Ryan Wild
In a significant move within the global AI and semiconductor landscape, Huawei Technologies has announced the development of its most advanced AI chip to date, the Ascend 910D.
This development positions Huawei to directly challenge Nvidia’s high-end H100 chip, especially in AI training performance. The announcement underscores China’s resilience and determination to achieve semiconductor independence, even as it faces stringent U.S. restrictions on chip-making tools and technology exports
Ascend 910D: A Leap Forward in AI Processing for Huawei
The Ascend 910D represents Huawei’s latest effort to close the performance gap with Nvidia’s leading AI chips. While previous iterations like the 910B and 910C fell short in matching Nvidia’s capabilities, the 910D is designed to offer enhanced performance in AI training tasks.
Huawei has approached domestic firms for initial testing and has refined chip integration technologies to boost performance, albeit at the cost of power efficiency.
The company plans to ship over 800,000 units of the earlier chip versions in 2025 to a mix of state and private clients, indicating a significant scale-up in production and deployment.
Huawei’s advancement comes at a time when U.S. sanctions have limited China’s access to advanced chip-making equipment and technologies. Despite these challenges, Huawei’s progress with the Ascend 910D demonstrates China’s commitment to overcoming technological barriers and reducing reliance on foreign technology.
The development also coincides with U.S. sanctions on Nvidia’s H20 chips, which have opened new opportunities for Chinese chipmakers like Huawei and Cambricon Technologies.
By innovating system-level solutions, Huawei aims to leverage its chip capabilities to maintain competitiveness in the AI sector.
To compensate for any remaining performance gaps, Huawei is focusing on building highly efficient systems like the CloudMatrix 384. This system connects multiple Ascend chips to rival Nvidia’s computing capabilities, showcasing Huawei’s strategy to enhance overall system performance rather than relying solely on individual chip advancements.
By using parallel computing and network optimisation, Huawei’s CloudMatrix 384 aims to minimize the latency between chips, improving training speeds for large-scale AI models — a crucial factor in keeping up with Nvidia’s DGX systems.
While the Ascend 910D marks a significant milestone, Huawei still faces challenges, including restricted access to advanced foundries and components due to ongoing U.S. sanctions.
Many of Huawei’s chips are manufactured using mature node processes rather than the cutting-edge 5-nanometer or 3-nanometer technology used by competitors like TSMC and Samsung.
This limitation could impact performance in the long term unless domestic foundries catch up.
Nevertheless, Huawei’s efforts are paying off. Reports indicate growing domestic interest in replacing foreign AI systems with homegrown alternatives, and the Chinese government continues to prioritize AI and semiconductor development as key strategic initiatives.
If Huawei successfully scales up production of the Ascend 910D and its integrated systems, it could redefine its position not only within China but also globally — especially among emerging markets seeking alternatives to Western technology providers.
As AI becomes increasingly central to industries ranging from healthcare to defense, control over powerful AI hardware will be a key determinant of technological leadership.
The ability to process massive datasets, train large language models, and deploy AI in real-world applications will define which nations and companies maintain competitive advantages in the next decade.
AI chips like Huawei’s Ascend 910D are not just technological products; they are strategic assets that underpin economic strength, cybersecurity, military capability, and global influence.
Huawei’s announcement this week sends a strong message: despite obstacles, China’s tech giants are moving aggressively to shape the future of AI. By investing heavily in research and development, fostering domestic supply chains, and creating integrated AI ecosystems, companies like Huawei are ensuring that China reduces its dependency on Western technology. This initiative aligns with broader national goals outlined in policies such as “Made in China 2025” and recent five-year plans that prioritize AI and semiconductor self-sufficiency.
If Huawei can continue scaling production, improving performance, and delivering real-world applications for its AI chips, it could shift the balance of power in global technology markets.
The Ascend 910D, and the systems built around it, may serve as a foundation for a new era where China becomes not just a major player, but a true leader in AI hardware and system integration — a development that will have far-reaching effects across the tech world.
Read more like this >
David jones
