On April 15, Nvidia announced it would incur $5.5 billion in charges following new U.S. government restrictions on the export of its H20 AI chip to China—a major market for the company. The H20 is Nvidia’s most advanced chip available for Chinese customers and has played a vital role in its efforts to maintain a presence in China’s rapidly growing AI industry.

U.S. authorities have tightened export controls to prevent cutting-edge chip technologies from being used in Chinese supercomputers, citing national security concerns. The Commerce Department confirmed new licensing rules for the export of AI chips, including Nvidia’s H20 and AMD’s MI308. Shares of both companies dropped in after-hours trading—Nvidia by 6%, AMD by 7%.

Nvidia had designed the H20 to comply with earlier export limits, but its high-speed connectivity and memory functions still make it suitable for use in powerful computing systems. These features raised concerns it could help China build supercomputers—an area under U.S. export restrictions since 2022. Several major Chinese firms, including Tencent, Alibaba, and ByteDance, had reportedly increased their H20 chip orders to support affordable AI models, such as those developed by startup DeepSeek. However, the Institute for Progress, a Washington-based think tank, suggested that some of these firms may already be using H20 chips in ways that violate U.S. export rules.

Nvidia stated it was officially notified on April 9 that H20 exports to China would require a license, with the restriction made permanent as of April 14. The company noted that the $5.5 billion charge relates to inventory, supply contracts, and reserves connected to the H20 chip. Nvidia declined to comment further beyond its official filing.

Disclaimer: This image is taken from Reuters.