Huawei Challenges Nvidia with Open Source AI
Beijing, Wednesday, 2 July 2025.
Huawei is intensifying its battle with Nvidia in China. The company is open-sourcing its AI models to boost the adoption of its Ascend AI chips. This strategy directly challenges Nvidia’s market dominance. Huawei’s approach mirrors Google’s chip-to-model strategy. Huawei focuses on specialized AI applications in government, finance, and manufacturing. Unlike Baidu, it tailors its AI for specific sectors. This move could reshape China’s AI landscape.
Huawei’s open source strategy
On June 30, 2025, Huawei announced it would open source two of its Pangu series AI models, along with its model reasoning technology [1]. This move aims to encourage the use of Huawei products, especially its Ascend AI chips, which compete directly with Nvidia [1]. Lian Jye Su, chief analyst at Omdia, believes this strategy will incentivize the adoption of other Huawei offerings [1]. Huawei’s approach involves specialized AI models for sectors like government, finance, and manufacturing [1]. This is unlike competitors such as Baidu, which have a broader focus [1].
Competitive dynamics and market impact
Huawei’s open-source strategy is designed to drive hardware sales, a different model from its competitors [1]. Marc Einstein, research director at Counterpoint Research, notes that Huawei collaborates with companies like DeepSeek and Baidu [1]. Huawei’s Pangu Pro MoE model, featuring 72 billion parameters, ranked fifth in open-source rankings, according to SuperCLUE data from May 2025 [3]. This positions it behind models from DeepSeek and Qwen [3]. Huawei’s integrated system, from chips to models, presents a high-performance alternative to Nvidia’s dominance [3].
Nvidia’s market position
Huawei’s increasing AI capabilities pose a challenge to Nvidia’s market share in China [1]. Paul Triolo, partner at DGA-Albright Stonebridge Group, describes Huawei as a “muscular technology juggernaut” across the AI hardware and software stack [1]. Huawei’s open-source strategy is expected to resonate in developing countries where enterprises are more price-sensitive [1]. The company’s focus on integrating its Ascend chips with its Pangu models creates a cohesive ecosystem [7]. This strategy mirrors Nvidia’s CUDA ecosystem, potentially attracting developers and enterprises [7].
Potential revenue and future outlook
By open-sourcing its AI models and inference technology, Huawei aims to embed its Ascend hardware into the AI industry [7]. Developers using Pangu models will likely rely on Ascend chips [7]. Enterprises implementing these models will need to procure Ascend servers or cloud services [7]. This approach could drive revenue for Huawei’s hardware and cloud divisions [7]. Huawei’s strategy includes open-sourcing model reasoning technology to facilitate better integration and adoption of its AI models [7]. This is projected to expand its reach in the AI market [1].
Bronnen
- www.cnbc.com
- q.stock.sohu.com
- wallstreetcn.com
- www.xhby.net
- www.bilibili.com
- cd.nbd.com.cn
- finance.sina.cn
- www.tmtpost.com