AI just tapped the brakes, and Wall Street spilled its coffee. Leading AI executives made a rare call to slow frontier development this week, rattling the hardware trade and challenging the push to build bigger models faster. The question now is who benefits if speed stops being the industry’s only scorecard.
Pressure point: The pressure hit companies tied to the AI buildout hardest, where demand for chips, memory, cloud capacity, and infrastructure has ridden on expectations for ever-larger models. Nvidia, Nebius Group, and CoreWeave all fell as Wall Street questioned how much spending relies on constant frontier-model upgrades. JPMorgan’s trading desk pushed back, arguing that “this is not a halt” with adoption still strong and slowing too much risking US leadership in AI.
- The iShares Semiconductor ETF fell 5% Monday morning, with Micron and Marvell among the chip names caught in the drop.
- The pressure quickly spread into the broader AI trade, pulling SoftBank, Samsung Electronics, Amazon, and HPE lower.
AI’s Slow Lane Has New Winners
Slowing the AI race could buy Alphabet and Meta Platforms some valuable time. Google recently released Gemini 3.8 Flash, while Meta’s Muse cracked the App Store’s top three last week. Both still trail Claude Fable 5.1, GPT-6 Astra, and Grok-4.6 on OpenLM’s leaderboard. New Street Research analyst Dan Salmon sees another opening, with Meta focusing more on data-center rentals and Google leaning into distribution and chips.
- Alphabet and Meta bucked the broader tech drop as investors warmed to companies that could benefit from a slower development cycle.
- Software also found buyers during the chip rout, with CrowdStrike and Palo Alto Networks leading the move.
AI changes gears: The slowdown trade is drawing a new line through AI. Hardware names built around nonstop training and infrastructure spending carry more risk, while software, cybersecurity, and inference could benefit if adoption keeps growing. Dario Amodei called for slower model improvements, while Sam Altman stressed that pacing does “not mean ‘stopping.’” AI spending isn’t disappearing. It may just start flowing to a different set of winners.