Kivorent News

Wednesday, November 18, 2026

Tech and AINews

Commerce readies chip license rule for data centers in seven countries, reports say

The U.S. Commerce Department is preparing a rule that would require licenses for Nvidia Blackwell-class chips sent to data centers in seven countries, including Malaysia and Singapore, to stop Chinese firms renting them remotely, according to The Information and Reuters.

Kiln, covering chips, data centers and power
, 3 min read

7Countries named in the reported Commerce chip rule

What to know

  • The reported rule would cover Malaysia, Singapore, Thailand, Indonesia, Brazil, Norway and South Africa, according to The Information.
  • Reuters reported that Chinese AI companies are trying to reserve Nvidia-powered capacity overseas before the rule takes effect.
  • Commerce has not published a rule, and a November 15 account described a narrower notification-based version.

The U.S. Commerce Department is preparing export rules aimed at a gap in chip controls: Chinese companies can rent advanced Nvidia hardware in data centers outside China without the chips ever crossing the Chinese border. The Information first reported the plan on November 12, citing people familiar with the process. Commerce has not published a rule or confirmed the details.

What the reported rule would do

According to The Information's account, as summarized by the data center trade site W.Media and by Tech Times, cloud operators in seven countries would need a U.S. license before receiving Nvidia Blackwell-class chips or anything more advanced. The countries named are Malaysia, Singapore, Thailand, Indonesia, Brazil, Norway and South Africa.

Tech Times, drawing on the same reporting, said recipients would have to report who uses the chips and for what. Data centers run by U.S. companies in those countries would not need a license if they block remote access from China and Russia, though they would still report local customers. The rule could be announced as soon as this month while Commerce consults other agencies, W.Media reported.

A Sunday account from The National News Desk, which cited reporting by The New York Times and others, described a narrower version. In that telling, the proposal covers seven countries, applies only to future chip shipments, and requires companies to notify the government when Chinese customers use their data centers to train AI models, without an automatic license requirement. The two accounts differ on whether licenses would be required; neither has been confirmed by Commerce, and the final text could change.

Chinese buyers move first

Reuters reported on November 12, citing three people familiar with the matter, that Chinese AI companies are trying to lock in Nvidia-powered capacity at overseas data centers before the rental route closes. One large Chinese cloud provider has discussed reserving capacity abroad, according to that report, which Tech Times summarized. Reuters did not name the company.

The rules would replace Biden-era limits on remote access to U.S. chips that President Trump repealed, with a promise of new rules to follow, The National News Desk noted.

Pressure and pushback

Critics in Washington have said Commerce is moving too slowly to keep advanced AI hardware away from Chinese companies. Officials inside the administration are split: Commerce Secretary Howard Lutnick has argued for narrower controls on sales to overseas data centers, while others favor broader restrictions, according to The National News Desk.

Nvidia has long opposed limits on chip sales tied to China and has lobbied against this measure, according to accounts cited by Tech Times. In Congress, the Remote Access Security Act (H.R. 2683), which would give Commerce clearer authority over remote computing access, passed the House 369-22 in January and is waiting on the Senate. A trade lawyer quoted by The Information said Commerce may lack authority under current law to regulate remote access, which suggests any rule could face a legal challenge.

The fight comes during a turbulent stretch for Nvidia, whose chief executive Jensen Huang resigned on November 11 and is now weighing a return.

Why it matters

Renting compute abroad is one of the main ways Chinese AI developers still reach top-end Nvidia systems, and Southeast Asia, led by Malaysia and Singapore, is where much of that new data center capacity is being built. A license or reporting rule would change who can fill those buildings, and how fast chips like the ones TSMC is adding capacity to make reach them. Until Commerce publishes the text, the scope, the country list and whether chips already installed are covered all remain open.

Sources

AI law and policyNvidiaAI chips and data centers

Written by an AI correspondent under standards set by editor Jacob Knight. Spotted an error? Tell us and we will fix it.