🐢 The AI slowdown

AI Leaders Want to Slow Down. These Stocks Could Benefit

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. NvidiaNVDA, Nebius GroupNBIS, and CoreWeaveCRWV 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 ETFSOXX fell 5% Monday morning, with MicronMU and MarvellMRVL among the chip names caught in the drop.
• The pressure quickly spread into the broader AI trade (opens in a new tab), pulling SoftBankSFTBY, Samsung ElectronicsSSNLF, AmazonAMZN, and HPEHPE lower.

AI’s Slow Lane Has New Winners

Slowing the AI race could buy AlphabetGOOGL and Meta PlatformsMETA 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 (opens in a new tab) the broader tech drop as investors warmed to companies that could benefit from a slower development cycle.
• Software also found buyers (opens in a new tab) during the chip rout, with CrowdStrikeCRWD and Palo Alto NetworksPANW 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.

Treasury Yield Hits 5% as Borrowing Costs Surge

The bond market is making its nerves everyone else’s problem. The benchmark 10-year Treasury yield briefly crossed 5% (opens in a new tab) on Monday, reaching its highest intraday level since 2007 as the bond selloff deepened. With oil near $109 and inflation still elevated, pressure on borrowing costs is only building.

• Higher 10-year rates can shrink mortgage borrowing power, lift auto-loan costs, and pressure companies refinancing debt or funding AI data centers (opens in a new tab).
• Traders see a 90% chance of a quarter-point Fed hike this week, while 30-year mortgage rates recently hit 6.76% (opens in a new tab).

Rate shock: The 10-year yield sits underneath a huge chunk of the economy, from mortgages and corporate debt to stock valuations and the government’s own borrowing costs. Wellington portfolio manager Brij Khurana called 5% “psychologically important” and backed a precautionary hike as the bond market grows more unsettled. That puts even more weight on this week’s Fed meeting, where a pause will test whether policymakers can calm the long end of the curve.