Brad Gerstner delivers a market check, arguing there's no AI bubble despite high valuations, driven by real revenue growth and the largest CapEx super cycle in tech history. He highlights the critical need for AI lab revenues to hit specific targets to sustain the trade, while flagging risks like 2026 interest rate hikes, overhyped compute buildout forecasts, and regulatory pressures. Gerstner advocates for a moderate, flexible investment stance rather than aggressive leverage.
Summarized by Podsumo
Markets are up 15% this year and 39% since Jan 2023, driven by AI infrastructure earnings, not multiple expansion—NASD and S&P multiples are actually down.
Semiconductors contribute 70% of the NASDAQ's return, with public infrastructure companies like Dell (up 5x) and Hynix (up 9x) delivering venture-capital-like returns.
The top three AI labs (Anthropic, OpenAI, SpaceX) are on track for ~$100B annualized revenue, but need to reach $180B by year-end to justify current CapEx levels.
Gerstner predicts only ~25 gigawatts of new compute will be stood up next year, below Dylan Patel's 43 gigawatt forecast, due to permitting and local opposition.
There's a 90%+ chance of an interest rate hike in 2026, which could boost bond yields to 5.5-6%, pressuring equity valuations and making leverage dangerous.
"This is not a program. This is a platform. It is the largest unlock of direct philanthropy in the history of the country."
"It's not a TAM issue. You only have to get to about 4% of that TAM or $1.2 trillion in order to pay for the CapEx."
"We make every child a capitalist. And also, thank you to all the people yesterday who came up, who took the CAC scan, the heart scan outside from the Center for Heart Attack Prevention."