According to three people briefed on the matter, Chinese customs authorities issued instructions this week instructing agents to bar Nvidia’s H200 artificial intelligence chips from entry into mainland China. The directive comes despite Chinese firms’ robust demand for high-performance AI accelerators. Sources say the ban applies broadly across all ports of entry and follows heightened scrutiny of advanced semiconductors previously exported under relaxed Trump administration rules. Analysts estimate that Chinese data-center operators had penciled in over 20,000 units of H200 systems for deployment in the first half of 2026, representing a potential revenue shortfall in the low hundreds of millions of dollars if the prohibition remains in place.

On the American side, the Commerce Department has approved Nvidia’s export of H200 chips to China but subject to a new set of certification and security requirements. All shipments must pass third-party functional and security testing, with detailed documentation on end-user compliance and upstream payment channels. The policy mandates that Nvidia’s customers in China undergo background vetting and commit to periodic audits. These measures are designed to address national security concerns raised by lawmakers, even as major Chinese cloud providers have reportedly placed orders for tens of thousands of H200 modules to support AI training workloads.

Following publication of the customs directive, Nvidia’s trading volume dipped by roughly 16.5% below its three-month daily average, reflecting investor caution over near-term Chinese sales. Despite this volatility, Nvidia’s most recent annual revenue exceeded $130 billion, with gross margins consistently above 70%. Industry strategists contend that while the Chinese ban may delay deployments, Nvidia’s broader AI roadmap—including partnerships in telecommunications and autonomous systems—offers diversified growth avenues. Long-term investors will be watching whether the dual regulatory environment constrains China’s AI ambitions or simply postpones the adoption curve for Nvidia’s most advanced accelerators.