Brad Gerstner: No AI Bubble, Semis Eat the Nasdaq & AI's Take Off Problem
All-In Podcast
A Battle For The Soul Of America 1:02
Brad Gersner opens by thanking the crowd for their response to the Trump accounts, a plan to make 70 million children under 18 direct owners in America through dedicated accounts. He frames this as the antidote to socialism, arguing that making every child a capitalist is how the country wins what he calls a battle for its soul. He also thanks attendees who took a CAC scan, a heart scan costing about a hundred dollars and taking fifteen minutes, from the Center for Heart Attack Prevention he helped start, saying that if it becomes as common as a mammogram it could save fifty thousand lives a year in the country.
Markets Are Up On Earnings 2:00
Turning to his market check, Gersner notes the market is up 15 percent this year and 39 percent since January of last year, despite worries about tariffs, geopolitics, and AI regulation. Gold is flat and Bitcoin is down 10 percent, while Nvidia revenue, hyperscaler capital spending, OpenAI and Anthropic valuations, and SpaceX have all roughly doubled or more. He stresses this is an earnings driven expansion, not multiple expansion, since earnings are up 26 percent while the multiple on the Nasdaq and S&P has actually fallen. Nvidia now trades at 14 times next year's fully taxed earnings, which he says makes this nothing like the bubble of 2000, though he notes the gains are narrow, with consumer discretionary, software, and financials barely moving while semiconductors account for 70 percent of the Nasdaq's return.
Anthropic's Revenue Lit The Fuse 4:33
Gersner recalls asking Sam Altman on a podcast last October how he could commit to a trillion dollars in capex with only 13 billion dollars of revenue, and being told to sell his shares instead of getting an answer. He says the real answer came later: Anthropic's revenue went from 2 billion in January to 4 billion in February to 11 billion in March, and that jump lit the fuse for the market's April and May rally. Growth cooled afterward when Anthropic's annual run rate came in at 65 billion instead of the expected 75 billion, causing some sideways movement since. He estimates the combined run rate of Anthropic, OpenAI, and SpaceX is about 100 billion now, and says they need to reach roughly 180 billion by year end just to keep the AI trade intact.
Someone Has To Pay The Rent 8:02
He explains that Microsoft, Google, and Amazon are not paying for their own data center buildouts, they are building to rent them out, which means offtake revenue from customers has to cover the cost. Exiting this year around 200 billion in run rate revenue, he says that figure needs to climb toward 450, then 800, then a trillion dollars to keep pace with expected capital spending. On compute, the country added about 19 gigawatts this year, with semi analyst Dylan Patel forecasting 43 gigawatts next year, though Gersner thinks that number is too aggressive and expects closer to 25 gigawatts, with half going to Anthropic and OpenAI, which he still believes is enough to support their revenue targets. He also points to demand signals like Jensen Huang's prediction of inference growing a billionfold, 47 quadrillion tokens expected this year, Codex users up 40 times in eight months, and enterprise AI spending up 17 times in eighteen months, arguing this shows the demand side is not the problem.
Three Risks To Watch 12:31
Gersner names regulation, power, and interest rates as the main threats. On regulation he warns against repeating past overreactions, citing the shutdown of 67 fission reactors in the country that left the United States disarmed against China on clean energy. On power, he doubts the industry can build 43 gigawatts in a single year given permitting delays, grid interconnection problems, labor shortages, and sold out equipment. On rates, he says there is now over a 90 percent chance of a rate hike, which raises the cost of borrowed money funding data centers and, quoting Warren Buffett, acts like gravity on stocks. He closes by describing his own positioning as medium sized and mentally flexible, watching whether monthly AI lab revenues reach 8 billion dollars as a signal of takeoff, while warning that unlike 2023 to 2025, simply betting on the AI trade is no longer enough since everyone already knows about it and it is priced in.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.
