
NVIDIA's China-exclusive version of the H20 AI chip will be delayed and is set to be released in the first quarter of next year.

Zhitong App learned that according to media reports citing informed sources, NVIDIA will postpone the release of its artificial intelligence (AI) chip for Chinese customers. It is reported that informed sources revealed that NVIDIA has informed its Chinese customers that due to issues encountered by server manufacturers in integrating the chip, the launch of the H20 AI chip will be postponed to the first quarter of next year.
Zhitong App learned that according to media reports citing informed sources, NVIDIA (NVDA.US) will postpone the release of its artificial intelligence (AI) chips for Chinese customers. It is reported that NVIDIA has informed its Chinese customers that the launch of the H20 AI chip will be delayed until the first quarter of next year due to integration issues faced by server manufacturers.
To comply with the new US export restrictions, NVIDIA has developed three chips for the Chinese market, namely H20, L20, and L2. H20 is the most powerful one. Informed sources revealed that H20 may be launched in February or March next year. This may further complicate the battle for market share between NVIDIA and local Chinese companies in the Chinese market.
In addition, informed sources revealed that L20 will be launched as planned, while no information is available regarding L2.
The report pointed out that the US export restrictions will create opportunities for NVIDIA's competitors. NVIDIA is betting on these chips in the hope of maintaining its market share in China.
Last year, due to the inability to provide A100 and H100 chips to the Chinese market, NVIDIA specifically developed A800 and H800 chips that comply with US regulations for the Chinese market in order to bypass US government restrictions. One year later, the US government tightened export regulations again, resulting in the ban of A800 and H800.
It is reported that the H20, L20, and L2 chips are all modified versions of the basic version of H100, and they include most of NVIDIA's latest features for AI work. However, some computing power has been reduced to comply with the new US regulations.
