Nvidia Corp (NVDA.O), the AI chip market's reigning champion, has modified its flagship product to meet export regulations for China. US regulators had barred Nvidia from selling its advanced A100 and H100 chips to Chinese customers last year, citing national security concerns. These chips are crucial to the development of generative AI technologies, including OpenAI's ChatGPT.
But Nvidia wouldn't let export regulations slow them down. In November, the company announced a China-compliant chip, the A800, that reduced some of the A100's capabilities. And now, Nvidia has developed the H800, a China-export version of its H100 chip that is already being used by Alibaba, Baidu, and Tencent.
Nvidia CEO Jensen Huang spoke excitedly about the new capabilities the H800 would offer to "young startups" who are taking the generative AI revolution by storm. With Nvidia's AI available through the cloud capabilities of the big players in Chinese tech, language models and AI training will be more accessible than ever.
US regulators are focused on slowing China's progress in the technology sector, and rules around AI chips include bans on chips with powerful computing capabilities and high chip-to-chip data transfer rates. A source in China's chip industry told Reuters that the H800 reduces the chip-to-chip data transfer rate to about half of the flagship H100.
Nvidia declined to provide further details on the differences between the H800 and H100, but confirmed that its 800 series products comply with export control regulations. It's clear that Nvidia won't let regulations slow them down, and they're still leading the way in AI chips.