Malaysia Data Centres Face New US Licence Rule on AI Chips
Malaysia data centres face a new US licence rule on advanced AI chips, adding pressure to an RM95.8 billion investment boom tied to Chinese demand.

Malaysia data centres face a new hurdle after the United States moved to require licences for shipments of advanced AI chips to the country. The US Commerce Department’s Bureau of Industry and Security (BIS) announced the interim final rule on 16 November 2026, naming Malaysia alongside Indonesia, Singapore, Thailand and Brazil as destinations with what it called “a higher-than-normal risk of diversion”.
The rule is scheduled for publication in the Federal Register on 19 November and takes effect on publication. It lands as Malaysia’s fastest-growing investment category comes under scrutiny over the end users of the computing power it hosts.
What the US AI chip licence rule requires
Exporters will need a licence to send specified advanced AI chips, servers and related technology to the five countries, or to transfer them within those countries, with applications reviewed case by case. The rule is not a blanket ban on shipments to Southeast Asia, but data centre operators in the four Southeast Asian countries will now need US approval to receive advanced chips from Nvidia and other American manufacturers.
Shipments to companies headquartered in the United States, or with a US-headquartered ultimate parent, are exempt, provided exporters file monthly reports on the companies involved, their ownership and the chips’ destination. Malaysian companies receiving chips directly cannot automatically claim this exemption.
A second element targets remote access. Suppliers may not knowingly ship covered items to any data centre, anywhere, that provides more than 12 million units of aggregate total processing performance (TPP) to companies headquartered in China, Macau or other Country Group D:5 destinations without a licence. Existing arrangements have 90 days from the effective date to obtain authorisation, while computing systems used only for commercial recommendation services, such as those run by online retailers and social media platforms, are exempt. Operators, exporters and chipmakers must also carry out deeper customer checks, including on indirect ownership or control by parties in restricted destinations.
The Commerce Department said the aim was to stop existing controls being bypassed through “remote computing arrangements involving data centres in third countries”. The rule also formally rescinds the Biden-era AI Diffusion Rule, which was issued in January 2025 but never enforced.
Why Malaysia data centres are exposed
The stakes for Malaysia data centres are high. Data centre and cloud projects accounted for RM95.8 billion (about US$23 billion) of approved investment in the first half of 2026, or about 44% of the RM218.5 billion total, according to the Malaysian Investment Development Authority (MIDA).
A significant share of that capacity is understood to serve Chinese clients, which is why the remote-access test matters. UOB Kay Hian said the impact on Malaysia would depend largely on who owns the receiving companies and who uses the computing capacity.
Malaysia already runs its own controls. Since July 2025, exports, re-exports, transshipments and transits of high-performance US-origin AI chips have required a permit under the Strategic Trade Act 2010. The country does not require a permit or licence to import AI chips.
Market reaction and government response
Data centre-linked stocks on Bursa Malaysia were already under pressure before the rule was announced. On 12 November, YTL Power fell 39 sen, or more than 8%, to RM4.40, while YTL Corp, NationGate, Gamuda, Sunway Construction and IJM Corp also declined, The Edge Malaysia reported.
The Ministry of Investment, Trade and Industry (MITI) said on 13 November that it had never cleared Singapore-based Megaspeed International or its Malaysian affiliate Speedmatrix, and that earlier visits to data centres were routine monitoring rather than regulatory clearance. The ministry said it was working with the Royal Malaysian Customs Department to determine whether chips brought into Malaysia had been re-exported and, if so, whether that complied with Malaysian law. MITI added that Malaysia does not enforce other countries’ unilateral export controls, although local companies are expected to take them into account. The new US rule does not name Megaspeed or Speedmatrix.
As of 17 November, MITI had not issued a separate statement on the new US measure, ProductNation reported. Singapore and Indonesia said they were studying the rule.
What happens next
BIS will accept public comments for 60 days after the rule appears in the Federal Register. Operators with existing arrangements that cross the computing threshold will have 90 days to seek authorisation.
For investors, the immediate question is how quickly licences will be granted and how many Malaysian projects depend on Chinese clients. Operators serving US cloud providers may be less affected, given the exemption for US-headquartered end users. Similar questions hang over neighbouring Thailand, where AI-led exports lifted third-quarter GDP growth, and over lenders with data centre exposure, a risk Japan’s megabanks have already flagged.
The timing adds a diplomatic layer. The rule was announced as US President Donald Trump headed to Manila for ASEAN meetings, ahead of the APEC leaders’ meeting in Shenzhen. How the licensing process works will matter for a sector that has become a pillar of Malaysia’s investment story, and for Malaysia’s semiconductor companies that supply it.







