Huawei is working with DeepSeek on AI chip software, TahawulTech.com reports. The collaboration targets the software layer that surrounds Huawei's processors — the toolchain developers use to run and optimize AI workloads on the company's silicon.

For Huawei, the partnership addresses a persistent gap. The company builds AI accelerators such as the Ascend series, but the surrounding software ecosystem has lagged behind the hardware. Developers outside China's major cloud providers have often found the toolchain harder to use than the CUDA environment that accompanies Nvidia GPUs. Teaming up with DeepSeek gives Huawei direct access to one of the most widely used AI model developers in China, and a concrete testbed for improving compatibility and performance.

DeepSeek brings credibility to the effort. The company's open-weight language models have gained significant traction since late 2024, and its inference-heavy workloads represent exactly the type of demand Huawei wants its accelerators to serve. Running DeepSeek models efficiently on Ascend hardware would demonstrate that domestic Chinese chips can support frontier-class AI systems without relying on US-origin GPUs.

The partnership also fits the broader strategic picture. US export controls have cut off Chinese firms' access to advanced Nvidia hardware and, increasingly, to the software ecosystems built around it. Chinese AI developers must now run their models on domestic chips, and domestic chips need mature software to be usable. A Huawei–DeepSeek alignment connects the two ends of that chain.

Details of the agreement remain limited. TahawulTech's report does not specify whether the collaboration involves engineering teams working directly on Huawei's CANN compute platform, joint optimization of specific DeepSeek models, or a broader co-development arrangement. Huawei has not published technical documentation, and neither company has announced a formal launch date or product milestone.

What the move signals, though, is clear. Huawei is trying to close the software gap that has held back adoption of its AI accelerators, and it is doing so by partnering with a model developer whose workloads carry real weight in the market. DeepSeek, in turn, gains a hardware partner invested in making its models run well on Chinese silicon.

For enterprise buyers in the Middle East and other regions where Huawei sells infrastructure, the development is worth tracking. Improved AI software support on Huawei platforms would expand the range of vendors capable of serving large-model workloads, and would strengthen Huawei's position in markets where US export rules already complicate procurement of American AI hardware.

The collaboration is the latest in a series of moves by Chinese technology firms to build a self-sufficient AI stack — chips, software, and models — under export-control pressure. Huawei and DeepSeek now form a key link in that effort.