President Trump signed a proclamation instituting a 25% tariff on advanced AI semiconductors, including Nvidia’s H200 series, that are produced overseas, pass through the United States, and are then exported to China. The measure formalizes the Commerce Department’s December decision to permit shipments of H200 chips to vetted Chinese customers. The tariff is applied at the border when the chips are re-exported, and it targets high-performance AI accelerators that Nvidia sells under strict licensing controls. While the surcharge raises the landed cost for Chinese buyers, Nvidia has publicly supported the policy, emphasizing that approved shipments will continue under existing export-control licenses.

Despite the extra duty, demand for H200 chips remains strong. Nvidia reported receiving over 100 early orders from major Chinese internet and cloud-service companies seeking to deploy large language models and generative AI workloads. To meet this appetite, Nvidia is considering a 30% increase in wafer starts for H200 production in its facilities in Taiwan and South Korea. Management has indicated that capacity could expand from 50,000 units per quarter to 65,000 by mid-year, pending approval of additional supply agreements with TSMC.

Beijing’s Ministry of Commerce is drafting guidelines to regulate foreign semiconductor purchases, potentially setting annual quotas for H200 imports. A leaked Nikkei Asia report suggests draft rules would allow Chinese companies to import up to 5,000 H200 units per quarter, subject to national security reviews. Any caps below current order levels could leave a backlog of 10,000 units unshippable, risking customer pushback and inventory build-up. Investors are watching for final quota figures expected in the next 60 days.

Analysts note that the tariff and quota regime may compress Nvidia’s China AI sales by up to 15% in fiscal 2026, but the company’s overall AI-data-center backlog exceeds $350 billion and remains diversified across hyperscalers in North America and Europe. Nvidia’s balance sheet, with $20 billion in cash flow generated in the last quarter, provides flexibility to absorb any near-term revenue shifts. Long-term, investors will focus on whether production scale-up and licensing efficiencies can offset the incremental duty and regulatory constraints in the world’s second-largest AI market.