Chinese Tech Firms Rush for Huawei AI Chips After DeepSeek V4 Launch
- tech360.tv
- 48 minutes ago
- 3 min read
Demand for Huawei’s Ascend 950 artificial intelligence chips has surged following the release of DeepSeek’s V4 AI model, which runs on the Shenzhen-based tech firm’s semiconductors. Major Chinese internet companies are actively seeking to secure new orders for these chips.
China’s largest internet firms, including ByteDance, Tencent, and Alibaba, are contacting Huawei about additional chip orders. Companies specialising in cloud computing and graphics processing unit rental services are also placing orders for the chips.
The 950PR chip significantly outperforms Nvidia’s H20 chip, which was the most powerful chip Nvidia was permitted to sell in China until its import was blocked. However, the 950PR still trails the American firm’s H200, a more advanced processor that is currently subject to regulatory uncertainty.
Despite obtaining export approvals from both the U.S. and China, the H200 has yet to be shipped to China. This delay is due to ongoing disagreements between Beijing and Washington regarding the conditions governing its sale, providing an opportunity for Huawei to sell its semiconductors.
The 950PR represents a significant breakthrough for Huawei after years of struggling to win large orders from China’s technology sector. Customer testing of the chip was successful, with firms including ByteDance and Alibaba planning orders after samples were distributed.
The heightened demand for Huawei’s chips underscores how the DeepSeek V4 release has boosted the need for domestic Chinese AI hardware. U.S. export controls continue to restrict access to Nvidia’s most advanced processors, further affirming the performance of Huawei’s chips.
DeepSeek’s decision to optimise its V4 specifically for Huawei’s chips marks a strategic shift away from American semiconductor dependence. This move aligns with Beijing’s priority of promoting China’s homegrown AI technology in pursuit of technological supremacy.
Huawei announced that its Ascend supernode infrastructure, built on the Ascend 950 series chips, fully supports DeepSeek V4 models. The entire Ascend SuperNode product line has been adapted for V4 inference, the process of using a trained AI model to answer queries and execute tasks.
Among Chinese chipmakers, Huawei’s Ascend 950 series, specifically the 950PR variant, is the only domestic chip to support a technique that processes AI calculations in a compressed numerical format. This allows it to handle more computations per second at a lower cost.
Alibaba Cloud’s Bailian platform made DeepSeek V4 available on the same day it was released, offering both the V4-Pro and V4-Flash variants at prices matching DeepSeek’s official rates. Tencent Cloud launched V4 preview services on its TokenHub platform concurrently.
Tencent deployed the model on both domestic nodes and its Singapore international gateway to serve global users. The rapid deployment by major cloud platforms means millions of users and developers can now access V4.
This access sharply increases the volume of AI queries that need to be processed, and with it, the demand for the underlying chips. DeepSeek is offering developers a 75% discount on its new model until May 5.
DeepSeek stated that V4-Pro pricing could decline significantly in the second half of 2026 once Huawei’s Ascend 950 supernodes ship at scale. However, the company acknowledged that supply constraints would persist until production increases.
These constraints reflect the tight supply of high-end homemade AI chips. Output of the 950 is expected to fall short of demand due to U.S. export restrictions on advanced chipmaking tools.
These restrictions prevent China from acquiring cutting-edge manufacturing equipment. Huawei planned to ship approximately 750,000 units of the 950PR this year, with mass production beginning in April.
Full-scale shipments are anticipated to start in the second half of 2026. DeepSeek’s V4 includes two versions: V4-Pro with 1.6 trillion total parameters and V4-Flash with 284 billion parameters. Both models support a one-million-token context window.
The models are available as open-source releases under the permissive MIT open-source licence. This licence allows companies to freely use, modify, and commercialise the models.
DeepSeek’s V4 AI model release has led to a significant increase in demand for Huawei’s Ascend 950 AI chips among major Chinese internet and cloud computing firms.
Huawei’s 950PR chip, a variant of the Ascend 950 series, outperforms Nvidia’s H20 and is the only domestic chip supporting compressed numerical format for AI calculations.
U.S. export controls on advanced Nvidia chips and chipmaking tools contribute to the demand for Huawei’s products and are expected to cause supply constraints for the 950.
Source: REUTERS
