Washington extends chip restrictions to remote computing, data centres worldwide face a new test
Listen to this article
Read by Anchor
Around 750 chips. That is the approximate threshold a data centre falls under according to a new rule issued by the U.S. Department of Commerce on 16 November 2026. As TechCrunch reported from the rule’s text, the restrictions apply to any centre or network capable of providing more than 12 million units of Total Processing Performance, roughly equivalent to 750 H100 or H200 chips from Nvidia.
The rule is issued by the Bureau of Industry and Security (BIS) of the department as a “temporary final rule”, meaning it takes effect before the comment period is completed. The bureau says in its statement that it was written at the direction of the Treasury Department and the National Security Council, and that its purpose is to prevent circumvention of existing controls on sensitive U.S. technology through remote-computing arrangements.
The loophole being closed
Until now, oversight focused on shipping the chip itself. But a Chinese company does not need to ship anything if it can rent computing capacity in a data centre located in a third country. The ministry describes the new rule as aligning remote-access protection with the controls that apply to physical shipments. TechCrunch cited Reuters that Chinese AI firms were racing the previous week to book overseas computing capacity in anticipation of tighter restrictions.
What the rule actually requires
According to TechCrunch, exporters will need a licence to export chips and servers, or to re-export or move them to data centres anywhere in the world if those centres can provide computing to companies in China or other restricted nations. This includes indirect access through intermediaries and brokers, and operators are required to trace the ownership structures and activities of their customers to ensure no party belongs to the restricted-countries group (D:5), which includes China, Russia, Iran, North Korea and 18 other states. Existing data centres that offer covered services have 90 days from the rule’s effective date to obtain a licence. The rule exempts commercial recommendation systems, such as those used in e-commerce and social-media platforms.
The rule also adds licence requirements for exporting advanced chips to Malaysia, Singapore, Indonesia, Thailand and Brazil, based on a “higher-than-usual diversion risk”. These requirements do not apply to chips destined for U.S., based companies. The rule also repeals the “diffusion rule” issued by the Biden administration in January 2025.
What is still unclear
The bureau’s statement does not specify an effective date, the entities covered or licensing timelines, and refers to the rule’s text in the Federal Register (document 2026-23529, titled “Remote Access to Advanced Semiconductors”). We could not read the full text directly, so the details above come from TechCrunch’s coverage. Until the official text is published and the effective date is known, the exact 90-day deadline remains uncertain.
Who holds the computing key?
The sovereign lens is clear: the chip may be physically inside a country’s borders, but the right to lease it to a specific client now depends on a licence from another country. Territorial sovereignty is no longer enough to control the computing decision, and the operator of a data centre effectively becomes the party responsible for verifying the identity and ownership of its customers.
What this means for the region
The bureau’s statement and TechCrunch’s coverage do not mention Gulf or Middle-East countries within the rule’s scope, and the five countries added to the licence list are not from the region. Therefore it is inaccurate to say the rule is aimed at the Gulf.
But its reach is global as described: any data centre outside the United States, including those in the region, that houses U.S. chips and exceeds the 12 million processing-unit threshold falls under the licence regime if it can provide computing to customers linked to restricted nations. This is a reading of the rule as reported, not a definitive judgment; confirmation requires the official text.
The likely practical impact for regional computing operators is tighter scrutiny of customer ownership and intermediary chains before signing large contracts, and preparation of licence applications if they already offer services that fall within the rule’s scope. The most important next step is the publication of the text in the Federal Register and the clarification of the effective date.