JPMorgan’s Playbook for a 10-15% Correction (or Worse) — ft. Michael Cembalest
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
This episode features Ed Elson and Scott Galloway interviewing Michael Cembalest, Chairman of Market and Investment Strategy for J.P. Morgan Asset & Wealth Management, about the current AI bubble and market conditions. Cembalest discusses how J.P. Morgan is positioning for what he sees as an inevitable market correction, sharing his analysis of the unprecedented capital spending in AI infrastructure and the risks it poses. (06:51)
- Main themes include AI bubble concerns, market valuations, and positioning for potential corrections ranging from 10-15% to potentially 40%
Speakers
Scott Galloway
Scott Galloway is a professor of marketing at NYU Stern School of Business and a bestselling author. He's known for his sharp analysis of business and technology trends, and hosts the popular Prof G podcast series.
Ed Elson
Ed Elson is co-host of Prof G Markets and works alongside Scott Galloway to deliver insights on market trends and economic analysis. He brings a measured perspective to financial discussions.
Michael Cembalest
Michael Cembalest is Chairman of Market and Investment Strategy for J.P. Morgan Asset & Wealth Management, overseeing strategy for $6 trillion in client assets. He joined JPMorgan in 1987 and previously served as chief investment officer for the firm's global private bank and head strategist for emerging markets fixed income.
Key Takeaways
AI Capital Spending Has Reached Historical Proportions
Cembalest reveals that tech capital spending in 2025 equals the combined investment of the moon landing, Manhattan Project, interstate highway system, electrification of farming, and major infrastructure projects like the Golden Gate Bridge and Hoover Dam. (12:54) This massive scale of investment, while impressive, raises serious questions about returns on investment. The context is particularly concerning when companies like Meta are spending 65-70% of their revenues on capital spending and R&D without clear destination paths - the same percentage they invested in the failed metaverse just two years ago. This historical comparison helps investors understand the unprecedented nature of current AI investments and why such massive capital allocation demands extraordinary returns to justify the spending.
Market Correction is Likely, But Scale Remains Uncertain
Cembalest's base case scenario anticipates a 10-15% market correction in 2026, which he describes as almost inevitable given current valuations. (23:26) However, he distinguishes this from more catastrophic 40% corrections seen in previous crashes. The key difference lies in how this bubble is being financed - primarily through internally generated cash flow rather than debt, which means it can continue longer but still requires substantial profit generation to justify the trillion-plus dollars spent since November 2022. This analysis suggests investors should prepare for volatility while recognizing that the financing structure may prevent the complete systemic collapse seen in previous bubbles.
Power Constraints Will Limit AI Expansion
One of the most practical limitations facing AI development is the power wall that will prevent companies from achieving their announced partnerships and expansions. (38:45) Cembalest points out that OpenAI's announced partnerships with Broadcom, Oracle, AMD, and NVIDIA would require 30 gigawatts of power - equivalent to 16 Hoover Dams - which is simply impossible to achieve. Microsoft's agreement to pay $130 per megawatt hour (double wholesale power prices) for Three Mile Island nuclear power demonstrates the desperation and unsustainable costs of powering AI infrastructure. This physical constraint provides a reality check on the ambitious AI expansion plans that are currently built into stock valuations.
Defensive Portfolio Positioning is Prudent
Rather than dramatically changing asset allocations, Cembalest recommends clients consider moving from growth portfolios to more balanced or conservative allocations. (31:54) A balanced defensive portfolio should contain 30-40% in cash equivalents, municipals, diversified hedge funds, and short-duration preferreds to reduce directional beta risk. This approach maintains the integrity of each portfolio type while allowing investors to match their risk tolerance to current market conditions. The strategy acknowledges that some institutional investors (like distressed pension plans) may still need growth exposure, but most investors should consider reducing risk given current valuations.
Individual Investors Have Advantages in Volatile Markets
Cembalest highlights that individual investors possess flexibility advantages over institutional investors during market corrections. (62:51) While large endowments and foundations typically meet quarterly and cannot respond quickly to market opportunities, individual investors can act when corrections create buying opportunities. Having dry powder (cash reserves) positioned to take advantage of sell-offs is crucial, as many quality assets often decline together during market stress, creating attractive entry points. The key is recognizing that most corrections tend to be V-shaped - rapid violent unwinding followed by equally rapid recovery - giving agile individual investors opportunities that institutional money cannot capture.
Statistics & Facts
- 75% of all revenues, profits, and capital spending since November 2022 have come from just 40 AI-related stocks, demonstrating the extreme concentration of market gains in AI companies. (11:51)
- Tech capital spending in 2025 is equivalent relative to GDP of the moon landing, Manhattan Project, interstate highway system, electrification of farming, and major infrastructure projects combined, showing unprecedented investment scale. (12:54)
- Last quarter, tech AI capital spending represented one-third of US GDP growth, and without it, US GDP growth would already be at 1%, revealing the economy's dangerous dependence on AI investment. (17:17)
