As global tensions reshape tech supply chains, Huawei is stepping up with a new AI processor designed to challenge Nvidia’s dominance. The Chinese tech giant hopes its latest chip, part of the Ascend series, can match the performance of Nvidia’s H100 graphics processing unit (GPU), according to Reuters.
This bold move comes as the U.S. tightens export controls, restricting China’s access to high-performance chips required for training advanced AI models. These controls have hit Chinese companies hard, blocking access to industry-standard processors from Nvidia and AMD. In response, Huawei is accelerating its efforts to achieve chip independence.
Huawei Ascend Series Targets China’s AI Future
Huawei’s latest chip, reportedly part of the Ascend 910D series, is a key step in a multi-generation strategy. The company is also planning mass shipments of its 910C AI processor by next month. Both chips are designed to support model training and inference for cloud and data center environments, and they can be integrated into scalable AI clusters and supercomputers.
This hardware push supports China’s broader goal of building a self-reliant technology sector. For years, Huawei and other domestic chipmakers have trailed behind U.S. rivals in AI processing power. Nvidia’s H100 and upcoming B200 chips have remained out of reach due to U.S. sanctions. These GPUs are critical for training large language models (LLMs) and running complex AI workloads.
The U.S. first blocked sales of Nvidia’s H100 chips to China in 2022, even before they officially launched. That restriction has since expanded to include newer models, including the B200, as Washington seeks to limit China’s military and AI capabilities.
Huawei’s solution? Develop chips at home that can stand toe-to-toe with the world’s best — even if that means taking a few years to catch up.
Software Ecosystems Create a New Hurdle
While Huawei’s chips may close the gap in hardware, software remains a significant challenge. Nvidia’s CUDA platform has become the global standard for AI model training. It provides developers with powerful tools and deep integration across major AI frameworks.
In contrast, Huawei uses its own Compute Architecture for Neural Networks (CANN). Although promising, CANN lacks the ecosystem maturity and global developer support that CUDA enjoys. This software gap could make it harder for Huawei to compete outside of China.
Yet, Huawei is undeterred. Since being placed on the U.S. Entity List in 2019 — which blocked access to key American technologies — the company has reinvented itself. It shifted focus from consumer smartphones to enterprise tech, cloud services, and now AI chips.
The Ascend 910D chip is a vital piece of this new puzzle. If successful, it could not only strengthen Huawei’s position in AI computing but also serve as a model for China’s ambitions to achieve semiconductor independence.
The Chinese government is backing this transformation with billions in R&D funding. AI and chip self-sufficiency are now seen as matters of national interest. But building a global alternative to Nvidia will require more than just money — it will take time, trust, and technical excellence.
Ren Zhengfei, Huawei’s founder and CEO, has stressed that innovation is key to survival. And with geopolitical pressures showing no signs of easing, Huawei appears committed to building a new tech stack from the ground up.
For now, Huawei’s AI processors may find more traction within China, where domestic companies face the same export challenges. As more firms adopt CANN and support grows, Huawei could eventually emerge as a serious global competitor in AI infrastructure.
But until the software ecosystem reaches maturity and developer adoption increases, Nvidia’s dominance remains firm — especially in Western markets.
Huawei’s story is still unfolding, but one thing is clear: the battle for AI hardware leadership is no longer just about chips — it’s about ecosystems, sovereignty, and long-term survival.
