US AI Chip Export Rule Targets Remote Access Through Overseas Data Centers

The US Commerce Department released an interim final rule on 16 November 2026 that requires licenses for advanced chips headed to data centers serving companies headquartered in China and other countries of concern, and adds license requirements for Indonesia, Malaysia, Singapore, Thailand and Brazil.

Christopher Ort
By Christopher Ort · 4 min read
Share
Rows of server racks in a large data center aisle seen from floor level.
Image: AI-generated, i10X
On this page
  1. Quick Take
  2. In plain English
  3. What happened
  4. How the rule works
  5. Why it matters
  6. What's still open
  7. Sources
  8. i10x Perspective

The US Commerce Department has moved to close the loophole that let Chinese companies rent American AI chips they are not allowed to buy. On 16 November 2026, the Bureau of Industry and Security (BIS) released an interim final rule that requires a license whenever advanced chips are headed to a data center, anywhere in the world, that serves large amounts of compute to companies headquartered in China or other countries of concern. The rule also adds license requirements for chip shipments to Indonesia, Malaysia, Singapore, Thailand and Brazil.

i10X Quick Take

Quick Take

  1. What changed: a new end-use control (§ 744.24) requires a license to ship advanced chips and servers to data centers that provide more than 12,000,000 “aggregate TPP” of compute to companies headquartered in Country Group D:5 (which includes China) or Macau.
  2. Five new license countries: Indonesia, Malaysia, Singapore, Thailand and Brazil, which BIS says present a higher-than-normal risk of diversion. Chips for companies whose ultimate parent is headquartered in the US are exempt if they file monthly reports.
  3. Existing deals: data centers already serving such customers must obtain a license within 90 days of the rule’s effective date or face future restrictions.
  4. Carve-out: commercial recommender systems, such as those used by online retailers and social networks, are excluded.
  5. Also in the rule: tighter presumptions for foundries and chip packagers, and the formal rescission of the Biden-era AI Diffusion Rule.

In plain English

The US already bans selling its best AI chips to Chinese companies. But a Chinese company could still rent those chips from a data center in, say, Malaysia and train its models remotely. This new rule makes that much harder. Anyone shipping top AI chips now has to check who the data center’s customers really are, and five Southeast Asian and Latin American countries need a license for these chips at all.

What happened

BIS said the rule was written “by and at the direction of the Department of Treasury and the National Security Council” and is designed “to ensure that existing restrictions on access to sensitive American technology cannot be circumvented through remote computing arrangements involving data centers in third countries.” Following Executive Order 14434, the rule uses the term “Super Intelligence” (SI) in place of “artificial intelligence” throughout.

The rule takes effect on the day it is published in the Federal Register. Law firm Nixon Peabody, in its client alert, said publication is expected on 19 November. Comments are due 60 days after publication.

How the rule works

The data center control. A license is needed when an exporter has “knowledge” that advanced chips are going to a data center that provides, or will provide, more than 12 million aggregate TPP (total processing performance) to or for a company headquartered in a D:5 country or Macau. The threshold counts one facility or a group of “linked data centers.” License applications are reviewed case by case.

Due diligence becomes the exporter’s job. According to Nixon Peabody’s reading, exporters are expected to obtain information about the foreign data center’s know-your-customer checks, including indirect ownership or control by D:5 entities. A data center that refuses to provide that information is a red flag.

The five-country license requirement. Shipments of advanced chips to or within Indonesia, Malaysia, Singapore, Thailand and Brazil now need a license. The exemption for US-parented companies comes with a monthly reporting duty: reports are due by the first business day of the month after each shipment. BIS says it will consider import restrictions, new license requirements, Entity List designations and other consequences for countries that tolerate evasion.

Foundries and packagers. The rule reduces the ways chip designers in license-required destinations can overcome the presumptions that apply when foundries and packagers export advanced-node logic chips.

Why it matters

For cloud and data center operators in Southeast Asia and elsewhere, compliance work just grew. Knowing your customer now means knowing your customer’s parent company, and operators that already serve Chinese-owned clients have a 90-day clock once the rule is effective.

For chip and server vendors, the Southeast Asian data center boom, much of it built on Nvidia hardware, now runs through a licensing process unless the buyer is a US-parented company that reports monthly.

For Chinese AI labs, renting overseas capacity was one of the few ways to reach frontier Nvidia hardware at scale. Closing that route adds pressure on compute supply at a time when domestic chips are still catching up.

The rule lands days after reports that the administration plans to name Nvidia, OpenAI and Anthropic among its AI “national champions”, and one day after the US and nine allies signed the San Francisco Declaration on AI supply chains.

What’s still open

  • How quickly BIS will process case-by-case license applications, and how many will be approved.
  • Which further countries the Secretary of Commerce might add to the five-country list.
  • How data centers will prove indirect ownership checks in practice, especially for resellers and multi-tenant clouds.
  • How the comment period, which closes 60 days after publication, might change the final text.

Sources

  • Bureau of Industry and Security, press release “BIS Issues Rule Addressing Remote Access to American Super Intelligence Technology,” 16 November 2026
  • Bureau of Industry and Security, interim final rule “Addressing Remote Access to Advanced Semiconductors” (RIN 0694-AK50), public inspection version
  • Nixon Peabody, export controls alert on the remote access rule, 16 November 2026

i10x Perspective

Export controls have always had a cloud-shaped hole, and this rule is the most serious attempt yet to plug it. The interesting shift is where the burden lands: not on Washington’s inspectors, but on every data center and chip distributor that now has to know who is behind its customers. Expect compliance costs to rise for Asian neoclouds and expect more capacity to move to US-parented operators who can use the reporting exemption. Follow the broader fallout in our Policy & Safety section.

Christopher Ort

Christopher Ort

Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). He is the author of “The Style Penalty”, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.

Timeline

What happened before

  1. Release

    Nemotron 3.5 Super & Lightning

    1M context · open weights

    research.nvidia.com (source, opens in a new tab)
  2. Acquisition

    NVIDIA to acquire Hugging Face for ~$12.9B

    To acquire Hugging Face · $12.9B

    Agreement to acquire Hugging Face for about $12.9B (incl. up to $1B retention equity); expected to close in H1 2027 subject to approvals.

    blogs.nvidia.com (source, opens in a new tab)
  3. Partnership

    NVIDIA guarantees up to $105B for OpenAI’s Ohio data-center campus

    With SB Energy, OpenAI · Compute · $105B

    Residual-value guarantee for OpenAI's 20-year lease of SB Energy's PORTS-Pike campus in Pike County, Ohio: 4.25 GW initially plus a 3.75 GW option, online from 2028. NVIDIA pays only on default; OpenAI must reimburse.

    investor.nvidia.com (source, opens in a new tab)
  4. Release

    Nemotron 3 Ultra

    1M context · open weights

    550B-parameter (55B active) hybrid Mamba-Transformer MoE with open weights.

    developer.nvidia.com (source, opens in a new tab)
  5. Partnership

    NVIDIA invests $2B in Marvell

    With Marvell · Investment · $2B

    cnbc.com (source, opens in a new tab)
Full Nvidia timeline ›

Editor's picks

  1. Microsoft: device code phishing can hijack AI coding agents like GitHub Copilot CLI
  2. Trinity College Philosophy Department Bans AI Use in All Courses, for Any Purpose
  3. Hour of AI 2026: CodeAI, Formerly Code.org, Launches 250 Free AI Activities for Students
  4. Qatar Launches a National AI Tutor for Schools and an AI Academic Adviser With Scale AI
  5. Novo Rebuilds Small Business Banking Around AI as Its Customers’ AI Tool Spending Jumps 493%