What the New Rule Actually Does
On November 16, 2026, the U.S. Commerce Department's Bureau of Industry and Security (BIS) announced an interim final rule that significantly tightens control over American AI technology. The core of the new measure is to treat renting powerful AI computing
power similarly to shipping the physical chips themselves. Specifically, data centers anywhere in the world now require a U.S. license if they provide significant computing capacity to companies headquartered in restricted nations like China, Macau, Russia, or Iran. This means US cloud providers and any data centers using advanced US technology must conduct due diligence on their foreign customers to prevent misuse.
Closing the Cloud Loophole
Previously, U.S. export controls focused on the physical sale and shipment of high-performance AI semiconductors, like those from Nvidia, to China. However, a major loophole existed: companies in restricted countries could simply bypass these hardware bans by renting computing power from data centers in third-party countries. These data centers, often located in Southeast Asia or the Middle East, use the very same restricted U.S. chips. By setting up proxy entities, Chinese firms could train their large AI models remotely, effectively nullifying the hardware export bans. This new rule directly targets this 'remote access' workaround, shifting the focus from controlling hardware to controlling access to computational power itself.
Who Is Affected by the Rule?
The primary targets are companies headquartered in or with ultimate parent companies based in China and other nations under a U.S. arms embargo. However, the regulation's ripple effects are much broader. Major U.S. cloud providers like Amazon Web Services, Microsoft Azure, and Google Cloud now face immediate and complex compliance obligations. They must implement robust 'Know Your Customer' (KYC) programs to verify the identity of foreign users and report suspicious activity. Furthermore, the rule introduces new licensing requirements for advanced chip exports to Brazil, Indonesia, Malaysia, Singapore, and Thailand, countries identified as having a 'higher-than-normal risk of diversion' where chips could be rerouted to restricted nations.
The Challenge of Enforcement
While the rule is clear in its intent, enforcement presents a significant challenge. Cloud providers must now monitor not just their direct clients, but also the customers of their foreign resellers. Verifying the true ultimate beneficiary of computing power can be difficult, especially with complex international corporate structures and the use of intermediaries. The rule sets a specific technical threshold for what constitutes 'significant' computing capacity—12 million aggregate Total Processing Performance (TPP)—which requires operators to track usage across potentially linked data centers. Existing data centers already serving restricted clients have a 90-day grace period to obtain the necessary licenses to continue their services.
The Bigger Picture for Global AI
This policy marks a fundamental change in the global AI landscape, treating computing access as a controlled commodity just like physical hardware. The immediate effect is a squeeze on Chinese AI companies, which have heavily relied on U.S. chips for training their most advanced models. In response, a Chinese Foreign Ministry spokesperson condemned the new rules, accusing the U.S. of misusing export controls and vowing to take necessary measures to protect its interests. This move could accelerate China's push for self-sufficiency in both semiconductor manufacturing and cloud infrastructure. For the rest of the world, including India's burgeoning tech sector, it introduces new compliance burdens and highlights the increasing fragmentation of the global technology ecosystem.
















