AivexaNewsSearch
AI news for builders and product teamsChecked every hour

US tightens AI chip export controls to deter Chinese companies from accessing compute remotely

Collected Nov 16, 2026

The Trump administration is preparing to introduce an interim final rule that expands U.S. export controls to cover remote access to AI computing infrastructure, not just the physical shipment of chips. Announced as taking effect Monday, the rule is aimed at stopping companies — particularly those in China — from sidestepping existing controls by renting or otherwise reaching AI data centers located in third countries. Under the measure, the Commerce Department will require licenses to export, reexport or transfer AI chips and servers to data centers anywhere in the world when those facilities may provide compute to companies in China or other restricted nations.

The scope is defined by the U.S. D:5 group of embargoed countries, which includes China, Russia, Iran, North Korea and 18 others. Washington already requires licenses to ship AI chips to firms based in, or owned by businesses in, that group. The new rule extends that logic to compute delivered remotely: the interim rule describes the control as a logical extension of existing restrictions and says it is intended to undercut the ability of companies headquartered in countries of concern to use advanced U.S. semiconductors to train advanced models.

A concrete threshold sits at the center of the rule. It applies to data centers or any network capable of supplying more than 12 million total processing performance (TPP), an export-control metric the U.S. uses to gauge computing power. Twelve million TPP roughly equals the maximum computing power of about 750 Nvidia H100 or H200 chips. That figure gives operators and exporters a quantifiable line for deciding whether a facility falls in scope.

The rule also reaches indirect arrangements. It covers access obtained through go-betweens or brokers, and it obliges exporters and data center operators to scrutinize their customers' ownership structures and activities to confirm they are not run by anyone from the D:5 group. This shifts part of the compliance burden onto the operators of overseas facilities, who must now assess who ultimately controls their tenants rather than only what hardware is being shipped.

Separately, the rule adds new licensing requirements for exports, reexports or transfers of advanced semiconductors to Malaysia, Singapore, Indonesia, Thailand and Brazil. The Commerce Department said those countries were included because they present a higher-than-normal risk of diversion. For those five, the license requirement does not apply to chips destined for U.S.-based companies.

On timing, existing data center operators already providing services covered by the rule must obtain licenses within 90 days of it taking effect. The rule does not apply to commercial recommendation systems, such as those used in e-commerce or social media, which suggests the intent is to target model training and advanced inference rather than everyday consumer personalization workloads.

The measure also rescinds the Biden administration's AI Diffusion Rule from January 2025, which had sought country-level caps on advanced AI chip exports and limits on transfers of certain powerful AI models. The Trump administration said in May that it would discard that framework, arguing it imposed unnecessary restrictions on U.S. companies and allies. In the interim rule's framing, the changes are meant to prevent certain advanced semiconductors from being used in data centers at a computing power level sufficient to train or run frontier models, which the government says could support military- and intelligence-related activities that threaten U.S. national security or foreign policy interests.

The timing is tied to a specific enforcement concern. The move comes days after Nvidia co-founder Jensen Huang resigned 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 has been buying thousands of Nvidia chips to supply compute to Chinese businesses and get around existing export controls. Reuters also reported last week that Chinese AI companies are rushing to secure computing capacity overseas in anticipation of tighter U.S. restrictions.

The practical effect for developers and infrastructure operators is a compliance regime built around who can reach the compute, not only where the hardware lands. In practice, a data center outside the U.S. that serves Chinese customers above the 12 million TPP line would need a license, and operators' customer due diligence becomes a licensing question. The 90-day window gives existing facilities a limited runway to review tenant ownership and apply for authorization.

A likely trade-off is added friction for global cloud and colocation providers, who may need to segment customers or relocate workloads by jurisdiction. The carve-outs for U.S.-based companies in the five named countries and for recommendation systems suggest the administration is trying to limit collateral damage to American firms and consumer services, but the ownership-look-through requirement could still complicate sales to customers with layered corporate structures. The rescission of the AI Diffusion Rule also removes country-level caps that some allies had criticized, replacing them with a targeted anti-diversion approach.

Open questions remain about how TPP is calculated across multi-tenant facilities and networked clusters, and how operators are expected to verify beneficial ownership in practice. Those details will determine how broad the rule proves in application.

Why it matters: Any developer or product team relying on AI capacity hosted outside the U.S. — or selling compute to customers with Chinese or other D:5 ties — now faces a licensing and due-diligence layer on top of hardware controls. Cloud and data center operators should expect to audit tenant ownership and map workloads against the 12 million TPP threshold, and companies in Malaysia, Singapore, Indonesia, Thailand and Brazil face new licensing for advanced chip transfers unless the buyer is U.S.-based. The 90-day transition is the immediate deadline to watch, and in my reading the biggest near-term risk is compliance uncertainty for multi-tenant providers rather than an outright ban.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

Under the interim rule, the U.S. will require firms to get licenses to export AI chips if they may provide compute to China or other restricted countries.