The Trump administration is rolling out new export controls that require licences to export, re-export or transfer AI chips and servers to data centres anywhere in the world if they may provide compute to companies in China or other restricted (D:5 group) countries.
The rule targets the remote-access loophole: instead of shipping chips to China, Chinese firms have been renting compute in data centres abroad. It applies to data centres and networks supplying over 12 million total processing performance (TPP), and it rescinds the Biden-era AI Diffusion Rule.
The move follows reporting that Chinese AI firms are rushing to secure overseas compute, and comes amid the reported resignation of Nvidia co-founder Jensen Huang in connection with an alleged Chinese spy scandal linked to a chip customer — reported, not settled, context for the timing.
If enforced as described, the rule would push AI-chip controls from where hardware ships to where compute is actually used — a much wider net, and one likely to be tested quickly by cloud providers and chipmakers alike.