U.S. AI chip export controls are tightening again! The plan is to bring seven countries—including Malaysia, Singapore, Thailand, Brazil, Indonesia, Norway, and South Africa—under stricter export licensing requirements.
According to a November 12 report by The Information, the U.S. Department of Commerce is finalizing a new set of AI chip export control rules. The rules focus on one issue: the United States will not allow restricted countries to bypass controls by remotely using advanced AI chips through overseas data centers in third countries.
The proposed rules cover Nvidia Blackwell and more advanced AI chips. Importers in the countries concerned may be required to disclose the ultimate recipients and intended uses of the chips to the U.S. Department of Commerce, and to submit to ongoing oversight.
However, data centers operated locally by U.S. companies may qualify for special licensing exemptions, provided they restrict remote use of chips above a specified computing-power threshold by countries under U.S. arms embargoes and meet customer information reporting requirements.
Notable points:
1. Southeast Asia is a key focus of regulation. Malaysia, Singapore, Thailand, and Indonesia are all on the proposed list, meaning overseas AI computing deployments will face stricter compliance reviews.
2. Nvidia’s overseas market is under pressure. Exports and deliveries of Blackwell and subsequent advanced chips, as well as data center deployments, could be affected.
3. The United States is tightening control over the flow of computing power. Controls are no longer focused solely on who buys the chips; they are also beginning to track who uses them remotely and how much computing power they use.
The new rules could be announced as early as November, but they are still under interagency consultation and have not yet been finalized.
This sends an important signal: U.S. AI chip export controls are extending beyond the flow of physical chips to encompass access to and use of global computing power.
The future contest will be about more than AI chips—it will also be about control over the world’s advanced computing power.
According to a November 12 report by The Information, the U.S. Department of Commerce is finalizing a new set of AI chip export control rules. The rules focus on one issue: the United States will not allow restricted countries to bypass controls by remotely using advanced AI chips through overseas data centers in third countries.
The proposed rules cover Nvidia Blackwell and more advanced AI chips. Importers in the countries concerned may be required to disclose the ultimate recipients and intended uses of the chips to the U.S. Department of Commerce, and to submit to ongoing oversight.
However, data centers operated locally by U.S. companies may qualify for special licensing exemptions, provided they restrict remote use of chips above a specified computing-power threshold by countries under U.S. arms embargoes and meet customer information reporting requirements.
Notable points:
1. Southeast Asia is a key focus of regulation. Malaysia, Singapore, Thailand, and Indonesia are all on the proposed list, meaning overseas AI computing deployments will face stricter compliance reviews.
2. Nvidia’s overseas market is under pressure. Exports and deliveries of Blackwell and subsequent advanced chips, as well as data center deployments, could be affected.
3. The United States is tightening control over the flow of computing power. Controls are no longer focused solely on who buys the chips; they are also beginning to track who uses them remotely and how much computing power they use.
The new rules could be announced as early as November, but they are still under interagency consultation and have not yet been finalized.
This sends an important signal: U.S. AI chip export controls are extending beyond the flow of physical chips to encompass access to and use of global computing power.
The future contest will be about more than AI chips—it will also be about control over the world’s advanced computing power.
