The economic rivalry between the United States and China is entering a new phase, one defined less by outright decoupling and more by a calculated struggle for leverage. Recent reports indicate that several Chinese tech giants, including Alibaba, ByteDance, and Tencent, have remotely accessed computing power from Nvidia's advanced GB300 chips via data centers in Thailand, Malaysia, and Japan. Days later, the White House accused Moonshot AI, another Chinese firm, of using one of Nvidia's most sophisticated processors.
Whether these actions are legal or expose gaps in US export controls, they underscore a fundamental reality: restricting direct access to chips does not necessarily prevent Beijing from tapping into America's advanced computing capabilities. As technology networks become increasingly global, controlling the physical movement of hardware no longer guarantees control over its use.
From decoupling to selective engagement
For years, both Washington and Beijing have pursued strategies aimed at reducing their vulnerabilities to each other. The United States has raised tariffs, restricted China's access to advanced technologies, and coordinated with allies on export controls targeting strategically sensitive industries. Beijing, in turn, has sought greater financial and technological autonomy, reducing reliance on US-dominated payment systems while leveraging its role in critical supply chains, including rare earths, to safeguard its economic and security interests.
China's technological progress has been rapid. According to the 2026 Stanford AI Index, the performance gap between leading US and Chinese AI models has narrowed. While America still produces more top-tier models and high-impact patents, China leads in publication volume, citations, patent output, and industrial robot installations. This points to China's growing ability to develop its AI sector indigenously—and the limits of US efforts to contain that development.
The United States hosts an estimated 5,427 data centers, more than ten times any other country, yet almost every leading chip is fabricated by Taiwan Semiconductor Manufacturing Company (TSMC), making Washington dependent on a single foundry. America's ambition to lead AI development thus relies heavily on international production networks.
President Donald Trump's decision to resume sales of Nvidia's H200 AI chips to "approved customers" in China was a calculated gamble. By allowing access to a less advanced generation while embargoing the cutting-edge Blackwell chips, Washington aimed to exert influence over China's AI ecosystem and strengthen its bargaining position in broader trade negotiations. But by capping H200 deployment in the mainland and using them to train models, China continues to support domestic AI growth while benefiting from American technology—without creating long-term dependence. This effectively blunts the leverage Trump seeks.
Beijing's perception that Washington is trying to constrain its technological advancement has only accelerated its push for a self-reliant semiconductor industry. Nvidia CEO Jensen Huang has himself blamed US withdrawal from the Chinese market for fueling China's semiconductor ambitions.
The limits of trade barriers
A White House report released in August, titled "The Great Transshipment Scam," accuses Chinese exporters of routing goods through more than 40 countries to evade US tariffs. The report reflects the limits of Washington's efforts to reduce economic dependence on China. While the United States can erect barriers to curtail direct imports, it cannot prevent China-linked supply chains from adapting. The report itself acknowledges that some of the shift stems from "legitimate changes" in production, investment, and sourcing—indicating that not every Chinese product reaching the US through a third country represents tariff evasion.
Research by the Peterson Institute of International Economics shows that Chinese goods and services remain deeply woven into US imports from third countries, despite years of tariffs. Tariffs have reduced direct bilateral trade but have not eliminated Chinese inputs across global supply chains. A Nikkei Asia survey similarly found that Chinese companies have expanded their global market share despite tariff barriers.
At the same time, China remains connected to American demand. Even as tariffs curtail bilateral trade, Beijing's exports of electronics, computers, and circuit-board assemblies to neighboring Asian countries have increased. Much of this production is tied to the US AI boom, allowing Chinese manufacturers to benefit indirectly from Washington's AI investments. This does not necessarily mean Chinese goods are being illegally routed into the United States; rather, it emphasizes that the underlying trade relationship remains intact.
With studies showing Chinese inputs entering the US through intermediaries like Vietnam and Mexico, the Trump administration appears to have realized that a complete commercial divorce between the world's two largest economies is unrealistic in the foreseeable future. The new approach moves away from outright decoupling toward extracting economic concessions from China—such as commitments on critical minerals supply chains, purchases of Boeing aircraft, and restored market access for US farm produce.
This does not mean Washington has abandoned economic pressure. It is pursuing selective commercial engagement while continuing to restrict access to technologies deemed critical to national security. The emerging strategy suggests a focus on increasing trade in non-sensitive sectors to retain leverage over Beijing. Meanwhile, China is doubling down on building its own capabilities, as seen in its record-setting robot sprint and its expanding visa-free travel push that reshapes regional ties. The rivalry is no longer about severing ties but about who holds the upper hand in a deeply interconnected world.
