The US Remote-Access AI Chip Rule Explained: What Changes for Cloud Computing Outside America

Awais Khalid

November 18, 2026

The US Remote-Access AI Chip Rule Explained: What Changes for Cloud Computing Outside America

On 16 November 2026 the US Commerce Department’s Bureau of Industry and Security (BIS) issued an interim final rule on remote access to advanced AI chips. Previous export controls stopped companies in China from buying Nvidia’s top processors directly, but they could still rent the same computing power from data centres in other countries. The new rule is designed to close that route.

What the rule does

According to the BIS announcement, the rule aligns controls on remote access to advanced computing with the existing controls on physically shipping chips. BIS says it was written by and at the direction of the Treasury Department and the National Security Council. Reporting from TechCrunch, SCMP and Silicon Report describes the main provisions:

  • Data centre licences: a licence is needed to export, re-export or transfer advanced AI chips and servers to data centres anywhere in the world that may supply large volumes of computing power to companies headquartered in China or other restricted destinations, including Russia and Iran.
  • Customer due diligence: operators must check who their customers are and who ultimately uses the compute.
  • Five more countries: exports of advanced AI chips and servers to companies headquartered in Indonesia, Malaysia, Singapore, Thailand and Brazil now need a licence, because BIS considers the risk of diversion higher there.

The rule runs to 132 pages, according to CyberScoop. Silicon Report says its provisions take effect 90 days after issuance. Some analysts also report that it removes the earlier controls on AI model weights from the Commerce Control List. Companies should read the Federal Register text, titled “Addressing Remote Access to Advanced Semiconductors”, rather than rely on summaries.

Why now

The timing reflects a fast-moving week. Scrutiny of chip diversion through Southeast Asian data centres intensified after reports about a Singapore-based operator, and Nvidia’s chief executive resigned on 12 November. Commerce Secretary Howard Lutnick has faced criticism over enforcement. The rule also landed as President Trump began a trip to China to meet President Xi Jinping. The “interim final” form lets Commerce skip parts of the normal rulemaking process, a route usually used when an agency argues that speed is essential.

Who is affected

Cloud and colocation providers outside the US carry most of the new compliance work: licences for advanced hardware and know-your-customer checks on who is renting compute.

Chipmakers and server vendors face slower sales into data centre hubs such as Singapore and Malaysia. Markets reacted quickly: Silicon Report notes the Philadelphia Semiconductor Index fell more than 5% on 16 November, with Nvidia closing over 2% lower and Micron down more than 7%.

AI companies and developers in affected countries may see higher prices or longer waits for top-tier GPU capacity as providers reassess customers. Nvidia’s latest outlook already assumes no Data Center compute revenue from China, as covered in our report on Nvidia’s Q3 results.

Analysis: what this means for the AI industry

Export controls have been chasing a moving target. Every time Washington restricted physical shipments, demand shifted to cloud access, which is harder to police because compute crosses borders as a service rather than as hardware. This rule moves enforcement to the data centre itself. Its effectiveness will depend on two things that are hard to verify: whether operators in third countries can reliably identify ultimate users, and whether BIS has the staff to process licences and audit compliance. The broader trend is clear: access to frontier compute is now treated as a national security matter, and companies building AI products on rented GPUs abroad should plan for more paperwork and less certainty about supply.

FAQ

Does the rule ban AI chip sales to Singapore or Malaysia?

No. It requires a licence for exports of advanced AI chips and servers to companies headquartered in Indonesia, Malaysia, Singapore, Thailand and Brazil. Licences may still be granted.

Does it affect ordinary cloud users?

Most users of consumer AI apps will not notice a direct change. The rule targets operators supplying large volumes of advanced compute.

Sources

Stay Ahead of AI

Get the latest AI news delivered to your inbox.

We don’t spam! Read our privacy policy for more info.

Published

Spotted an error? Read our corrections policy or email editorial@perplexityaimagazine.com.