Chinese Tech Giants Scramble for Huawei Ascend Chips as DeepSeek V4 Triggers Supply Crunch

The launch of DeepSeek V4 has triggered an unprecedented scramble for domestic artificial intelligence hardware in China. Reuters reports that major technology conglomerates are rushing to secure Huawei Ascend 950 processors. The sudden surge in demand underscores a pivotal shift in China’s semiconductor ecosystem, as domestic alternatives are beginning to demonstrate viability at scale amid tightening United States export controls.

According to multiple sources familiar with the matter, internet giants including ByteDance, Tencent, and Alibaba have approached Huawei in recent days to place substantial new orders for the Ascend 950 series. The urgency stems directly from the successful deployment of DeepSeek V4, which was optimized specifically to run on Huawei’s infrastructure. This development has provided the clearest evidence yet that Chinese foundational models can achieve state-of-the-art performance without relying exclusively on Nvidia hardware.

The DeepSeek Catalyst

The catalyst for this hardware rush was the release of DeepSeek V4, a highly capable open-source model that has disrupted the global AI landscape. When DeepSeek announced that its V4 model, including the massive 1.6 trillion parameter V4-Pro version, was fully supported by Huawei’s Ascend 950-based infrastructure, it sent a powerful signal to the broader industry.

For months, Chinese technology companies have been exploring alternatives to Nvidia, driven by the necessity of US export restrictions that have cut off access to top-tier chips like the H100 and H200. However, widespread adoption of domestic alternatives was hindered by concerns over software ecosystem maturity and real-world performance at scale. DeepSeek’s successful integration with Huawei hardware has effectively validated the Ascend platform for large-scale model training and inference.

Following the launch, major cloud providers, including Alibaba Cloud and Tencent Cloud, immediately rolled out V4 services, expanding access to developers and enterprise customers. This rapid deployment has sharply boosted the demand for AI processing capabilities, forcing these companies to rapidly expand their hardware infrastructure. The scramble is not limited to the largest players; companies specializing in cloud computing and graphics processing unit (GPU) rental services are also aggressively pursuing Ascend 950 allocations to meet the anticipated surge in developer demand.

Performance and the Nvidia Gap

The Huawei Ascend 950PR, the specific variant driving the current demand spike, represents a significant breakthrough for China’s domestic semiconductor industry. Industry benchmarks and testing indicate that the 950PR outperforms the Nvidia H20—the downgraded chip specifically designed by Nvidia to comply with US export controls for the Chinese market.

However, the Ascend 950PR still lags behind Nvidia’s more advanced H200 processor. The critical factor is that shipments of the H200 to China remain completely stalled amid ongoing regulatory disputes and export restrictions. This dynamic has created a crucial opening for Huawei. With the H200 unavailable and the H20 offering compromised performance, the Ascend 950PR has emerged as the most compelling option for Chinese companies requiring high-performance AI compute.

Huawei has capitalized on this opening, positioning the 950PR as the only domestic chip capable of enabling highly efficient, lower-cost AI computations at the scale required by frontier models. The successful testing and deployment by firms like ByteDance and Alibaba represent a hard-won victory for Huawei after years of limited traction in the data center GPU market, which has historically been dominated by Nvidia’s CUDA software ecosystem.

Supply Chain Constraints and Future Outlook

While demand for the Ascend 950 series is surging, significant supply constraints threaten to bottleneck the transition to domestic hardware. Huawei aims to ship approximately 750,000 units of the 950PR this year, with mass production having commenced in April 2026. Full-scale deliveries are planned for the second half of the year.

However, the supply of the Ascend 950PR is widely expected to lag behind the rapidly accelerating demand. This shortfall is primarily due to the ongoing impact of US export curbs on advanced chipmaking equipment, which severely limit China’s access to the cutting-edge manufacturing tools required to produce these complex processors at high yields. The inability to freely procure advanced lithography machines and other critical semiconductor manufacturing equipment remains the most significant hurdle for Huawei and its foundry partners.

The implications of these supply constraints are already rippling through the AI ecosystem. DeepSeek, which is currently offering developers a 75 percent discount on its new model, has indicated that pricing for the V4-Pro could decline significantly in the second half of 2026, but this is contingent on Huawei’s Ascend 950 supernodes shipping “at scale.” The company has acknowledged that constraints will persist until production ramps up, reflecting the tight supply of high-end homemade AI chips.

This dynamic creates a complex landscape for China’s AI ambitions. On one hand, the successful pairing of DeepSeek V4 and Huawei hardware demonstrates remarkable progress toward technological self-reliance, aligning perfectly with Beijing’s strategic objectives. On the other hand, the physical limitations of semiconductor manufacturing under a strict sanctions regime mean that scaling this success will be a formidable challenge. As the scramble for chips intensifies, Huawei’s ability to navigate these manufacturing bottlenecks will be a decisive factor in the trajectory of China’s artificial intelligence sector over the coming years.