AiGpu

Policy·

U.S. Tightens AI Chip Export Controls to Block Remote Compute Access

New interim rule requires licenses for any data center worldwide that could supply compute to firms in embargoed nations, targeting remote access loopholes used by Chinese companies.

Illustration of U.S. export control rule affecting AI chip shipments to global data centers

The U.S. Commerce Department has issued an interim final rule that extends existing export controls to any data center or network capable of delivering more than 12 million total processing performance (TPP) — roughly the equivalent of 750 Nvidia H100/H200 GPUs. Operators must now obtain a license before exporting, re‑exporting, or transferring advanced AI chips and servers to facilities that may serve companies headquartered in the D:5 group of embargoed countries, which includes China, Russia, Iran, North Korea and 18 others.

The rule also adds licensing requirements for shipments to Malaysia, Singapore, Indonesia, Thailand and Brazil, citing a higher‑than‑normal risk of diversion. Existing data‑center providers that fall under the new threshold have 90 days to secure the necessary approvals. Notably, commercial recommendation systems such as those used in e‑commerce or social media are exempt.

For cloud and GPU‑as‑a‑service providers, the regulation means tighter due‑diligence on customer ownership structures and usage patterns. Operators must monitor indirect access through brokers and ensure that compute capacity is not funneled to restricted entities, adding compliance overhead and potential delays in capacity provisioning.

Why it matters for GPU and AI infrastructure

GPU cloud platforms like AiGpu must now embed export‑control checks into their provisioning pipelines, verify end‑user eligibility, and maintain audit trails for every high‑performance allocation. This raises operational costs but also creates a competitive advantage for providers that can demonstrate robust compliance, especially as enterprises seek guaranteed, regulation‑ready compute for training and inference workloads.

  • aigpu
  • ai gpu
  • ai gpu cloud
  • aigpu dubai
  • export controls
  • ai chips
  • data centers
  • gpu cloud

By AiGpu Editorial · Editorial rewrite based on public reporting (TechCrunch AI)

← All articles