366. This Economist Predicted the Last Crisis. What’s the Next One?

Could the current AI boom be a “parlay bet” where massive capital expenditures are gambling on near-term productivity gains that may never materialize? Gary Gensler, a former chair of both the SEC and CFTC and current MIT professor, suggests that we are seeing a familiar historical pattern: a massive investment phase in a general-purpose technology that creates a bubble before eventually finding its equilibrium. With AI spending potentially hitting 3% of the U.S. GDP, Gensler warns that if revenues don’t catch up to the infrastructure costs, the resulting plateau could trigger a significant market correction.

The conversation delves into the structural risks of the American economy, specifically the unsustainable trajectory of federal debt and the concentration of financial power. Gensler argues that the U.S. has become overly reliant on the world’s willingness to buy dollar-denominated assets, a trend that cannot continue indefinitely without real interest rates rising. He notes a widening gap between empirical economic analysis and political will, where “entitlement” spending dominates the budget, leaving policymakers hamstrung by party loyalty and a lack of public consensus for reform.

Looking toward the future, Gensler predicts a “transitional period” of political angst and social disruption throughout the 2030s and 40s. He views AI not as a total job-killer, but as a disruptor of specific tasks that will reshape the labor market. While he remains an optimist about the long-term utility of AI, he cautions against the “winner-take-all” model and the emergence of “good enough” AI from China, which could commoditize the market and erode the high profit margins currently enjoyed by U.S. tech giants.

Surprising Insights

• The “Volkswagen” of AI: While the U.S. is building the “Maseratis and Ferraris” of AI models, China is developing “good enough” models that may eventually dominate the market due to lower costs and high utility for most general tasks.
• The Warren Buffett Index: The U.S. stock market is currently hovering around 235% of GDP—an all-time high compared to the 25-year average of roughly 110-120%.
• Infrastructure vs. Revenue: There is a staggering gap in the AI sector where capital expenditures are roughly $750 billion, while actual revenues are estimated to be only $150–$200 billion.
• Insider Trading and Trust: Contrary to the academic argument that allowing insider trading in prediction markets makes them more efficient, Gensler argues it destroys the “public good” of trust, which ultimately raises the cost of capital for everyone.

Practical Takeaways

• Focus on Task, Not Job: When assessing your career risk regarding AI, analyze which specific tasks within your role can be automated rather than fearing the loss of the entire job category.
• Beware of “Sentiment-Only” Assets: Evaluate investments based on underlying fundamentals rather than market sentiment; assets that rely purely on sentiment (like many crypto assets) are the most susceptible to total washouts.
• Avoid Vendor Lock-in: For businesses implementing AI, pursue “orchestration”—using multiple models to avoid being dominated by a single provider who can arbitrarily raise prices.
• Practice the “Premortem”: To protect a project or investment, imagine it has already failed and work backward to identify the causes. This allows you to create “insurance policies” and amend your strategy before the launch.