Huawei will Trial Produce AI Chips to Benchmark Nvidia

The US restrictions have unexpectedly driven independent innovation among Chinese tech giants. According to The Wall Street Journal, Huawei is preparing to trial produce the new generation AI processor Ascend 910D, aiming to challenge Nvidia’s high-end chip products. It is reported that Huawei will receive the first batch of samples of the chip as early as the end of May, and has invited multiple domestic technology companies to assist in testing its performance.

The goal of this new chip is to surpass Nvidia’s H100 GPU released three years ago. Although H200 has replaced H100 as Nvidia’s latest flagship product, H100 is still widely popular in the fields of AI computing and model training. Companies such as Google, Meta, and Tesla have purchased it in large quantities for $30000 to $40000 per chip.

The tariffs and export control policies of the US government have essentially blocked the channels for Nvidia and AMD to supply chips to China. Jack Gold, founder of J. Gold Associates, pointed out that this creates excellent opportunities for Huawei, not only to meet domestic market demand, but also to attract other international customers who are unable or unwilling to purchase Nvidia technology.

Gold told Fierce earlier this month that Huawei has successfully developed chips such as 5G, and AI GPUs are naturally the next target. He believes that “this will greatly enhance the competitiveness of companies such as Huawei in the global market and promote the sales of their independently developed chips.” If the performance of the Ascend 910D approaches that of the H100, it is expected to be widely used in complex tasks such as AI model construction and large-scale facial recognition.

DeepSeek has proven that there is a demand in the market for AI models with higher cost-effectiveness, and companies such as Alibaba, BYD, and Lenovo have begun to adopt its technology. Huawei’s AI chips are expected to replicate this successful model in the hardware field, providing new options for the global AI industry.