So are we in an AI bubble? Here are clues to look for.

Key Takeaways Copied to clipboard!

• Economists have identified four clues for detecting bubbles—high valuations, volatility, issuance, and acceleration—though these indicators only predict popped bubbles about 60% of the time, challenging Nobel laureate Eugene Fama's skepticism.  Copied to clipboard!
• The potential AI bubble's economic damage upon popping might be less severe than the 2008 crisis because AI companies are primarily borrowing from private credit rather than directly from traditional banks.  Copied to clipboard!
• A provocative theory suggests that some bubbles, like the dot-com era's investment in fiber optics, might yield positive societal benefits (silver linings) by funding necessary research and development that markets otherwise underinvest in.  Copied to clipboard!

Segments Expand All Collapse All

Defining and Spotting Bubbles

Copied to clipboard!

(00:00:06)

• Key Takeaway: A bubble is defined as buying and selling an asset at prices irrationally above its actual worth, often involving new and exciting, yet uncertain, industries like AI.
• Summary: The episode opens by framing the current stock market surge, driven heavily by AI companies like NVIDIA, as potentially the biggest bubble in years. A bubble is characterized by prices detached from rational value, which keeps rising until it inevitably pops. Bubbles typically form around new, exciting technologies where future potential is highly uncertain.

Fama’s Skepticism and Bubble Clues

Copied to clipboard!

(00:05:34)

• Key Takeaway: Researchers developed four statistical clues—high valuations, volatility, issuance, and acceleration—to challenge Eugene Fama’s efficient market hypothesis by attempting to predict bubbles before they pop.
• Summary: Nobel laureate Eugene Fama is skeptical of bubbles, arguing that if markets are efficient, they cannot be reliably predicted before collapse. In response, researchers identified four indicators present in historical market spikes: high price-to-earnings ratios, stock price volatility, high issuance of new shares, and accelerating price increases. The current AI boom shows some, but not all, of these warning signs, leading to an ’early bubble’ assessment.

Policymaker Debate: Lean vs. Clean

Copied to clipboard!

(00:13:35)

• Key Takeaway: Policymakers debate whether to ’lean’ against suspected bubbles by intervening to shrink them or to ‘clean’ up the economic damage after they collapse, a debate intensified by the dot-com and housing crises.
• Summary: The ’lean versus clean’ debate centers on whether government intervention is warranted to deflate a suspected bubble or if inaction is better to avoid unintended harm. The dot-com bust and the housing crisis, which triggered the global financial crisis, made macroeconomists seriously consider the systemic risks posed by asset bubbles. Bubbles cause damage either when they pop, especially if fueled by debt, or subtly beforehand through wasted investment in the wrong areas.

AI Bubble Impact and Silver Linings

Copied to clipboard!

(00:20:36)

• Key Takeaway: While an AI crash could erase $35 trillion, the immediate threat to the financial system’s backbone may be lower due to less direct bank borrowing, and past bubbles show that overinvestment can lead to beneficial long-term infrastructure like broadband.
• Summary: An AI crash could wipe out $35 trillion globally, but the risk to core banking systems might be mitigated because AI companies rely more on private credit than traditional bank loans. Furthermore, the massive investment in AI data centers might create societal benefits, similar to how ‘dark fiber’ from the dot-com bubble eventually enabled modern broadband. This suggests that bubbles, by fixing market failures like underinvestment in R&D, can sometimes have positive externalities.