What The US AI Chip Rule Means For Remote Access And Licences
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

The US Commerce Department issued a rule requiring licences for certain advanced AI chip exports tied to remote data center access, including access by companies headquartered in China, Macau and other designated countries. It takes effect upon Federal Register publication, scheduled for November 19, 2026; existing data centers have 90 days to obtain licences to continue covered service.

The US Commerce Department’s Bureau of Industry and Security (BIS) issued a rule on November 16 requiring licences for certain advanced AI chip exports when overseas data centers provide substantial computing capacity to companies headquartered in China, Macau or other designated countries. The measure also adds licensing requirements for chip exports to five named countries, extending US controls to some remote computing arrangements as well as chip shipments.

The rule, titled “Addressing Remote Access to Advanced Semiconductors,” sets a licence threshold for data centers anywhere in the world that provide companies headquartered in China, Macau or countries in US group D:5 access to more than 12 million aggregate TPP of computing power. Aggregate TPP is the rule’s measure of processing capacity. The text also requires a licence for covered advanced-chip exports to Indonesia, Malaysia, Singapore, Thailand and Brazil.

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

The rule is scheduled to take effect on November 19, 2026, the date listed for its publication in the Federal Register. The 90-day period is for data centers already serving covered companies: they must obtain a licence within that period to continue the covered service. A separate provision allows certain shipments already in transit to proceed if completed within 30 days of publication.

Exporters are expected to obtain “know your customer” information from foreign data centers, including details about who owns or controls their customers. BIS identifies a data center’s inability or unwillingness to provide this information as a red flag. The rule exempts computing used only for commercial recommender systems, such as online retail recommendations; chatbots and image generators do not qualify under that exemption.

At a glance
updateWhen: Issued November 16, 2026; scheduled to…
The developmentThe Bureau of Industry and Security issued an interim final rule extending US export licensing requirements to certain advanced AI chips used by overseas data centers that provide remote computing access.
Graphic titled How big is the line in the new data-centre test? Highlighted: 12,000,000 aggregate TPP, the sum of the computing power of all covered chips a data centre provides. Bars: chips at 4,800 TPP, the lower limit of the main 3A090 chip class, 2,500 chips; chips at 21,000 TPP, a line in the rule's China licensing policy, about 570 chips. Our arithmetic: 12,000,000 divided by the TPP per chip; BIS gives no chip count; more powerful chips reach the line with fewer units
The 12,000,000 aggregate TPP line in chip terms, our arithmetic. Source: BIS interim final rule, Federal Register document 2026-23529 (public inspection), pp. 25, 30, 39 and 67. Graphic: Thorsten Meyer AI.

Cloud Access Now Faces Export Review

The rule addresses a gap between buying advanced chips and renting computing capacity powered by them. A company facing restrictions on purchasing chips could potentially access computing through a data center in another country. BIS says the measure is designed to prevent existing restrictions on sensitive US technology from being circumvented through remote computing arrangements.

For cloud and data center operators, the requirements may mean new licence applications, customer ownership checks and reporting before they can serve covered users or make certain shipments. Customers could face changes in access to high-capacity computing if a provider cannot obtain a licence. The rule’s actual impact will depend on how BIS reviews applications and enforces the requirements.

Products Worth Considering

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

As an affiliate, we earn on qualifying purchases.

From Chip Shipments To Remote Use

US export controls already restricted the sale of certain advanced chips to buyers in China. The new rule broadens the regulatory focus to some cases where the chips are exported to a third country and computing access is provided remotely. It creates a worldwide data center test based on customer headquarters and computing capacity, alongside a separate country-based test for exports.

For the five named countries, shipments to companies headquartered in the United States or with a US ultimate parent do not require a licence under the specified exception, provided the exporter files monthly reports. BIS released the interim final rule on November 16. The Federal Register scheduled publication for November 19, when the rule is set to take effect. Public comments are due 60 days after publication.

““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.””

— Bureau of Industry and Security

Licensing And Enforcement Questions

The rule’s practical reach depends in part on how BIS interprets and applies its 12 million aggregate TPP threshold and reviews licence applications. The source material does not establish how quickly licences will be issued, how enforcement will work in practice, or whether BIS will add more countries. Those issues remain open as the rule approaches its effective date.

The scheduled Federal Register publication date is November 19, and the rule says it takes effect on publication. If publication proceeds as listed, the 90-day period for existing covered data center services will run from that date. The source material does not describe the outcome of any later public comments or possible changes to the interim rule.

Publication Starts The Compliance Clock

The next milestone is Federal Register publication, scheduled for November 19, 2026. If it occurs as scheduled, the rule takes effect that day, the 30-day in-transit provision begins, and comments will be due 60 days later. Data centers already serving covered companies will have 90 days from the effective date to obtain licences to continue those services.

Exporters and data center operators will need to review customer headquarters, computing capacity and ownership information against the rule’s requirements. BIS’s handling of applications, enforcement and public comments will clarify how the requirements operate in practice.

Key Questions

When does the US AI chip rule take effect?

The rule is scheduled to take effect upon Federal Register publication, listed for November 19, 2026.

Which remote data center services require a licence?

A licence is required under the rule when a data center anywhere in the world provides more than 12 million aggregate TPP of computing power to companies headquartered in China, Macau or other countries in US group D:5.

Which countries are covered by the separate export requirement?

The rule requires a licence for covered advanced-chip exports to Indonesia, Malaysia, Singapore, Thailand and Brazil, subject to the stated exception for qualifying US-headquartered companies or companies with a US ultimate parent and monthly exporter reports.

Can existing data centers continue serving covered customers?

They must obtain a licence within 90 days of the effective date to continue covered services. The rule is scheduled to become effective on November 19, 2026.

Are chatbots and image generators exempt?

No. The exemption described in the rule covers computing used only for commercial recommender systems. The source material says chatbots and image generators do not qualify.

Source: Thorsten Meyer AI

You May Also Like

Will GPT-6 Be Released By December 31, 2026?

Speculation surrounds GPT-6’s release date, with market signals suggesting high confidence but no official confirmation from OpenAI.

MCP 2026-07-28 Specification: Transport Going Stateless

The MCP 2026-07-28 specification introduces a shift toward stateless transport systems, impacting future network architecture and security.

How OpenAI’s Compute Campus Plans Put Power And Water In Focus

OpenAI confirmed it is considering the term “compute campuses.” The label comes as questions grow about electricity costs, grid upgrades and water.

Ai‑Driven Drug Discovery: Accelerating Therapeutic Development

From faster compound screening to predicting drug efficacy, AI-driven discovery is revolutionizing medicine—discover how it’s shaping the future of therapeutics.