Delayed Launch of Nvidia’s H20 AI Chip Could Shift Market Dynamics in China

In a twist of events, Nvidia, a leading manufacturer of artificial intelligence (AI) chips, has informed its Chinese client base about the delay in the release of its latest AI chip, the H20, until early next year. The information comes from insiders who requested anonymity due to the confidential nature of the information.

The H20 chip, the most robust among three China-centric chips Nvidia developed to comply with the U.S. export laws, has had its launch postponed until the first quarter of the subsequent year. Initially, Nvidia was anticipated to introduce the H20 chip and other new products as early as November 16, according to the chip industry newsletter, SemiAnalysis. Some insiders suggest the launch could possibly occur around February or March.

The delay, it appears, is triggered by certain integration-related issues that server manufacturers are encountering with the H20 chip. However, one of the other two chips under Nvidia’s new export-compliant series, the L20, is expected to launch as per its original timeline. Details about the third chip, the L2, remain unconfirmed by the sources.

These new export-compliant chips, namely the H20, L20, and L2, were part of Nvidia’s strategy to maintain its market share in China. The three new chips encapsulate most of Nvidia’s latest features for AI work but have reduced computing power to align with the U.S. export rules, as per SemiAnalysis’ analysis of these chips’ specifications. The action came in response to the company’s prohibition from shipping products like its advanced A800 and H800 AI chips due to tightening U.S. export rules.

The ripple effects of the U.S. export constraints reveal a silver lining for Nvidia’s competitors like Huawei, who have received a sizable order of their chips from the Chinese tech company Baidu.

As a result, “Nvidia shares closed down 1.9% on what was a shortened trading day in the U.S.”