Sources: DeepSeek CEO Liang Wenfeng says training on Huawei chips is one of DeepSeek's biggest bets and Huawei is set to deliver training chips in Q4 or Q1 2027

Access to domestic AI training chips speeds up model development for Chinese companies.

DeepSeek CEO Liang Wenfeng has stated that a significant strategic focus for the company is the increased utilization of domestic chips for model training. This approach represents a key bet for DeepSeek. The company is reportedly anticipating the delivery of training chips from Huawei in either the fourth quarter of 2026 or the first quarter of 2027. This initiative underscores a broader trend among Chinese AI companies to reduce reliance on foreign hardware and bolster their domestic supply chains. The collaboration with Huawei signifies a potentially substantial shift in the resources available for large-scale AI model development within China.

DeepSeek's strategic pivot towards Huawei chips signals a significant effort by Chinese AI firms to build out independent, robust domestic hardware ecosystems. This move is crucial for China's ambitions to achieve AI self-sufficiency, potentially accelerating the pace of AI development and deployment within the country by mitigating supply chain risks and geopolitical dependencies. The anticipated delivery of Huawei's training chips in late 2026 or early 2027 could unlock new capabilities for Chinese AI companies.

This reliance on domestic hardware, particularly from a major player like Huawei, positions Chinese AI developers to potentially gain a competitive edge through tailored chip solutions optimized for their specific model architectures. The success of this strategy will be closely watched as it could influence global AI hardware manufacturing and research dynamics, fostering a more bifurcated AI development landscape.