DeepSeek CEO Liang Wenfeng has informed investors that the company plans to shift its artificial intelligence training strategy toward domestic hardware. This significant move marks a transition away from a heavy reliance on American-made chips like those produced by Nvidia. The company aims to integrate more Chinese technology into its core operations to ensure long-term stability and growth. This strategic pivot highlights the growing importance of self-reliance within the Chinese technology sector.

Strategic Bet on Huawei Hardware Liang Wenfeng described the decision to train models using Huawei chips as one of the company's biggest strategic bets. DeepSeek currently relies on a massive infrastructure of Nvidia chips, but future development will prioritize domestic alternatives. The CEO emphasized that using local hardware is now a major priority for the firm’s upcoming AI projects. This shift follows previous attempts to test Huawei’s Ascend hardware, which the company is now ready to scale. By moving to Huawei, DeepSeek aims to navigate the complexities of international trade and hardware availability.

Timeline for Chip Deliveries Huawei is expected to begin delivering these critical AI training chips to DeepSeek as early as the fourth quarter of 2026. Some sources suggest the delivery window could extend into the first quarter of 2027 depending on production schedules. These chips will likely be part of a massive deployment at DeepSeek's specialized data facilities. This timeline allows the company to prepare its software stack for the transition to a new architecture. The phased rollout will ensure that existing training processes are not immediately disrupted by the hardware change.

Bypassing Export Restrictions The move is largely seen as a response to tightening U.S. export controls on advanced semiconductor technology. By securing a domestic supply chain, DeepSeek can continue training powerful AI models without fearing future international trade bans. This strategy aligns with broader national goals in China to develop a self-sufficient semiconductor ecosystem. Using Huawei’s Ascend 910C or newer iterations provides a viable path for the company to maintain its competitive edge. This transition reflects a broader trend among Chinese AI labs seeking independence from foreign hardware providers.