Nvidia Chips Drive Chinese AI Growth Despite Export Controls

Nvidia (NASDAQ:NVDA) highlighted the rising demand for its chips in China after DeepSeek, a leading Chinese AI firm, showcased its advancements using Nvidia's H800 GPUs. The statement followed Nvidia's shares dropping 17% to $118.58 on concerns that DeepSeek achieved significant progress using fewer Nvidia chips compared to competitors like OpenAI. Shares of rival Advanced Micro Devices (NASDAQ:AMD) also fell, sliding over 6% to $115.01.

DeepSeek utilized approximately 2,000 Nvidia H800 chips, compliant with U.S. export regulations introduced in 2022. These restrictions aimed to curb China's development of supercomputers for AI and nuclear capabilities but have not significantly slowed Chinese AI advancements. According to Jimmy Goodrich, a technology expert at RAND Corp, China's AI firms, including DeepSeek, have long been building their expertise with existing Nvidia-powered supercomputers.

Nvidia emphasized that DeepSeek's demand for inference processing underscores the importance of its GPUs and networking solutions. Inference, the process of serving AI-generated responses to users, requires substantial computational power. Nvidia's H20 chip, designed for compliance with export rules, remains a critical tool for this process. Goodrich noted that the H20 is likely the best chip globally for inference, despite its limited AI training capabilities under export restrictions.

DeepSeek's ongoing progress demonstrates the resilience of Chinese AI innovation, even under U.S. export controls. Nvidia, however, sees a bright future for its chips in meeting the growing demands of Chinese firms, highlighting their reliance on advanced GPU technology for both AI training and inference workloads. As demand for AI solutions surges, Nvidia's products are expected to remain integral to global AI development.