Washington and Beijing are moving toward a new phase of technological confrontation just as their leaders prepare to reinforce a fragile trade truce. The immediate trigger is a proposed US rule restricting Chinese companies’ remote access to advanced American artificial-intelligence chips through overseas data centers. The broader dispute is about who controls the infrastructure, capital and standards that will shape the next generation of computing.

The US Commerce Department’s measure would require overseas data centers serving companies in China and several other countries to identify end users, while imposing additional licensing requirements on chip shipments to countries viewed as vulnerable to diversion. The objective is to prevent Chinese firms from obtaining restricted processors indirectly through facilities in places such as Singapore and Malaysia, rather than importing them directly.

For Washington, the policy reflects a basic security argument: access to the most powerful chips can accelerate military systems, cyber capabilities and surveillance as well as commercial AI. US officials and lawmakers have therefore treated export controls as a way to slow China’s access to frontier computing while allied countries build a common technology-security framework. Nine countries, including Singapore and the United Kingdom, recently signed a US-backed declaration on protecting sensitive technologies, according to CNBC.

Why the loophole matters

Restrictions on physical chip shipments are harder to enforce when computing can be rented remotely. A Chinese company may not need to own a restricted processor if it can access one in a foreign data center. That makes cloud infrastructure an extension of the export-control battlefield and explains why the new proposal reaches beyond customs borders.

Analysts cited by Reuters Breakingviews estimate that nearly two-thirds of China’s projected 2027 computing capacity could be exposed to the new restrictions. The figure is a forecast rather than a measurement of current capability, but it illustrates the potential scale: the policy could affect not only Chinese technology companies, but also foreign cloud providers, data-center operators and chip distributors.

American technology companies face a conflicting set of incentives. China remains a major market and an important manufacturing hub, yet every transaction can attract scrutiny from Washington and countermeasures from Beijing. The resulting uncertainty may encourage firms to diversify supply chains and reduce exposure, even where no formal ban exists. That could make technology more resilient in the long term, but more expensive and less efficient in the short term.

Beijing’s dilemma

China has denounced the proposed rules as economic coercion and long-arm jurisdiction, arguing that they interfere with global supply chains. Its government has also expanded its own export-control toolkit, including restrictions affecting rare earths and critical minerals, and has proposed controls covering technologies such as AI, biotechnology and solar power, according to The Boston Globe.

That gives Beijing leverage, but using it carries costs. Strong retaliation could threaten the trade understanding that has helped stabilize relations, including an apparent exchange in which Washington delayed some technology measures while critical minerals continued flowing to US industries. China could also damage its own exporters if foreign customers accelerate efforts to find alternative supplies.

Beijing therefore has several possible responses: impose targeted restrictions on American companies, tighten licensing for strategic materials, challenge the US measure diplomatically, or retaliate gradually while preserving room for negotiation. A restrained response would not mean acceptance. It could reflect a calculation that China benefits from time to expand domestic chip production and reduce dependence on American equipment.

What comes next

The immediate test will be whether the leaders’ summit can separate crisis management from the deeper technology contest. Reuters reported that the two sides were exploring cooperation on AI safety, including a possible mechanism for notifying each other about serious incidents, although that arrangement had not become operational. Hardware restrictions could make such cooperation politically harder, even as shared risks make it more necessary.

The dispute is unlikely to end with one rule or one summit. Washington is moving from controlling products to controlling access, finance and infrastructure. Beijing is responding by broadening the range of technologies and materials it can regulate. Other countries are being asked to choose how closely they will align with either system, while companies must plan for duplicated supply chains and competing standards.

The central question is whether guardrails can remain narrow enough to protect national security without turning the global technology economy into rival blocs. If both governments use controls selectively and preserve channels for dialogue, the rivalry may remain manageable. If each new restriction is answered with a broader countermeasure, the trade truce could become little more than a pause between escalations.

Sources