The U.S. Commerce Department has introduced new regulations targeting Chinese companies’ access to advanced AI chips located in data centers outside of China. This rule affects five countries identified as “high-risk”: Indonesia, Malaysia, Singapore, Thailand, and Brazil, and will require new export licenses for businesses sending advanced AI chips to these nations.
The interim final rule from the Bureau of Industry and Security (BIS) is set to take effect around November 16, 2026. The main focus of the rule is to mitigate concerns regarding remote access to high-end chips, such as those produced by Nvidia and AMD. Chinese firms will no longer be able to freely rent computational resources from data centers abroad without regulatory oversight.
Facilities providing significant computing power to Chinese firms are now required to obtain licenses and must conduct enhanced due diligence if their operations exceed an aggregate threshold of over 12 million aggregate TPP (Trade Promotion Programs). Following the rule’s implementation, operators will have a 90-day compliance period; failure to meet these requirements could lead to further restrictions.
However, there is a notable exemption: U.S.-based operators are not subject to certain licensing demands, but they must still submit monthly compliance reports.
This new regulation arises partly from concerns surrounding Megaspeed, a data center operator that invested heavily—at least $4.6 billion—in Nvidia servers from 2023 to 2025. Allegations have surfaced that these chips were diverted to allow major Chinese clients, including Alibaba, remote access to advanced computational resources. While these claims remain allegations, they have prompted swift regulatory action.
Financial markets reacted quickly to the announcement, with shares of Southeast Asian firms linked to data centers experiencing drops of up to 20% following the news.
As the policy unfolds, there are notable contradictions. Earlier in 2026, nearly 900,000 advanced AI chips received export approval for Chinese companies, some with military connections. This raises questions given the recent push for tighter regulations on remote access to these very technologies.
For operators in Southeast Asia, the immediate implications include increased operational friction, as the new licensing and due diligence requirements will likely lead to delays and higher costs. Additionally, as U.S.-headquartered providers are exempt from some licensing requirements, they may gain a competitive edge as the preferred choice for large computational needs in the region.
As the rollout approaches, the significance of key deadlines will be critical for the affected operators. The effective date of the rule in November 2026, coupled with the following 90-day compliance window, underscores the urgency for these firms to adapt their practices accordingly.


