US moves to close China loophole on remote AI servers with new export control plan

The Trump administration is reportedly preparing a new round of export controls aimed at cutting off China’s access not just to physical AI chips, but also to powerful compute delivered via remote servers in third countries. Until now, US rules have mainly focused on blocking the shipment of high‑end GPUs and accelerators to Chinese entities. The proposed measure would go a step further by targeting cloud‑based AI services hosted in nearby hubs such as Singapore or the Middle East that Chinese firms can tap over the internet.

According to reports, the draft rule would apply to remote AI servers above a certain performance threshold and could require cloud and data‑center providers to vet the location and identity of their customers, and in some cases restrict or deny service. The goal is to close a perceived loophole that allows Chinese companies to train and run cutting‑edge AI models without importing the underlying hardware.

The “remote access” restriction could be circulated to industry as soon as September, giving chipmakers, hyperscale cloud providers, and infrastructure operators little time to adapt. Supporters frame the plan as necessary for national security, while critics warn that enforcement will be technically complex and could further fragment the global AI ecosystem and cloud market.