The Daily Ledger

TechnologyNews

US Tightens AI Chip Restrictions, Raising Concerns for Alibaba and Tencent

New US export rules target remote access to advanced American computing chips through overseas data centres, closing a route Chinese technology firms used to obtain AI computing power. Alibaba says the measures could affect some global data-centre expansion plans, while Tencent warns that tighter access may slow future model development.

The Daily Ledger Staff4 min read
Editorial illustration of data-centre analysts reviewing advanced AI chip access routes across a global cloud network, with a secure barrier representing new export controls
Illustration generated with AI

New US export controls on advanced computing technology are creating fresh uncertainty for Chinese technology companies that rely on overseas data centres to access powerful artificial intelligence chips. The rules, issued by the Commerce Department's Bureau of Industry and Security, extend restrictions beyond the physical shipment of chips to certain remote computing arrangements.

The measures are intended to prevent companies headquartered in China and other restricted destinations from using data centres in third countries to access large amounts of computing power based on controlled US technology. The change could affect how cloud providers and AI developers plan infrastructure outside China.

Why the rules reach beyond chip shipments

Earlier export controls restricted the direct sale or transfer of some advanced US semiconductors to China. However, companies could also rent computing capacity from data centres located elsewhere, allowing them to use advanced chips remotely without importing the hardware themselves.

The new interim final rule introduces licensing requirements for certain advanced computing items when they are destined for data centres that provide significant computing capacity to companies headquartered in China, Macau or other specified restricted jurisdictions. It also adds licensing requirements for covered advanced chips and related items sent to Indonesia, Malaysia, Singapore, Thailand and Brazil, which the US identifies as higher-risk diversion routes.

The rule requires data-centre operators and other parties involved in relevant transfers to examine customers and ownership structures. The US government says the objective is to close a gap between controls on physical exports and the growing ability to rent computing power across borders.

Alibaba reviews the impact on its expansion plans

Alibaba said the new restrictions could affect some of its global data-centre expansion plans and that it was assessing the possible consequences. The company has been investing in cloud infrastructure, chips and AI models as demand for computing services grows.

For the July-to-September quarter, Alibaba reported that revenue from its AI cloud and computing services increased 53% year over year to nearly 61 billion yuan. Its capital expenditure rose 60% to 50.5 billion yuan as it continued investing in infrastructure.

The figures underline why access to advanced computing capacity matters to cloud providers. Training and serving increasingly capable AI models requires large amounts of specialised hardware, and restrictions can influence where companies build facilities and which customers they can serve.

Tencent warns of pressure on model development

Tencent has also warned that the restrictions reduce the advanced computing resources available to Chinese technology companies. Company president Martin Lau said the rules could affect the timelines for training future versions of its AI models.

Chinese AI developers have relied heavily on advanced Nvidia processors, including through computing services located outside mainland China. Researchers have estimated that remote access to chips through overseas data centres could represent a substantial share of China's available AI computing capacity in the coming years.

Domestic chip development may gain urgency

The restrictions could increase pressure on Chinese companies to develop and deploy domestic alternatives. Alibaba has been expanding its own chip and model programmes, while other technology groups are also investing in local semiconductor supply chains and computing infrastructure.

However, replacing leading imported processors is not an immediate process. Chip performance, manufacturing capacity, software compatibility and the availability of large-scale computing clusters all affect how quickly domestic systems can support advanced AI training.

What the rules mean for global data-centre operators

The changes also matter to companies operating data centres outside China. Operators may need stronger customer checks, clearer information about ultimate ownership and additional licensing procedures before providing certain levels of computing capacity to restricted entities.

Countries including Singapore and Malaysia are important locations in the regional data-centre market. The new requirements could therefore affect compliance planning, hardware procurement and commercial arrangements for operators serving international customers.

A more complex global AI infrastructure market

The new controls show how competition over AI is increasingly shaped by access to infrastructure as well as model research. Governments are looking beyond where chips are shipped and examining how computing capacity is accessed, rented and used across borders.

For Alibaba, Tencent and other Chinese developers, the immediate task is to assess which infrastructure plans and customer arrangements are affected. For data-centre operators and chip suppliers, the challenge is to understand the new licensing and due-diligence obligations while continuing to serve legitimate global demand.

Sources

This story was prepared with the help of automated drafting tools and was reviewed and fact-checked by our editors before publication. See our Editorial Policy.

Report a problem with this story · Copyright & takedown