US BIS Issues New Rule Extending AI Chip Export Controls to Remote Cloud Computing Power
The US Department of Commerce's Bureau of Industry and Security (BIS) issued an interim final rule titled "Restricting Remote Access to Advanced Semiconductors," extending export controls from physical hardware shipments to cloud computing power. If overseas data centers provide large-scale computing power with a total processing performance (TPP) exceeding 12 million to companies in D:5 country groups such as China or in Macau, they must obtain a US export license and undergo case-by-case review; companies in Singapore, Malaysia, Thailand, Indonesia, and Brazil purchasing advanced AI chips from Nvidia or AMD must also apply for special export licenses. Affected overseas data centers must apply to BIS for a license within 90 days, or they will face subsequent export restrictions. The rule takes effect on the date of publication in the Federal Register and includes a 60-day public comment period.
Why read this
AI score 77
The new rule extends export controls from hardware shipments to remote cloud computing power and sets a TPP threshold and a 90-day grace period, which can be used to determine the compliance boundaries for cross-border computing power leasing.
Topics
Body · AI translation
On November 17, it was reported that on November 16, 2026, local time in the United States, the U.S. Department of Commerce's Bureau of Industry and Security (BIS) issued a new export control regulation on advanced AI chips, further extending the scope of U.S. export controls from physical hardware shipments to cloud computing power.
This interim final rule (IFR) titled "Addressing Remote Access to Advanced Semiconductors" runs 132 pages and was drafted under the leadership of the U.S. Department of the Treasury and the White House National Security Council, aiming to control remote access to advanced superintelligence (SI) chips and servers.
In its announcement, BIS stated that the new rule is intended to ensure that existing restrictions on access to sensitive U.S. technology are not circumvented through remote computing arrangements involving third-country data centers. With the rapid development of cloud computing and SI infrastructure, it has become increasingly easy to obtain advanced U.S. computing capabilities remotely—even when those computing capabilities are provided by data centers located outside countries subject to U.S. export controls. In response to this evolving technological landscape, the IFR further refines protections against remote access to advanced computing capabilities, aligning them with existing controls on the physical delivery (physical shipment) of sensitive U.S. technology.
Key details of the new U.S. rule
· Regarding remote computing power: If an overseas data center provides large-scale computing power to enterprises headquartered in China and other D:5 country groups or Macau, and the total processing performance (TPP) exceeds 12 million, a U.S. export license must be obtained, and BIS will review it on a case-by-case basis.
· Expanding the scope of licensing: Enterprises in Singapore, Malaysia, Thailand, Indonesia, and Brazil must now apply for special export licenses if they want to purchase advanced AI chips from U.S. companies such as Nvidia or AMD. Washington has listed the above countries as high-risk hubs for chip transshipment.
· Customer screening: Exporters and data centers must conduct strict due diligence (KYC, know your customer) to verify end-user identities and prevent circumvention of controls through shell companies; the rule still applies even if U.S. chips have already been integrated into foreign hardware. Commercial recommendation systems (such as ordinary retail or social media algorithms) are not within the scope of remote access licensing controls.
· 90-day compliance grace period: Affected overseas data centers must apply to BIS for export licenses within 90 days in order to continue providing services to Chinese customers; otherwise, they will face subsequent U.S. export restrictions.
· Effective date: The IFR formally replaces the Biden administration's previously issued "AI Diffusion Rule." The latest IFR takes effect immediately upon publication in the Federal Register and includes a 60-day public comment period.
Chain reaction from the Jensen Huang incident
Following the end of the latest U.S. midterm elections and Jensen Huang's resignation, national security hawks within the U.S. government and Congress quickly took action to tighten controls. The new rule was drafted directly by the U.S. Department of the Treasury and the National Security Council, bypassing the Department of Commerce.
Reports say that Chinese AI giants such as Alibaba and ByteDance have increasingly rented server computing power from data centers in Southeast Asia (such as Malaysia and Singapore) to train advanced AI models, and these data centers use Nvidia Blackwell and Hopper series chips that are subject to export restrictions. The new rule directly threatens this channel and may cut it off entirely.
The Financial Times reported earlier this month that Oracle had canceled a $7 billion chip leasing agreement with Tencent. Last month, it was reported that Oracle had signed a five-year agreement with Tencent to lease 100,000 advanced AI chips to it, involving multiple Oracle data centers in Southeast Asia.
Kate Koren, who once worked at the U.S. Department of Commerce, pointed out that the impact of the new rule announced on Monday on China far exceeds that of adding a single entity to the Commerce Department's trade "Entity List."
He said, "Nvidia and Jensen Huang himself were once the main promoters of the proposition of selling chips to China as a strategy. The information disclosed last week called this proposition into question, and now the wind is shifting—toward stricter export controls."
AI Trends