News

US moves to tighten AI chip export rules over remote compute access, TechCrunch reports

Published 3 min readAI-assisted · human approved
TL;DR

TechCrunch reports the US Commerce Department is preparing an interim rule that would require export licenses for AI chips going to large data centers that could serve China or other restricted countries. Existing operators get 90 days.

Why it matters

Teams that build or rent large GPU clusters outside the US may need to check who their customers are and how those customers are owned. The details come from one media report, so read the rule text before acting.

Image from TechCrunch
Image: TechCrunch. It belongs to its owner; see our image policy.

Original source

TechCrunch reported on 16 November 2026 that the US Commerce Department is preparing an interim rule requiring export licenses for AI chips and servers sent to data centers worldwide that might supply computing power to companies in China or other restricted countries. This summary relies on TechCrunch's report.

What happened

According to TechCrunch, the Trump administration wants to stop companies, especially Chinese ones, from getting around US export controls by renting AI infrastructure located in other countries. Existing rules already require a license to ship AI chips to buyers based in, or owned by a business in, the D:5 group of embargoed countries. TechCrunch says that group includes China, Russia, Iran, North Korea and 18 others.

The new rule extends that logic to remote access. Even when the hardware sits in a third country, a license is needed if the facility could serve customers from that group. TechCrunch quotes the rule as calling the measure "a logical extension of existing controls" and describing a goal of limiting how far firms headquartered in countries of concern can use US chips to train advanced models.

Key details

  • Threshold: The rule covers data centers or networks that can supply more than 12 million total processing performance (TPP), an export-control metric for computing power. TechCrunch's own conversion puts 12 million TPP at roughly the maximum capacity of about 750 Nvidia H100 or H200 chips.
  • Indirect access: The rule also reaches access through go-betweens or brokers. Exporters and data center operators must watch customers' ownership structures and activities to confirm none are run by anyone from the D:5 group.
  • Added countries: New licensing requirements apply to advanced chip exports, reexports or transfers to Malaysia, Singapore, Indonesia, Thailand and Brazil. Commerce said these "present a higher-than-normal risk of diversion." For them, the requirements do not apply to chips destined for US-based companies.
  • Deadline: Existing data center operators offering covered services must obtain licenses within 90 days of the rule coming into effect. TechCrunch's report does not state an effective date.
  • Exclusion: Commercial recommendation systems, such as those used in e-commerce or social media, are not covered.
  • Earlier policy: The rule rescinds the Biden administration's AI Diffusion Rule from January 2025, which set country-level limits on advanced chip exports and restricted transfers of certain powerful AI models. TechCrunch notes the Trump administration said that May it would discard that framework.

TechCrunch connects the timing to Nvidia co-founder Jensen Huang's resignation days earlier, amid a scandal involving an alleged Chinese spy linked to one of the chipmaker's customers. OpenBeat covered that separately in its report on Huang's resignation from Nvidia.

TechCrunch adds that the US government has reportedly wanted to put that customer, Megaspeed, on a special entity list, believing it bought thousands of Nvidia chips to supply compute to Chinese businesses and evade existing restrictions. That detail is framed as reported, not as a confirmed action. TechCrunch also cites a Reuters report from the prior week that Chinese AI companies are rushing to secure computing capacity overseas, fearing tighter US limits.

Why it matters

For infrastructure teams, the compliance question shifts from where a chip ships to who ultimately uses the compute. These are possible effects, not reported outcomes:

  • Cloud and colocation operators with large GPU clusters may need customer-ownership checks and license applications.
  • Buyers in Malaysia, Singapore, Indonesia, Thailand or Brazil may face new paperwork for advanced chip purchases.
  • Smaller deployments below the 12 million TPP level appear to fall outside the remote-access provision, based on TechCrunch's description.

What is still unconfirmed

TechCrunch describes the administration as preparing to roll out the rule. The article gives no effective date and no license criteria, and it includes no responses from affected companies. Treat the figures above as TechCrunch's summary of the interim rule and check the published text before making compliance decisions.

Checked claims

ClaimStatusEvidence
The interim rule applies to data centers or networks above a 12 million TPP computing-power threshold.confirmedThe rule applies to data centers or any network that can supply over 12 million total processing performance (TPP)
Existing data center operators providing covered services must obtain licenses within 90 days of the rule taking effect.confirmedExisting data center operators providing services that fall under this new rule are required to get licenses within 90 days of the rule coming into effect.
The rule describes itself as an extension of existing controls.confirmedThis new control is a logical extension of existing controls,
Commerce added Malaysia, Singapore, Indonesia, Thailand and Brazil because of diversion risk.confirmedpresent a higher-than-normal risk of diversion.
The rule adds licensing requirements for advanced chip exports to five additional countries.confirmedThe rule also adds new licensing requirements for chip exports, reexports, or transfer of advanced semiconductors to Malaysia, Singapore, Indonesia, Thailand, and Brazil.

Found a mistake? Use Report above; an editor reads every one. Editorial policy

Question for readersIf you run or rent GPU capacity outside the US, how would you verify the ownership and activities of your customers under a rule like this?

Discussion · 0

No comments yet. Start the discussion.

Sign in to join the discussion.