Technology 32 sources · over 4 days Latest coverage 19 Nov 2026, 6:33 am UTC

US Extends AI Chip Export Curbs to Remote Access and Southeast Asia Ahead of Trump-Xi Talks

Washington's new interim rule targets cloud rentals and Asian data centres as routes to China, landing days before Trump meets Xi and prompting a Beijing threat of countermeasures.

By Hushread Stories, from 32 outlets · First published 18 Nov 2026 · Chapter 4 of 5

Earlier: US Closes Remote-Access Loophole on AI Chips as Trump Heads to China

In brief

  1. The Commerce Department's interim rule blocks Chinese and other restricted entities from remotely accessing advanced American AI chips.
  2. Data centre operators in Singapore, Malaysia, Thailand, Indonesia and Brazil now need US licences to buy advanced chips.
  3. China's commerce ministry strongly opposes the rules and says it will take countermeasures, while urging dialogue before the Xi meeting.
  4. Nvidia has stopped shipping licensed Hopper-series chips to China and reportedly warned customers of possible lawsuits over the rule.
  5. The administration is reportedly weighing a requirement that US firms disclose any Chinese AI use, adding a further layer to the curbs.
US Extends AI Chip Export Curbs to Remote Access and Southeast Asia Ahead of Trump-Xi Talks
Source: Technology

Timeline · 12 moments

12 moments Open the full timeline →

Tencent says remote GPU ban will push China toward domestic chips

The Register ↗

Alibaba faces pressure to deliver AI plans under chip limits

Bloomberg Technology ↗

Export rule closes offshore data centre loophole; divisions persist

Christian Science Monitor ↗

Nvidia co-founder Huang expected in Shenzhen with Trump

France24 ↗

US commission proposes tougher export certification rules

South China Morning Post ↗

Nvidia stops shipping licensed Hopper chips to China

Technology ↗

China warns of retaliation, urges dialogue

Bloomberg Technology ↗

Amazon, Google and Meta press Trump; he proceeds

India Today ↗

Singapore defends status as trusted business hub

Fortune ↗

Rule extends bans to Indonesia, Malaysia, Singapore, Thailand, Brazil

The Register ↗
2 more See the rest →

Update 19 Nov 2026, 6:33 am UTC

Since the original piece, the story has widened beyond the export rule itself. Reuters reported the administration is weighing a requirement that US firms disclose their use of Chinese AI, and Bloomberg Technology reported Alibaba faces mounting pressure to show its AI plans can work under the new chip limits. The Register quoted Tencent's Martin Lau saying the US ban on remote GPU access will hurt in the short term but push China to build domestic alternatives faster. The headline and core facts still hold, so the existing framing stands with these developments added to the timeline.

Update 19 Nov 2026, 0:27 am UTC

Coverage since the rule's release has moved from the export rule itself toward the politics around it. A Christian Science Monitor report says divisions over AI and China persist inside Washington even as the Commerce Department closes the offshore data centre loophole. France24 reports that Nvidia co-founder Jensen Huang was expected to accompany Trump to Shenzhen, raising questions about whether the chip dispute will shape the Xi summit. A Reddit post linking to NBC News adds that Chinese companies could rent top chips online, a point the original piece already touched on.

How it started

The US has spent years limiting what advanced AI hardware reaches China. The gap that kept coming up was indirect access: a Chinese firm that cannot buy a top Nvidia chip can still rent time on one sitting in a data centre somewhere else.

That is the loophole the Trump administration moved to close on Monday, 16 November, just as President Donald Trump began a trip to Asia to meet President Xi Jinping. ThePrint reported that Washington said it would further restrict China's access to the best chips from companies like Nvidia.

The backdrop was an accelerating AI race. Alibaba had just unveiled a new flagship model, and a separate report said China's top AI firms were limiting offensive cyber capabilities in their latest releases.

How it unfolded

The Commerce Department's Bureau of Industry and Security issued an interim final rule on 16 November. Business Line described it as preventing remote access from circumventing existing limits, so third-country data centres cannot be used to bypass restrictions. Global News reported that overseas data centres serving customers in countries including China, Russia, Iran and Cuba must identify their end users.

The geographic reach was wider than many expected. The Register said the rule bans remote access to certain GPUs and servers in data centres in Indonesia, Malaysia, Singapore, Thailand and Brazil. Fortune and the Japan Times reported that operators in Southeast Asia now need a US licence to buy advanced chips from Nvidia and other American makers.

Southeast Asia reacted quickly. The New York Times said the rules could put billions in investment at risk and force an even harder balancing act between Washington and Beijing. Singapore defended its status as a trusted business hub, according to Fortune. MLQ.ai reported that Malaysia's trade minister told operators to comply, said the country would seek US clarification, and would notify all 117 approved data centre projects.

Companies pushed back too. India Today reported that Amazon, Google and Meta pressed Trump over the curbs and that he went ahead anyway. Tom's Hardware cited CNBC as saying Nvidia warned customers the rule could expose them to legal trouble over serving Chinese clients. ZeroHedge claimed the crackdown froze a $480 million loan linked to ByteDance, a report that rests on that outlet alone.

On 17 November, China's commerce ministry said it strongly opposed the rules and would hit back if Washington proceeds, while urging dialogue, according to Bloomberg and The Next Web. The same day, a Reuters investigation, relayed by bdnews24, described Chinese entities using output from leading American AI models to train domestic rivals as chip curbs bite.

On 18 November, CNBC reported that Nvidia's interim CEO Colette Kress said the company has stopped shipping Hopper-series AI chips to China. The Financial Express noted this came despite $2.4 billion in quarterly sales. A bipartisan US commission also urged Congress to require exporters of advanced chips to China to certify the shipments would not divert US manufacturing capacity, according to the South China Morning Post.

Where it stands

The rule is in place as an interim measure, and the diplomatic temperature is high. Tom's Hardware described Beijing's threat of countermeasures as coming ahead of Thursday's Xi meeting, with the trade war potentially renewed.

The administration is not speaking with one voice. Reason described a divide inside the government on AI chip policy toward China, and the South China Morning Post reported fractures across Washington and Silicon Valley, tied to a leadership crisis at Nvidia. Fox Business referred to the ouster of Jensen Huang, though the coverage here does not detail it.

Some coverage is conflicting on emphasis. Reports differ on whether the rule is final or still proposed, with Fortune calling it a proposal while Business Line describes an interim final rule.

What to watch

The first test is the Trump-Xi meeting on Thursday, and whether chips are bargaining material or a sideline. Watch for the specific countermeasures Beijing has threatened but not yet named.

Also open are how Southeast Asian governments and data centre operators handle licensing, whether Congress acts on the commission's proposal, and whether Nvidia's Hopper halt is followed by further changes to its China business.

Spotted an error? Email [email protected] and we'll fix it.

More in Technology

All →