The AI Bubble: A Dot-Com Analogy with Broader Implications

Jared Bernstein, a Stanford Institute for Economic Policy Research policy fellow and former Biden CEA chairman, recently offered a stark assessment of the current AI investment landscape, declaring it the third economic bubble of the century. Speaking with CNBC's "Squawk Box" hosts, Bernstein and his co-author Ryan Cummings contend that despite the genuine technological advancements and profitability of some key players, the sector exhibits classic bubble characteristics, potentially leading to a significant negative wealth effect should it burst.

Bernstein outlined several markers pointing to an AI bubble, chief among them the "rapidly rising asset prices" and "very extreme valuations" seen in the AI space. He highlighted Nvidia, trading at roughly 55 times earnings, as a prime example. More critically, he noted a historical comparison: "The share of the economy devoted to AI investment is nearly a third greater than the share of the economy devoted to internet-related investments back during the dot-com bubble." This quantitative measure suggests an unprecedented level of capital flowing into AI.

The interviewers challenged Bernstein's thesis by pointing out a significant difference from the dot-com era: many of today's leading AI companies, the "Mag 7" like Microsoft, Meta, and Amazon, are highly profitable, unlike the speculative, often profitless startups of the late 1990s. This distinction, they argued, might suggest a more robust foundation for the current AI surge.

Bernstein acknowledged the profitability of these tech giants but quickly pivoted to a crucial nuance. "You can have a bubble with completely profitable and technologically innovative firms investing into the bubble," he stated. He elaborated that the true definition of a bubble lies in "the gap between the level of speculation, the level of investment, and credible, reasonable expectations of future profits is extremely wide." For many of these companies, their vast profits stem from established business lines, ads, cloud services, e-commerce, rather than directly from their burgeoning AI investments. The AI-specific investments, he argued, represent a smaller, more speculative share of their overall financial picture.

This distinction is vital. While companies like Nvidia directly benefit from AI's infrastructure demands, the broader tech landscape's AI investments are often speculative bets on future, unproven revenue streams. Open AI, for instance, projected a trillion dollars in investment this year against only $13 billion in AI revenue. This divergence between investment scale and current AI-derived profitability fuels Bernstein's concern.

Related Reading

• Former Intel CEO Declares "Of Course" We're In An AI Bubble
• The AI Economy: Bubble or Breakthrough Demand?
• AI Stocks: Navigating Breakouts and Breakdowns Amidst Market Euphoria

A primary worry for Bernstein is the potential for a "really large negative wealth effect" if the AI bubble pops, estimating it could amount to "hundreds of billions of dollars." Unlike the dot-com bust, which saw a minimal contraction in GDP despite significant market losses, the current landscape involves a broader base of "regular mom and pop investors" whose portfolios could be heavily impacted. Such a widespread loss of wealth would inevitably dampen consumer spending, feeding into existing "underlying fragilities in the real economy." The scale of current investment, relative to the dot-com bubble, suggests a potentially more impactful downturn on household finances.

The current market enthusiasm for AI, driven by transformative technology, echoes past speculative frenzies. However, Bernstein’s analysis emphasizes that even fundamentally sound companies can engage in over-speculation within a rapidly evolving sector. The sheer volume of capital chasing AI, coupled with a widening chasm between investment and immediately attributable AI profits, paints a picture of a market susceptible to a significant correction.