When Xi Comes to Washington

Council on Foreign Relations | Rush Doshi

Chinese President Xi Jinping arrives in Washington on September 23, 2026, for a high-stakes state visit with U.S. President Donald Trump amid asymmetrical economic leverage and lingering structural trade tensions. Washington offered unprecedented ceremonial honors—including fighter jet flyovers and tarmac greetings—despite Beijing throttling rare earth exports and expanding economic pressure.

Beijing previously leveraged supply chain controls to roll back U.S. tariffs in the October 2025 "Busan deal," cementing China's status as an equal peer competitor. Washington holds unexecuted retaliatory measures, such as the September 18 Sanctioning Russia and Iran Act authorizing 100 percent tariffs on importers of Russian energy. However, the Trump administration has repeatedly prioritized short-term stability by postponing Section 301 tariff findings and delaying Taiwan arms packages. Strategic competition remains frozen. The bilateral summit is expected to yield minimal substantive deliverables beyond a temporary extension of the November 10 trade truce, limited agricultural purchase commitments, and a preliminary artificial intelligence safety dialogue.

Comment

The licensing of Nvidia H200 microprocessors to selected Chinese entities illustrates the friction between targeted technological containment and state-directed market substitution. Washington’s export control regime aims to restrict frontier artificial intelligence training, yet Beijing’s domestic steering toward Huawei hardware demonstrates how external chokepoints accelerate import-substitution ecosystems. Consequently, unilateral export restrictions yield diminishing leverage as Chinese technology firms establish parallel silicon supply chains.

This structural decoupling is further compounded by industrial-scale model distillation techniques that bypass compute bottlenecks entirely. By extracting structural parameters from American frontier models into localised systems, Chinese developers mitigate hardware deficits without relying on advanced foreign silicon. The strategic bottleneck has thus shifted from Nvidia hardware acquisition to counter-distillation defenses across American frontier laboratories.

Strategic Question for Discussion

If industrial-scale model distillation allows Chinese developers to achieve competitive capabilities without top-tier silicon, which factor will dictate long-term AI dominance—unilateral hardware export controls on Nvidia H200 processors or proprietary algorithmic defense mechanisms within American frontier laboratories?

The pattern suggests that hardware export controls face diminishing returns as model distillation techniques effectively uncouple algorithmic performance from silicon availability. My assessment is that proprietary counter-distillation defenses and data-sanitisation protocols will become the primary determinant of technical edge. Consequently, controlling software architecture access will prove far more decisive than attempting to restrict semiconductor hardware distribution across global supply chains.

Share your assessment in the comments below.

To read in details click here