Huawei is testing its latest AI chip, the Ascend 910C, with potential clients in China. This chip aims to provide a strong alternative to Nvidia’s high-end GPUs, especially after US restrictions limited Nvidia’s sales to China. Samples of the Ascend 910C have been offered to large Chinese server companies for testing and hardware configuration.

The Ascend 910C is an upgraded version of Huawei’s Ascend 910B chip, which has already been used in China as an alternative to Nvidia’s A100 chip in various industries, including AI training.

What is the impact of US Sanctions on Nvidia?

Since August 2022, US sanctions have prohibited Nvidia from selling its A100 and H100 GPUs to China. In response, Nvidia developed modified versions, such as the A800 and H800, but these too faced further export restrictions in 2023. Despite these limitations, Nvidia remains a major player in China’s AI market, with new offerings like the H20, L20, and L2 GPUs.

Nvidia’s H20 chip is expected to generate significant revenue in China, with projected sales of US$12 billion in 2024, despite earlier low demand.

Huawei’s Growing Opportunity in China

The US sanctions on Nvidia have created an opportunity for Huawei to expand its AI infrastructure and computing resources within China. Huawei’s rotating chairman, Eric Xu Zhijun, emphasized that the company has built two computing businesses over the last five years to support the domestic AI industry. This has positioned Huawei as a strong competitor in the AI chip market.

While Huawei’s AI chips, such as the Ascend 910C, show great potential, the company faces challenges. Huawei typically bundles its AI chips with other services like network and storage solutions, which may deter some clients. Additionally, the majority of Huawei’s AI chips in use are still the older 910B models.

However, as the rivalry between Huawei and Nvidia intensifies, Huawei’s continued innovation in AI technology could position it as a key player in China’s AI chip market, particularly as it moves toward self-sufficiency in semiconductor production.

Related:

Comments

RELATED ARTICLESMORE FROM AUTHOR