nvidia, AMD, CoreWeave, AWS, and Microsoft—as factors likely to further strain finances. OpenAI is also facing criticism that its valuation is excessively inflated relative to revenue. According to U.S. market research firm CB Insights, OpenAI’s current valuation stands at approximately 300 billion dollars.

Still, some analysts argue that an AI bubble burst would not necessarily signal a retreat of the industry itself. John Turner, an economist at Queen’s University Belfast, notes that after the dot-com bubble collapsed, the number of academic papers in electronics, computer science, and scientific fields actually increased, while the spread of internet and mobile technologies continued unabated. In other words, research momentum persisted independently of the industry shock.

Scholars suggest a similar pattern could emerge even if an AI bubble bursts. Brent Goldfarb, an economist at the University of Maryland, argues that while an AI crash could trigger mass layoffs at startups and younger firms—leaving giants such as Google and Nvidia as likely survivors—AI talent could migrate to other areas, eventually sparking new waves of innovation. Turner echoes this view, recalling that after the collapse of the U.K. bicycle market in 1896, displaced bicycle engineers went on to develop the motorcycle, automobile, and aviation industries. Likewise, the internet spread even more rapidly after the dot-com bubble burst.

David Kirsch, a technology historian at the University of Maryland, shares a similar perspective. He suggests that if AI professionals displaced by an industry downturn return to academia, they could develop technologies far more beneficial to society. Such a shift, he argues, could give rise to breakthroughs akin to DeepMind’s AlphaFold—an AI tool that helped solve the 50-year-old challenge of protein structure prediction.