Episode Details
Back to EpisodesTRUSTED SOURCE — Steve Eisman's 'Big Short Investor Explains How the AI Bubble Could Burst': the technology doesn't need to fail for the stocks to disappoint
Published 1 week, 4 days ago
Description
A Trusted Source episode. Marco, the macro desk, walks through Steve Eisman's framework from episode 83 of The Real Eisman Playbook, published November 8, 2026 — his figures attributed to him throughout.
Eisman handicaps four stories: bull 30 percent, bear 30 percent, an intermediate overbuild case at 40 percent — citing PAA Research's view that US data-center capacity could be double demand by the end of 2027 — and doomsday at roughly zero. Underneath: by his numbers, over 65 percent of the S&P 500 is tied to AI and over half of this year's corporate bond issuance is AI-related. It is, in his phrase, all one trade.
Then the five risks: the shift from capital light to capital heavy (Google's equity raise, Oracle's debt and downgrade, Meta's off-balance-sheet vehicle, Nvidia's circular investment); power and permits; customer concentration (70 percent of Nvidia's receivables from five customers, roughly 70 percent of hyperscaler AI revenue from Anthropic and OpenAI); OpenAI as a single point of failure, with his safety rotation naming LVHD, SPLV and KBWP; and the absence of moats setting up a token price war. His preference: the picks and shovels — Nvidia, Micron, Arista, Cisco, GE Vernova and Eaton.
The desk note: today's AI Trade package agrees with Eisman on the overbuild tension and the concentration case, but diverges on names — Micron and Arista sit on the desk's avoid list and Eaton is its most contested name, and his 40 percent overbuild weight is more bearish than the desk base case. The full filing, with the cross-read, is in the desk records.