Beijing/Singapore - U.S. chipmaker Nvidia is set to initiate mass production in the second quarter of 2024 for its artificial intelligence (AI) chip designed specifically for China, in adherence to U.S. export regulations. The H20 chip, the most powerful among three developed by Nvidia to comply with restrictions imposed in October, was initially slated for launch in November but faced delays due to integration issues reported by server manufacturers.
The H20 chip production, starting in Q2 2024, is anticipated to have limited initial volume, with Nvidia primarily focusing on fulfilling orders for major customers. The company, however, remains tight-lipped about specific details, stating, "Nvidia declined to comment."
Chinese companies have reportedly shown reluctance in adopting the downgraded H20, opting to explore domestic alternatives amid concerns that the U.S. might further tighten restrictions. Baidu, a leading search engine company, shifted towards Huawei Technologies for AI chips, signaling a move away from Nvidia, as reported by Reuters.
Nvidia's strategy to navigate the U.S. export restrictions involves the development of two additional chips besides the H20, namely the L20 and L2, all designed to comply with the new regulations. However, the chipmaker has not yet announced the sale of any of the three chips.
In late December, Nvidia unveiled a modified version of an advanced gaming chip, adapting it to meet the new export rules. The company's focus on these chips is pivotal for retaining its market share in China after the tightened restrictions prevented the shipment of products like the A800 and H800 AI chips.
Introduced as alternatives for Chinese customers in November 2022, the A800, H800, H20, L20, and L2 chips incorporate Nvidia's latest AI features, albeit with reduced computing power to align with the newly imposed regulations. According to an analysis by SemiAnalysis, these chips maintain the essential features for AI work despite the power cutback.
