A Commerce Department rule targeting remote access to advanced chips puts Indonesia, Malaysia, Singapore, Thailand and Brazil under a new licensing regime

Sat 21 Nov 2026 · 06:50 EST · 3 min read

In Regulation

Share

Washington has spent years stopping advanced AI chips from physically reaching China. It turns out the chips never had to travel at all.

On Monday, the US Commerce Department issued a rule aimed at Chinese companies that tap high-end AI hardware sitting in data centers outside China. The move lands hardest on Southeast Asia, a region that has been building server farms at a brisk clip.

What the new rule actually does

The rule comes from the Commerce Department’s Bureau of Industry and Security, or BIS. It is an interim final rule, with an effective date of approximately November 16, 2026.

Its core target is remote access. A Chinese firm does not need to own an Nvidia or AMD chip to use one. It can simply rent compute from a facility abroad and run its workloads over the internet.

Under the new framework, companies that want to export advanced AI chips to five countries labeled “high-risk” now need export licenses. Those countries are Indonesia, Malaysia, Singapore, Thailand and Brazil.

The rule also reaches data centers directly. Any facility providing significant compute to Chinese entities needs a license and must carry out enhanced due diligence once it crosses a threshold of over 12 million aggregate TPP.

Operators get a 90-day compliance window after the rule takes effect. Those that miss it may face future restrictions.

There is one notable carve-out. US-headquartered operators are exempt from certain licensing requirements, though they must submit monthly compliance reports.

The crypto and markets news you actually need—every morning.

Daily. Free. Join 34,000+ readers across crypto, finance, and policy.

We respect your privacy. Unsubscribe anytime.

The Megaspeed saga that forced the issue

The rule is largely a response to the scandal around Nvidia and Megaspeed, a data center operator. Megaspeed spent at least $4.6 billion on Nvidia servers between 2023 and 2025.

The controversy centers on allegations that chips were diverted and that major Chinese clients, including Alibaba, gained remote access to compute. Those are allegations, not findings, but they were enough to push regulators into action.

Markets responded quickly. Shares of Southeast Asian companies linked to data centers fell after the announcement, with some dropping by as much as 20%.

A policy with a built-in contradiction

The timing invites some awkward comparisons. In early 2026, nearly 900,000 advanced AI chips had been approved for export to Chinese companies, including some with military connections.

So the same government that cleared a large volume of chips for direct sale is now tightening the screws on rented access abroad.

What this means for Southeast Asia’s data center boom

For the region’s operators, the immediate issue is friction. Licensing, due diligence and customer screening all take time and money, and firms tied to these data centers could face delays and higher operating costs.

The competitive picture may also tilt. Because US-headquartered providers are exempt from certain licensing requirements, they could become the preferred option for large compute buyers in the region.

Brazil’s inclusion is worth noting too. The list is not purely a Southeast Asia story, which suggests BIS is mapping where compute could flow to China rather than drawing a regional boundary.

For investors, the key dates sit on the calendar. The rule’s effective date of approximately November 16, 2026, and the 90-day window that follows, mark the period when operators must show they can comply.

Disclosure: This article was edited by John Chen. For more information on how we create and review content, see our Editorial Policy.