AI Infrastructure Investment Debate Intensifies as Tepper Rotates, Eisman Warns and CFTC Eyes Compute Derivatives

A wave of developments on August 17, 2026 sharpened the debate over where to place bets in the AI infrastructure cycle. Billionaire David Tepper's Appaloosa Management, whose portfolio returned 32% in the first half of 2026 largely on memory-chip holdings, disclosed in its Q2 2026 13F filed August 14 that it trimmed Micron Technology by 690,000 shares to 975,000, a stake still valued at roughly $1.1 billion, and fully exited its 281,250-share SanDisk position, rotating proceeds deeper into the AI stack. Separately, Steve Eisman, known for the Big Short trade, flagged what he calls the Achilles heel of the AI boom: the entire trade's success depends heavily on just two companies, OpenAI and Anthropic, and a price war with cheaper Chinese models could destabilize both.

Eisman said he is not yet shorting the sector, noting hyperscalers continue to spend heavily and Nvidia is working with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to mobilize more than $500 billion for AI infrastructure. On the regulatory front, the Commodity Futures Trading Commission said it wants to take public comment on derivatives tied to computing capacity, a move prompted by plans already in motion at CME Group and Silicon Data to launch two compute futures contracts on October 5, 2026. OpenAI President Greg Brockman, appearing on CNBC's Squawk Box, framed compute as the new oil of the AI age and addressed capex spending and the state of the AI tech race.

On the infrastructure side, Vertiv and Super Micro Computer both reported recent quarterly results that analysts described as telling two very different stories, with Vertiv posting second-quarter net sales of $3.27 billion, up 24% year over year with 18% organic growth, while SMCI reported its full fiscal year 2026 figures separately.