He Invested Through Five Bubbles | Andy Constan on What They Taught Him About AI

Key Concepts

• Bubble Regime: A market environment characterized by unsustainable price appreciation driven by human nature (FOMO), excessive leverage, and speculative behavior, often following a "something new" catalyst.
• Root Conditions: The foundational triggers for a bubble, such as technological breakthroughs, regulatory changes, or significant monetary easing.
• Escalation Events: Secondary developments (e.g., financial engineering, excessive deal-making, or central bank overreaction) that accelerate the bubble.
• Peaking Phase: The final stage of a bubble where expectations become parabolic and disconnected from long-term economic reality.
• Compute-Driven Growth: The current AI-centric market cycle where capital expenditure (CapEx) is heavily funneled into data centers and hyperscalers to build compute capacity.
• Circular Financing: The reliance on debt and corporate issuance to fund massive CapEx, which in turn drives earnings for semiconductor and AI-related firms.

---

1. The Framework of a Bubble

Andy Constan defines a bubble not by its inevitable pop, but by the regime in which it operates. He argues that predicting the exact top is the "holy grail" and is impossible in foresight. Instead, investors should focus on identifying the characteristics of a bubble regime:

• Root Conditions: Every bubble starts with a "something new" event. Examples include:
  • 1982–1987: Financial deregulation and the rise of the LBO (Leveraged Buyout).
  • 1995: The invention of Netscape Navigator (internet connectivity).
  • 2005–2008: Globalization ending inflation, leading to excessive bank leverage.
  • COVID-19: ZIRP (Zero Interest Rate Policy) and QE (Quantitative Easing) creating a bond bubble.
  • Today: The ChatGPT/AI moment, representing the culmination of a 40-year evolution in compute power and machine learning.
• Escalation: Once the root condition is established, escalation events—such as the 1998 Long-Term Capital Management (LTCM) crisis or the 2023 banking crisis—often lead to central bank interventions that "add rocket fuel" to the market by easing financial conditions unnecessarily.

2. The "Peaking" Phase and Market Dynamics

Constan identifies the current market as being in the peaking phase. Key indicators include:

• Step-Change in Expectations: A shift in earnings growth projections (e.g., semiconductor earnings moving from 60% to 100% growth expectations).
• FOMO (Fear Of Missing Out): The psychological impact of seeing neighbors and peers get "enormously rich suddenly," which is a hallmark of bubble behavior.
• The "Big Base" Problem: Unlike the 1990s, where tech started as a small portion of the S&P 500, tech is now a massive component of the economy. Constan argues that it is mathematically difficult for an industry that is already a significant share of GDP to continue growing at parabolic rates without cannibalizing the rest of the economy.

3. The Role of Policy and Financing

• Policy Errors: Constan notes that policymakers often stimulate bubbles through subsidies (e.g., housing) or by overreacting to financial instability (e.g., the Fed’s response to LTCM or the 2023 banking crisis). He warns that when the "hollowing out" of the middle class becomes too severe, governments eventually intervene to redistribute wealth from capital to labor.
• The Funding Chain: The current AI boom is funded by hyperscalers (e.g., Meta, Google, Microsoft) using cash hoards and corporate debt. Constan highlights that these companies are canceling stock buybacks to fund massive CapEx. The risk lies in whether the market can continue to absorb the massive supply of corporate debt and upcoming IPOs required to sustain this cycle.

4. Notable Quotes

• "The reason why a bubble regime is so difficult is because it plays precisely on human nature."
• "I’m pretty cynical about the ability for the government to do anything but steal from the future and to give to the current."
• "If you’re already at 50x [share of the economy], you can’t go to 60x. You can’t get a 4x [return]."

5. Synthesis and Conclusion

The main takeaway is that the current AI-driven market is a classic bubble regime characterized by a "compute-first" investment cycle. While the technology itself is transformative, the sustainability of the market depends on whether the massive CapEx being deployed can generate a positive Return on Investment (ROI) without relying on perpetual, massive leverage. Constan suggests that while bubbles can theoretically "grow into" their valuations over years, the more likely outcome is a period of volatility as the market grapples with the limits of financing and the social consequences of technological disruption. Investors are advised to prioritize discipline and recognize that in a bubble, the most costly mistakes are those driven by the desire to "keep up with the Joneses."