The Commerce Department extended export controls to cover advanced AI models while authorizing release to specific trusted partners, according to Mayer Brown's analysis of the rule. This follows the existing chip export control framework and effectively creates a tiered system where the most capable AI — and by extension the hardware needed to run it — faces restrictions on where it can go.
Who's exposed: Nvidia (NVDA) is the most directly affected public company. Its H100 and successor chips are the primary hardware used to train and run the advanced AI models now subject to controls. Every new export restriction layer adds compliance cost and, more importantly, shrinks the addressable market for its highest-margin products. Advanced Micro Devices (AMD) faces the same dynamic with its MI-series AI accelerators — it's the number-two player in AI compute and subject to the same export licensing requirements. Both companies have already taken revenue hits from China chip restrictions; AI model controls extend the perimeter of what's restricted.