According to foreign media reports on the 21st, the United States is considering approving the export of NVIDIA's H200 AI chip to China. The report, citing sources familiar with the matter, stated that the U.S. Department of Commerce, which oversees U.S. export controls, is reviewing changes to export restrictions on China, and that the plans may be subject to change.
As of now, the U.S. Department of Commerce has not responded, and NVIDIA has not directly commented.
It is understood that the H200 chip, released two years ago, has more high-bandwidth memory than its predecessor, the H100 chip, enabling it to process data more quickly. It is estimated that the H200 chip's performance is twice that of NVIDIA's H20 chip.
NVIDIA CEO Jensen Huang has repeatedly emphasized the importance of the Chinese market. He believes that entering the Chinese market is crucial for the United States' competitiveness in the field of artificial intelligence.
The NVIDIA H200 uses the NVIDIA Hopper architecture and is the first GPU equipped with 141 GB of HBM3e video memory, boasting read/write speeds of up to 4.8 TB/s. The H200's larger capacity and faster memory accelerates generative AI and LLM operations, while driving scientific computing for high-performance computing (HPC) workloads with higher energy efficiency and lower total cost of ownership.
