🔍 Read the full analysis: The US AI Chip Rule: Remote Access, Licences, And Effective Dates on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
The US Commerce Department’s Bureau of Industry and Security issued a rule covering advanced AI chips used by data centres that provide substantial computing access to certain foreign companies. It is scheduled to take effect November 19, 2026; the 90-day period is for existing data-centre relationships to obtain licences, not a delay before the rule starts.
The US Commerce Department’s Bureau of Industry and Security issued a rule on November 16 requiring licences for some exports of advanced chips used by overseas data centres, including facilities that give certain foreign companies remote access to computing capacity. The rule is scheduled to take effect November 19, 2026, and also requires licences for covered chip exports to Indonesia, Malaysia, Singapore, Thailand and Brazil.
The interim final rule, titled “Addressing Remote Access to Advanced Semiconductors”, amends the Export Administration Regulations. One new test applies worldwide: a licence is required for covered chips sent to a data centre that provides, or will provide, more than 12 million aggregate TPP of computing power to companies headquartered in China, Macau or other countries in the US control group D:5, or whose ultimate parent is based there. TPP means total processing performance, a measure used by US export rules.
The threshold can cover capacity at linked data centres, including digitally linked facilities, and the rule can also apply when an intermediary serves a customer that trains or provides “SI” models to a D:5 company. It covers specified advanced integrated circuits and computers containing them. The rule separately requires licences for covered chip exports to five named countries: Indonesia, Malaysia, Singapore, Thailand and Brazil. For shipments to companies headquartered in the United States, or with a US ultimate parent, in those countries, the source says an exception is available if the exporter submits monthly reports.
The rule is effective on its scheduled Federal Register publication date, November 19, rather than after 90 days. Existing data-centre relationships that would need a licence may continue during a 90-day period, but providers must obtain a licence within that window to keep serving the covered customers. A separate provision allows certain shipments already on their way to be completed within 30 days of publication. Public comments are due 60 days after publication.
US export rule · explainer
How the new rule closes the cloud route to US AI chips
Buying the chips was already restricted. Renting them abroad was the gap.
The route (schematic)
Before: rent instead of buychips shipped to a data centre in a third country, computing power sold remotely New: a licence gateabove 12,000,000 aggregate TPP for companies headquartered in China, Macau or other D:5 countries Already serving them? 90 daysto obtain a licence, or face future export restrictions
What the rule does, step by step
- BuyCompanies headquartered in China already need a licence to buy advanced US AI chipsexisting rule
- RentRemote access through data centres in third countries was the workaroundthe gap
- GateNew licence test for data centres anywhere above 12,000,000 aggregate TPPsection 744.24
- CheckExporters need the data centre’s know-your-customer checks; no answer is a red flagdue diligence
- FiveEvery advanced-chip export to Indonesia, Malaysia, Singapore, Thailand or Brazil needs a licence, US-headquartered buyers exceptedcountry test
Source: US Commerce Department, Bureau of Industry and Security, interim final rule “Addressing Remote Access to Advanced Semiconductors”, Federal Register document 2026-23529 (public inspection), pp. 4, 10, 24, 30 and 32; BIS press release, Nov 16, 2026. The diagram is a schematic.
Graphic: Powered by Thorsten Meyer AI
Cloud Access Becomes an Export-Control Issue
The rule reaches beyond the physical sale or shipment of chips. A company barred from buying advanced US chips may still depend on computing services at a foreign data centre; the new test seeks to bring some of that remote access within export licensing. This makes the location and ownership of data centres, their customers and connected facilities relevant to compliance.
For cloud providers, the requirements may mean reviewing customer relationships, tracing corporate ownership and determining how computing capacity is allocated across linked facilities. Exporters are expected to obtain data centres’ know-your-customer information, including details of who owns or controls their customers. The source says a facility’s inability or refusal to supply that information is a red flag. Customers may also face service changes while providers determine whether a licence is needed.
The rule is a licensing requirement, not a blanket prohibition on all chip exports or all data-centre services in the countries named. Its practical effect will depend on the threshold, customer links, exemptions and how BIS processes applications. Those distinctions matter to businesses planning capacity, as well as governments and firms concerned about the reach of US export controls.
Products Worth Considering

Samsung Galaxy Tab A11+ Tablet (6GB RAM, 128GB Storage, Gray)

Bose QuietComfort Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, Eucalyptus Green

DJI Mic 3 (2 TX + 1 RX + Charging Case), Wireless Microphone
As an affiliate, we earn on qualifying purchases.
From Chip Shipments to Remote Use
Earlier US controls focused on the export, re-export or transfer of advanced chips and related systems. The new rule adds a specific test for data centres providing computing access, addressing a gap BIS says could allow restrictions to be bypassed through third-country facilities. It applies to covered chips even when incorporated into a foreign-made GPU board or server, according to the supplied account of the rule.
BIS released the text on November 16, and the Federal Register scheduled publication for November 19 as document 2026-23529. Because it is an interim final rule, it applies upon publication while comments are collected. The source reports that the rule says the changes can be imposed without prior notice and comment. Confusion over the dates has arisen because the text includes a 90-day period, but that period concerns existing customer relationships, not the effective date.
The rule uses “SI,” meaning “Super Intelligence,” in place of “AI,” following an executive order, while retaining the cited chip classifications 3A090 and 4A090. The supplied material does not provide the full technical schedule or a complete account of how every possible service arrangement will be assessed.
Licensing Scope and Enforcement Remain Open
The rule’s practical reach will depend on how BIS interprets and applies the 12-million aggregate TPP threshold, linked data-centre arrangements and the provisions involving intermediaries. The source does not specify how quickly licences will be decided, how many applications are expected or what enforcement actions may follow. It also does not establish whether additional countries will be added.
Some exemptions are described, including one for commercial recommender systems such as those used by online retailers. The source says chatbots and image generators are not covered by that exemption. Further operational guidance may be needed for providers deciding how to classify services and demonstrate compliance. The scheduled November 19 publication date is stated in the source; if publication timing changes, the effective date would need to be checked against the Federal Register.
Publication, Comments and Licence Reviews
The rule is scheduled for publication in the Federal Register on November 19, 2026, when it takes effect if that schedule holds. Existing providers relying on the 90-day window will need to seek licences within the period measured from the effective date. Exporters and data centres will also need to review customers, ownership links and computing capacity against the new tests.
Comments are due 60 days after publication. BIS may receive feedback before deciding whether to revise the interim rule. Governments and companies are studying the requirements, but the source material does not identify a later decision date or provide a timetable for licence processing.
Key Questions
When does the US AI chip rule take effect?
It is scheduled to take effect on November 19, 2026, the date set for Federal Register publication. The 90-day period is a window for certain existing data-centre relationships to obtain licences, not a delayed start date.
Which remote data-centre arrangements may require a licence?
A licence may be required when covered chips go to a data centre providing more than 12 million aggregate TPP of computing power to companies headquartered in China, Macau or other D:5 countries, or to companies with an ultimate parent there. Linked facilities and certain intermediary arrangements may count.
Which countries are named in the separate export test?
The rule separately names Indonesia, Malaysia, Singapore, Thailand and Brazil. Covered chip exports to those countries require a licence, subject to the stated exception for qualifying US-headquartered companies or companies with a US ultimate parent when monthly reports are filed.
Does the rule ban all chip exports to those countries?
No. The supplied material describes a licensing requirement, not a blanket ban. Whether a particular shipment can proceed depends on the rule’s conditions and any applicable exception or licence.
What remains unresolved?
The source does not say how quickly BIS will issue licences, how enforcement will work in practice or whether more countries will be added. The application of the capacity and customer-link tests to specific service arrangements may also require further guidance.
Source: ThorstenMeyerAI.com
