Bloomberg Television9 min
Is AI a Bubble? History Offers Some Clues
Justin Fox, David Gura, Christina Ruffini
Interview
Market Sentiment and Bubble Indicators
- A recent dramatic sell-off has reignited debate regarding a potential AI bubble, though no investor or academic has a reliable track record for predicting the exact timing of a market peak or collapse.
- Historically, finance professors shifted from the "Rational Market" hypothesis (pre-1990s) to acknowledging market irrationality following the dot-com crash, a change partly driven by Robert Shiller's research.
- Shiller's Cyclically Adjusted Price-Earnings (CAPE) ratio for AI-related earnings briefly reached levels comparable to the 1999–2000 peak one month ago before receding.
- While some academic indicators suggest a 40% probability of a market collapse within three years, they simultaneously imply a 60% probability of stability.
Historical Context and Cyclical Risk
- The current bull market has persisted for approximately 17 years, covering a full generation of millennials and aligning with historical 15-year credit cycles identified by 19th-century banker John Mills.
- Regulatory laxity and the rotation of market authorities often lead to a collective forgetting of past downturns, causing the industry to repeat previous mistakes.
- Justin Fox, author of The Myth of the Rational Market (originally finished in 2007), noted that publishing an updated edition during a market downturn is an ominous sign for investors.
Debt Levels and Financial Vulnerability
- Unlike the 2008 financial crisis, which was exacerbated by high corporate and household debt, the dot-com bubble burst was less devastating economically due to lower leverage.
- Current concerns focus on significant government debt and specific high debt loads within the tech sector, particularly for companies building infrastructure like data centers.
- Warren Buffett's quote regarding "swimming naked" when the tide goes out highlights that the true extent of financial vulnerability often remains hidden until a market correction occurs.
AI Technology and Future Scenarios
- Fox characterizes AI as a fundamentally transformational technology comparable to railroads, the internet, or electrification, rather than a temporary trend.
- Two extreme future scenarios for AI are dismissed: the "AI extinction" scenario (bad for markets) and the Elon Musk-style "no money/no work" future (also bad for stocks but deemed unlikely).
- The prevailing view is that AI will likely supplement human tasks rather than replace all labor, though access and monetization may remain uneven.
- While stocks may hold increased value over the next 25 years, the intermediate 5-to-10-year horizon is expected to involve volatility and potential declines.
Valuation Comparisons and Cryptocurrency
- Recent massive IPOs, such as those involving Anthropic and OpenAI, show more "solidity" regarding capital than late-90s tech companies, though these valuations may be further along the investment cycle.
- Academic consensus increasingly accepts the term "bubble" to describe market phases where prices exceed fundamentals, despite the Efficient Market Hypothesis (Eugene Fama) arguing against predictability.
- Cryptocurrency markets, particularly Bitcoin, are noted for lacking underlying fundamentals, with Fama previously suggesting such assets could go to zero.
- Investors are advised to prepare for deflation of bubble valuations by managing debt and avoiding blind enthusiasm, even if the timing of the correction remains unknown.