Huawei Technologies is significantly ramping up production of its most advanced artificial intelligence processor, the Ascend 950PR, aiming to ship approximately 750,000 units in 2026. A Strat News Global report, citing sources familiar with the company’s plans, indicates that mass production commenced in April, with full-scale deliveries scheduled for the second half of the year. This aggressive scaling effort comes as demand for domestic AI hardware surges, driven largely by the successful launch of the DeepSeek V4 foundational model.
The production target represents a critical milestone for Huawei and China’s broader ambitions to achieve technological self-reliance in the semiconductor sector. However, the ambitious goal is heavily constrained by ongoing United States export controls, which severely limit China’s access to the cutting-edge manufacturing equipment necessary to produce these complex chips at high yields. The resulting supply-demand imbalance is reshaping the landscape of China’s AI industry, forcing companies to adapt their strategies and infrastructure planning.
The DeepSeek V4 Demand Shock
The immediate catalyst for the surge in demand for the Ascend 950 series is the release of DeepSeek V4. The highly capable open-source model, which includes a massive 1.6 trillion-parameter V4-Pro version and a 284 billion-parameter V4-Flash version, was specifically optimized to run on Huawei’s hardware infrastructure. Both versions offer a one-million-token context window and were released under the permissive MIT open-source license, allowing for free use, modification, and commercialization.
DeepSeek’s decision to optimize its flagship model for Huawei chips signals a definitive shift away from US semiconductors toward domestic AI hardware. This alignment with Beijing’s tech ambitions has had an immediate impact on the market. Major Chinese internet firms, including ByteDance, Tencent, and Alibaba, have approached Huawei to place new chip orders, recognizing the viability of the Ascend platform for large-scale model training and inference.
Furthermore, cloud computing providers and graphics processing unit (GPU) rental services are scrambling to secure allocations of the 950PR. Major platforms like Alibaba Cloud and Tencent Cloud launched V4 services immediately after the model’s launch, expanding access to developers and sharply boosting demand for AI processing capabilities. Huawei has stated that its Ascend 950-based infrastructure fully supports V4, positioning the 950PR as the only Chinese chip currently enabling highly efficient, lower-cost AI computations at this scale.
(Related: Chinese Tech Giants Scramble for Huawei Ascend Chips as DeepSeek V4 Triggers Supply Crunch)
Performance vs. Supply Constraints
The Ascend 950PR represents a significant technological achievement for Huawei. According to industry sources, the chip outperforms the Nvidia H20, the downgraded processor specifically designed by Nvidia to comply with US export controls for the Chinese market. However, the 950PR still lags behind Nvidia’s more advanced H200 processor. Crucially, shipments of the H200 to China remain stalled amid ongoing US-China regulatory disputes, creating a vital opening for Huawei to capture market share.
Despite this opening, Huawei faces formidable challenges in meeting the surging demand. The supply of the Ascend 950PR is widely expected to lag significantly behind market needs due to the stringent US export curbs on advanced chipmaking equipment. These restrictions limit China’s access to the extreme ultraviolet (EUV) lithography machines and other critical tools required to manufacture advanced semiconductors efficiently.
The impact of these supply constraints is already evident in the pricing strategies of AI developers. DeepSeek, which is currently offering developers a 75 percent discount on its V4-Pro model until May 5, has indicated that pricing could decline significantly in the second half of 2026. However, this price reduction is entirely contingent on Huawei’s Ascend 950 supernodes shipping “at scale.” DeepSeek has acknowledged that constraints will persist until production ramps up, reflecting the tight supply of high-end homemade AI chips.
Strategic Implications for China’s AI Ecosystem
Huawei’s effort to scale production of the Ascend 950PR to 750,000 units is a central component of China’s strategy to insulate its technology sector from US sanctions. The successful deployment of DeepSeek V4 on Huawei hardware demonstrates that Chinese companies can build and operate world-class AI models without relying exclusively on Nvidia’s ecosystem.
However, the ongoing supply constraints highlight the vulnerability of China’s semiconductor supply chain. The inability to freely procure advanced manufacturing equipment means that scaling production of chips like the 950PR will remain a slow and costly process, characterized by lower yields and higher manufacturing complexities than those faced by Western competitors.
As Huawei pushes forward with its mass production plans, the ability of the company and its foundry partners to overcome these manufacturing bottlenecks will be closely watched. The success or failure of the Ascend 950PR rollout will have profound implications for the trajectory of China’s artificial intelligence sector, determining whether domestic hardware can truly support the massive compute requirements of the next generation of foundational models.
(Related: China’s Politburo Calls for Technology Self-Reliance and Full Implementation of ‘AI Plus’ Initiative)