Huawei Technologies is set to begin mass shipments of its advanced Ascend 910C and upcoming Ascend 920 artificial intelligence chips, marking a significant milestone for China´s self-reliance in the face of mounting US export restrictions. The new Ascend 910C, constructed by combining two previous-generation chips into one module, is designed to closely compete with Nvidia´s flagship H100 accelerator, which has been banned from the Chinese market since 2022. Supporting mainstream artificial intelligence frameworks alongside Huawei´s MindSpore, these chips offer Chinese companies vital alternatives to Western hardware and ensure continuity for large model training and analytics within China´s technology ecosystem.
This rapid development by Huawei exemplifies China’s broader push toward semiconductor autonomy, a move intensified by Washington’s bans on high-end chips from Nvidia and AMD as well as stricter curbs on supporting technologies. Responding to these constraints, Beijing has heavily invested in its semiconductor sector, with Huawei leading China’s efforts to sidestep the US technology blockade. While the Ascend 910C leverages domestic foundries like SMIC and at times circumvents sourcing limitations through intermediaries, the ingenuity and resourcefulness displayed signal that sanctions, rather than stalling mainland advancement, are catalyzing significant innovation and competitive momentum within China’s artificial intelligence hardware industry. Other domestic players, including Baidu, Alibaba, and Biren Technology, are also rushing to fill the market vacuum, further accelerating the decoupling from Western chip supplies.
The emergence of Huawei’s Ascend series challenges Nvidia´s longstanding market dominance and introduces a credible competitor in one of the world’s largest tech markets. As Chinese companies and data centers pivot to homegrown solutions, Nvidia faces not only immediate financial setbacks, such as inventory write-offs and revenue losses, but also the longer-term threat of eroded influence in China. This divide is emblematic of a growing bifurcation in the global artificial intelligence ecosystem: the US and its allies pressing forward with Western-developed chips, while China forges an increasingly independent path with its own silicon, software, and standards. This trend could reshape international technology flows, create new barriers for collaboration, and spur a more distributed global innovation environment as both spheres race to advance artificial intelligence infrastructure and capabilities independently.
