What Is The New Rule?
On November 15, 2026, the US Department of Commerce's Bureau of Industry and Security (BIS) announced an interim final rule aimed at preventing companies in restricted nations, like China, from using cloud services to bypass export controls. Previously,
export bans focused on the physical shipment of high-performance AI semiconductors. This new regulation extends controls to the act of accessing these chips remotely. It requires US cloud providers and other data centers worldwide to obtain a license if they provide significant computing power to companies headquartered in China, Macau, and other designated countries of concern. This effectively treats renting access to a supercomputer the same as selling the physical hardware.
Closing a Critical Loophole
The rule addresses a critical gap in US national security policy. While Washington had already barred the direct sale of top-tier AI chips from companies like Nvidia to China, Chinese firms could still tap into that same computing power by renting it from data centers in third countries. This 'cloud loophole' meant that the intended restrictions were being circumvented. The new measures force data center operators to implement robust 'know-your-customer' (KYC) procedures to verify the ultimate parent company of their clients, preventing them from unknowingly providing services to restricted entities through complex ownership structures or intermediaries.
Who Is the Primary Target?
The policy is squarely aimed at slowing China's development of advanced AI, which the US fears could be used for military applications. The regulations target companies headquartered in or whose ultimate parent company is based in China, Macau, or other nations in the US arms-embargoed 'Country Group D:5', which includes Russia and Iran. Furthermore, the rule identifies five countries—Indonesia, Malaysia, Singapore, Thailand, and Brazil—as presenting a 'higher-than-normal risk of diversion'. Consequently, any advanced AI chips shipped to these countries now also require a license, unless the end-user is a US-headquartered company.
Impact on Big Tech and Data Centers
The rule places a significant compliance burden on US cloud giants like Amazon Web Services, Microsoft Azure, and Google Cloud, as well as other data center operators globally. These companies must now invest in sophisticated monitoring and verification systems to ensure they are not in breach of the new regulations. The rule sets a specific threshold for what constitutes 'significant' computing power, requiring a license for providing access to over 12 million aggregate Total Processing Performance (TPP). Data centers already providing services to restricted companies have a 90-day grace period from the rule's effective date to obtain the necessary licenses to continue their services.
What Happens Next?
This is an 'interim final rule', which means it takes effect almost immediately upon publication in the Federal Register, but a 60-day period is provided for public and industry comment. This allows for potential adjustments based on feedback from the tech industry, which may raise concerns about the complexity and cost of implementation. China's foreign ministry has already voiced its opposition, calling the move 'economic coercion'. The rule marks a new phase in the US-China tech rivalry, shifting the focus from controlling hardware exports to governing access to computational power itself. This escalation will likely have far-reaching consequences for global AI development and the international operations of Big Tech.
















