President Donald Trump announced in early December 2025 that Nvidia could export its H200 artificial intelligence chips to approved customers in China under conditions preserving U.S. national security, but sales stalled nearly two months later amid U.S. and Chinese government reviews.
Trump conveyed this position in a Truth Social post directed to Chinese President Xi Jinping. The message specified permission for Nvidia to ship chips to designated buyers while ensuring strong national security protections. Following this signal, the Commerce Department’s Bureau of Industry and Security issued formal export rule changes in mid-January 2026. These alterations shifted the licensing process from a presumption of denial to evaluations conducted on a case-by-case basis, contingent on meeting specific security criteria.
The established requirements encompass multiple safeguards. Chips undergo third-party testing within the United States prior to any shipment. Exporters pay a 25 percent tariff on the value of H200 units sent to China. Companies must certify sufficient availability of chips for domestic U.S. supply chains. Recipients face prohibitions against any military applications. Shipments to China cannot exceed half the total volume of H200 chips sold to U.S. customers. Despite this regulatory framework, individual export licenses remain pending completion.
Nvidia CEO Jensen Huang addressed the status in late January 2026. He stated that the U.S. side was “nearing completion” of the licensing process, yet Nvidia awaited formal consent from Beijing before proceeding with shipments. Chinese customers have withheld H200 orders pending resolution on license approvals and attached conditions, as reported by the Financial Times. This pause highlights ongoing frictions between the United States and China regarding access to advanced semiconductor technology essential for artificial intelligence advancement.
China’s response involves internal deliberations balancing AI computing demands against support for local semiconductor producers such as Huawei. In mid-January 2026, customs officials in Hong Kong intercepted the initial arriving H200 shipments. This action surprised Nvidia and led component suppliers to suspend production lines. Subsequently, Beijing issued conditional purchase approvals to major technology enterprises, including ByteDance, Alibaba, Tencent, and AI startup DeepSeek. These firms received authorization for more than 400,000 H200 chips in aggregate.
The attached conditions to these approvals undergo further examination by China’s National Development and Reform Commission. Nvidia had projected demand exceeding 1 million units from Chinese buyers. Suppliers operated around the clock to ready shipments targeted for March 2026. The H200 chip provides roughly six times the performance of the H20 variant, which faces existing restrictions for the Chinese market.
Huang projected that Nvidia’s annual sales to China could reach $50 billion once stabilized. Amid the uncertainty, Nvidia implemented new terms for Chinese transactions, mandating full upfront payments without options for refunds or cancellations. This adjustment transfers financial exposure to the purchasers during the regulatory delays. Previously, Nvidia recorded a $5.5 billion inventory write-down after the Trump administration imposed a sudden ban on H20 chip sales to China.
Approval authority rests with both the U.S. and Chinese governments over these transactions. Neither side has granted full sanction to the arrangements, sustaining uncertainty for Nvidia and its prospective Chinese clients.
