DeepSeek Partners Huawei on AI Chip Tools
Chinese AI company DeepSeek has partnered with Huawei Technologies to develop programming tools optimised for Huawei’s Ascend AI chips. The move strengthens cooperation between Chinese technology companies seeking alternatives to Nvidia’s ecosystem.
DeepSeek announced the partnership on September 30, 2026. Moreover, the company said it is open-sourcing programming infrastructure for Huawei’s Ascend platform. The release includes computing and communication libraries.
The collaboration comes as China works to build a more independent AI computing ecosystem. Meanwhile, US export restrictions have limited Chinese companies’ access to Nvidia’s most advanced processors.
DeepSeek Opens Ascend Programming Tools
DeepSeek is releasing six software modules designed for Huawei’s Ascend processors. These tools build on programming infrastructure previously developed for Nvidia hardware.
One key component is TileLang, an open-source programming language for developing high-performance AI kernels. Furthermore, an Ascend-compatible version now supports Huawei’s Ascend 950 accelerators.
DeepSeek says TileLang can simplify AI chip programming while still providing access to hardware performance. The company also describes it as a simpler programming model than Nvidia’s CUDA.
As a result, developers could have a more accessible software layer for building applications on Huawei hardware. However, Nvidia’s CUDA ecosystem remains substantially more mature and widely adopted.
Huawei and DeepSeek Develop Supernode Systems
The partnership also covers a supernode system based on 128 Huawei Ascend 950 chips. DeepSeek said the system jointly optimises computing and communication performance.
Huawei recently unveiled its next-generation AI processors and supernode computing systems. At the time, the company said it expected its AI systems to see broader use for model training in 2027.
Therefore, the DeepSeek partnership could help improve the software layer surrounding Huawei’s expanding AI hardware portfolio. At the same time, better programming tools could make it easier for AI developers to adapt workloads to Ascend processors.
Building an Alternative AI Computing Ecosystem
The collaboration reflects a wider shift among Chinese technology companies toward domestic AI hardware. Export controls have increased pressure on companies to develop alternatives to Nvidia processors and software.
DeepSeek has previously explored Huawei hardware for AI workloads. In April, the company released a preview of DeepSeek-V4 adapted to run on Huawei chips.
However, reducing dependence on Nvidia involves more than replacing processors. Developers also need mature compilers, libraries, programming frameworks and communication systems.
Consequently, the latest partnership focuses on an important layer of the AI infrastructure stack. DeepSeek and Huawei are seeking to make domestic AI hardware easier to program and deploy, while building a broader software ecosystem around Ascend chips.
