What the New 'Remote Access' Rule Says
On November 16, 2026, the U.S. Commerce Department's Bureau of Industry and Security (BIS) announced an 'interim final rule' targeting a critical loophole in its export controls. Previously, restrictions focused on the physical sale and shipment of high-performance
AI chips to countries of concern, primarily China. The new rule now requires a license for exporting advanced AI chips to any data center in the world if there is knowledge that it provides significant computing power to companies headquartered in China, Russia, Iran, and about 20 other nations under a U.S. arms embargo. This is a direct attempt to prevent firms from bypassing hardware bans by simply renting processing power from cloud providers in third-party countries. The rule is designated as 'interim final', meaning it takes effect immediately upon its scheduled publication on November 19, with a 60-day period for public comment to follow.
Closing a Cloud Loophole
The core of this policy shift is the focus on 'remote access'. As AI development becomes increasingly reliant on massive cloud infrastructure, the ability to rent computing power has become as valuable as owning the physical chips. U.S. officials grew concerned that companies in restricted nations were using data centers in friendly or neutral countries to train their advanced AI models, thereby circumventing the spirit of existing export controls. The rule places a new due diligence burden on chip exporters and data center operators. They must now perform robust checks on their customers to identify any direct or indirect ownership or control by entities from the restricted countries, effectively making them gatekeepers of U.S. technology. The threshold for 'significant computing power' has been defined, applying to any network that can supply over 12 million total processing performance (TPP), a metric used to measure computational power.
Five More Countries Under Scrutiny
In a related but distinct measure, the rule also imposes a new licensing requirement for the export of advanced AI chips to five specific countries: Brazil, Indonesia, Malaysia, Singapore, and Thailand. The Commerce Department identified these nations as presenting a 'higher-than-normal risk of diversion', meaning they could potentially be used as transshipment hubs to smuggle chips to restricted destinations. This move has significant implications for Southeast Asia's burgeoning data center market. However, there is a notable exemption: U.S.-headquartered companies operating in these five countries can continue to procure advanced chips without a license, provided they file regular reports with the government. This carve-out appears designed to favor American cloud giants while increasing the compliance burden for regional operators.
The Bigger Picture: A Tech Cold War Heats Up
This rule is the latest move in a broader geopolitical strategy by the United States to slow China's technological and military advancement, particularly in the field of artificial intelligence. By expanding controls from just hardware to 'compute-as-a-service', Washington is signaling its intent to regulate not just the tools of AI, but access to the capability itself. The rule formally rescinds a previous, unenforced policy from January 2025 known as the 'AI Diffusion Rule', replacing it with this more targeted and enforceable framework. For the global tech industry, this represents a major escalation. Companies in India and other nations will need to navigate these complex regulations carefully. While India is not directly targeted, its ambitious national AI mission and growing data center capacity could face new compliance hurdles and potential disruptions in accessing the most advanced hardware, underscoring the far-reaching impact of this U.S. policy shift.
















