Huawei is poised to make waves in the AI market with its cutting-edge Ascend 910C AI chips, a development that could significantly challenge the dominance of NVIDIA. Reports suggest that Huawei’s new technology will integrate into the robust CloudMatrix 384 AI cluster, showcasing superior performance compared to NVIDIA’s GB200, albeit with a higher power consumption.
The upcoming Ascend 910C chips are part of Huawei’s strategic move to establish a strong foothold in AI-driven solutions, particularly with its groundbreaking rack-scale architecture. The CloudMatrix 384 (CM384) AI cluster, which will feature these chips, is designed to deliver 300 PetaFLOPS of BF16 computing power. This represents nearly double the capability of NVIDIA’s GB200 NVL72, bolstered by increased HBM capacity for superior computing outcomes.
The CM384 AI cluster’s architecture employs an “all-to-all topology,” connecting 384 Ascend 910C chips. Despite consuming nearly four times the power of its NVIDIA counterpart, China is well-equipped to support such high-energy demands, mitigating concerns over perf/watt figures which might weigh heavily elsewhere.
For Huawei, the focus appears to be on expanding domestic technological infrastructure and diminishing reliance on foreign technology alliances. The CM384 represents more than just computational prowess; it signals a robust domestic solution that aligns with national priorities.
As global supply chain dynamics shift, Huawei’s production of these AI clusters depends on a broad network of partners including TSMC and Samsung. With changing geopolitical landscapes, including U.S. restrictions affecting market operations, companies like NVIDIA face increasing pressure. In this competitive environment, Huawei’s advances push back against these challenges, targeting China’s sizable tech market with promising new solutions.
In essence, Huawei’s Ascend 910C chips spotlight China’s advancements in AI, as the nation aims to lead innovation, independently of Western technology dependencies. As Huawei narrows the technological gap with NVIDIA, the global AI industry could witness significant shifts in leadership and market dynamics.
