Remote AI Chip Access: A 90-Day License Checklist

Wide dusk view of an unmarked data-center aisle with a distant worker walking away.

Cloud and data-center operators need a remote AI chip access checklist now, because a new U.S. license rule can reach overseas compute that serves restricted countries. On November 16, 2026, the Bureau of Industry and Security issued an interim final rule on remote access to advanced super intelligence chips and servers. This guide maps total processing performance, customer end use, and the 90-day license window.

Updated November 16, 2026

Key Takeaways

  • BIS issued the remote-access interim final rule on November 16, 2026.
  • The rule covers networks that can supply over 12 million TPP.
  • Existing covered operators have 90 days after the rule comes into effect.
  • Five-country chip licenses do not cover chips for U.S.-based companies.

What does the remote-access rule cover?

The interim final rule was written by and at the direction of the Department of Treasury and the National Security Council. The Bureau of Industry and Security announcement said it is designed so remote computing arrangements in third countries cannot circumvent existing restrictions on sensitive American technology. TechCrunch’s account of the interim rule reports that companies must get licenses to export, reexport, or transfer AI chips and servers to data centers anywhere if those sites may provide compute to companies in China or other restricted countries.

The rules target companies based in the D:5 group, including China, Russia, Iran, North Korea, and 18 others. They apply to a data center or any network that can supply over 12 million total processing performance (TPP), roughly the maximum computing power of about 750 Nvidia H100 or H200 chips. Indirect access through go-betweens or brokers is covered. Commercial recommendation systems used in e-commerce or social media are not.

How do you check remote AI chip access before the 90-day clock?

Existing data center operators providing services that fall under the new rule must get licenses within 90 days of the rule coming into effect. Exporters and operators must watch customers’ ownership structures and activities to ensure they are not run by anyone from the D:5 group.

  1. Confirm whether the data center or network can supply more than 12 million TPP.
  2. Identify any path, including go-betweens or brokers, that may provide that compute to companies in China or other D:5 countries.
  3. Review customer ownership structures and activities for control by anyone from the D:5 group.
  4. If covered services are already being provided, obtain licenses within 90 days of the rule coming into effect.
  5. For advanced semiconductors headed to Malaysia, Singapore, Indonesia, Thailand, or Brazil, confirm a license unless the chips are destined for a U.S.-based company.

Which destinations require an advanced-chip license?

The rule requires licenses to export, reexport, or transfer advanced semiconductors to Malaysia, Singapore, Indonesia, Thailand, and Brazil. Commerce said those countries “present a higher-than-normal risk of diversion,” as TechCrunch reported. That requirement does not apply to chips destined for companies based in the United States.

CBS News reported that President Trump said in White House remarks on Monday, “Selling the other chips is a good thing, not a bad thing.” Existing regulations on chip exports apply only to the most advanced semiconductors.

Control What it reaches Stated limit or exception
License to export, reexport, or transfer AI chips and servers Data centers anywhere that may provide compute to D:5 companies, including through brokers Over 12 million TPP; existing operators have 90 days after the rule comes into effect
Advanced-semiconductor destination license Malaysia, Singapore, Indonesia, Thailand, and Brazil Does not apply to chips destined for U.S.-based companies
Commercial recommendation systems Systems used in e-commerce or social media The new rule does not apply

What happened to the AI Diffusion Rule?

This IFR rescinds the Biden administration’s AI Diffusion Rule from January 2025, which sought country-level limits on exports of advanced AI chips and restrictions on transfers of certain powerful AI models. That May, the Trump administration said it would throw out that framework, arguing that it added unnecessary restrictions on U.S. companies and the country’s allies.

This new control is a logical extension of existing controls.

TechCrunch quoted that line from the interim rule. The diffusion rescission does not remove the remote-access license test or the destination licenses. For related supply context, see Nvidia leadership and roadmap risk and GPU debt and neocloud credit.

Frequently asked questions

Who needs a license under the new remote-access rule?

Companies must get licenses to export, reexport, or transfer AI chips and servers to data centers anywhere if those sites may provide compute to companies in China or other restricted countries. Existing operators providing covered services must get licenses within 90 days of the rule coming into effect.

What does 12 million TPP mean here?

The rule applies to a data center or any network that can supply over 12 million total processing performance. TechCrunch reported that level is roughly the maximum computing power of about 750 Nvidia H100 or H200 chips.

Why were five countries added for diversion risk?

Commerce said Malaysia, Singapore, Indonesia, Thailand, and Brazil present a higher-than-normal risk of diversion, so advanced-semiconductor exports, reexports, or transfers there need a license. That requirement does not apply to chips destined for companies based in the United States.

Does the rule cover recommendation systems or the AI Diffusion Rule?

The rule does not apply to commercial recommendation systems used in e-commerce or social media. It also rescinds the Biden administration’s AI Diffusion Rule from January 2025.

Notice an error?

Help us improve our content by reporting any issues you find.