China reportedly allows ByteDance and Tencent to import 10,000 H200 chips

China has reportedly authorized ByteDance and Tencent to each import 10,000 Nvidia H200 AI chips, signaling a potential easing of access to advanced US-made silicon despite ongoing export controls. The Financial Times reports that other Chinese firms could receive similar permissions, raising questions about enforcement of American trade restrictions.

According to the Financial Times, Chinese authorities have granted ByteDance and Tencent permission to import 10,000 Nvidia H200 graphics processing units each — a significant development given Washington's sweeping export controls designed to limit China's access to cutting-edge AI hardware. The H200 is one of Nvidia's most powerful chips for AI workloads, making it a closely watched commodity in the global semiconductor race.

The reported approvals suggest Beijing may be finding ways to channel advanced chips to its biggest tech players despite American restrictions. Analysts will be watching closely to see whether additional Chinese companies receive similar quotas, and how the US government responds — particularly as tensions over AI supremacy between the two nations continue to intensify.

China has reportedly given the green light for ByteDance, the parent company of TikTok, and gaming-and-tech giant Tencent to each import 10,000 units of Nvidia's H200 chip, one of the most capable AI accelerators currently available. The information comes from the Financial Times, which also noted that other major Chinese technology firms could be in line for comparable approvals.

The H200 sits near the top of Nvidia's data center lineup, offering substantial improvements in memory bandwidth and AI inference performance over its predecessor. Access to chips at this tier is precisely what US export controls have aimed to restrict, with the Biden and Trump administrations both citing national security concerns about China potentially using such hardware to accelerate military AI development or close the gap with American AI capabilities.

This development raises immediate questions about how these chips are being imported — whether through direct licensing exceptions, third-party intermediaries, or some form of bilateral arrangement not yet publicly disclosed. The scale of 10,000 units per company is not trivial; it represents meaningful compute capacity that could fuel large-scale model training or inference infrastructure.

Why it matters: If China is successfully routing high-end Nvidia silicon to its top tech companies despite export controls, it would represent a significant policy challenge for Washington. Export restrictions are only as effective as their enforcement, and any systematic workaround could erode the strategic advantage the US has sought to maintain in AI hardware. This could prompt a fresh round of regulatory tightening or diplomatic friction between the two countries.

The broader takeaway is that the global semiconductor chessboard remains highly dynamic. Companies, governments, and investors will be watching to see whether these imports are confirmed officially, how Nvidia responds publicly, and whether US regulators move to close whatever gap may have allowed this level of access to occur.