US Requires Licenses for AI Chips in Overseas Data Centers
The Commerce Department's new interim rule requires licenses for AI chips and servers at any data center worldwide that may serve Chinese firms, closing the remote-compute loophole.
By Grace Kim
4 min read

What's News
- The interim rule requires licenses for AI chip and server exports to any data center worldwide that may provide compute to companies in China or other D:5 countries.
- The threshold is 12 million TPP, roughly the max computing power of about 750 Nvidia H100 or H200 chips.
- New licensing requirements apply to advanced semiconductor exports to Malaysia, Singapore, Indonesia, Thailand and Brazil.
- Existing data center operators must obtain licenses within 90 days of the rule taking effect.
- The rule rescinds the Biden administration's AI Diffusion Rule from January 2025.
By Grace Kim
The Commerce Department will now require licenses to export, reexport or transfer AI chips and servers to data centers anywhere in the world if those facilities may provide compute to companies in China or other restricted countries. The Trump administration rolled out the interim rule on Monday, aiming to stop Chinese firms from accessing U.S. AI infrastructure remotely and sidestepping existing export controls.
The stakes are straightforward: Washington wants to cut off every path to advanced American compute, not just the direct shipping routes.
Who does the rule target?
The remote-access provisions cover companies based in the U.S.'s D:5 group of embargoed countries — China, Russia, Iran, North Korea and 18 others. The U.S. already requires chip companies to obtain licenses to ship AI chips to companies based in, or owned by businesses in, those countries.
The new rule extends that logic beyond ownership to access. "This new control is a logical extension of existing controls," the interim rule reads. "These restrictions will undermine the ability of companies headquartered in countries of concern from using U.S. advanced Super Intelligence (SI) semiconductors to train advanced SI models."
In plain words: if a data center anywhere serves restricted customers, its chip supply now needs a U.S. license.
What's the compute threshold?
The rule applies to data centers or any network that can supply over 12 million total processing performance (TPP) — an export-control metric the U.S. uses to measure computing power. Twelve million TPP roughly matches the maximum computing power of about 750 Nvidia H100 or H200 chips.
The rule also reaches indirect access through go-betweens or brokers. Exporters and data center operators must monitor their customers' ownership structures and activities to ensure they aren't run by anyone from the D:5 group.
One carve-out: the rule does not apply to commercial recommendation systems, like those used in e-commerce or social media.
Why now?
The timing follows a scandal at Nvidia. Co-founder Jensen Huang resigned days earlier amid a scandal involving an alleged Chinese spy linked to one of the chipmaker's customers. The U.S. government has reportedly wanted to place that customer, Megaspeed, on a special entity list, believing the company bought thousands of Nvidia chips to provide compute to Chinese businesses and get around existing export-control restrictions.
The pressure had been building. Reuters reported last week that Chinese AI companies are rushing to secure computing capacity overseas amid fears of tighter restrictions from the U.S.
Which countries face new licensing?
The rule adds new licensing requirements for chip exports, reexports or transfers of advanced semiconductors to five countries: Malaysia, Singapore, Indonesia, Thailand and Brazil. The Commerce Department said it included them because they "present a higher-than-normal risk of diversion."
There's an exception for these markets: the license requirements won't apply to chips "destined" for companies based in the U.S.
Existing data center operators providing services covered by the new rule must obtain licenses within 90 days of the rule taking effect. That's a hard compliance clock for operators already running restricted workloads.
What happens to the Biden-era framework?
The rule also rescinds the Biden administration's AI Diffusion Rule from January 2025, which sought to impose country-level limits on exports of advanced AI chips and restrict transfers of certain powerful AI models. In May, the Trump administration said it would discard that framework, arguing it added unnecessary restrictions on U.S. companies and the country's allies.
The new interim rule replaces country caps with a compute-and-customer approach. "By making these changes, this IFR prevents the use of certain SI items in data centers and at a computing power level that can be used for training or running of advanced SI models," the rule reads. "Access to such computing power can be used to train and run advanced SI models capable of supporting military- and intelligence-related activities that could threaten U.S. national security or foreign policy interests."
For chipmakers and data center operators, the message is blunt: watch who your customers serve, not just who they are — because Washington now is.
Original: bis.gov
More from Grace Kim
Show full bio
Market editor covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.
2,178 articles