DeepSeek and Huawei have launched an open-source programming toolkit aimed at Huawei’s Ascend AI chips, offering developers an alternative software stack to Nvidia’s CUDA. The toolkit includes TileLang plus compute and communication libraries, according to Mobile World Live’s 30 September report.
DeepSeek Watch analysis
What happened
Mobile World Live says DeepSeek described TileLang as simpler to programme, with the aim of improving development efficiency and simplifying code. DeepSeek also credited Huawei with supporting research and development for its Ascend 950 chips. The article reports that the toolkit is free to download and open source.
That is a concrete attempt to make Huawei’s AI hardware easier to program, not evidence that developers have already moved away from CUDA or that the software matches it in performance. The report attributes the toolkit details to DeepSeek’s statements on WeChat.
Why it matters
AI chips need software developers can use, not just silicon with impressive specifications. A programming toolkit, alongside compute and communication libraries, is part of the infrastructure needed to build and run AI models on a platform. If DeepSeek and Huawei can make Ascend easier to work with, they could give developers another option in a field where Nvidia’s CUDA is widely used.
The practical test will be adoption and results: the announcement does not establish how well the toolkit performs, how complete it is, or how many developers will choose it. Software ecosystems are built one working project at a time, which is less dramatic than a chip launch and often more consequential.
Our read
This is a meaningful move in the contest to make AI computing less dependent on Nvidia’s software stack. Free and open-source access lowers the barrier to trying the tools; it does not, by itself, make switching painless. Developers will want working documentation, reliable libraries and performance they can reproduce before they rewrite anything important.
What to watch
- Whether DeepSeek or Huawei publishes benchmarks and practical examples using the toolkit.
- How the tools support Ascend 950 and other Ascend chips.
- Whether external developers adopt TileLang and contribute to the project.
Discussion spark: What would persuade AI developers to choose an open-source Ascend toolkit over the established CUDA ecosystem: better performance, easier programming, or a stronger library and support base?
Sources and evidence
- DeepSeek, Huawei team on AI chip software – mobileworldlive.com (30 September 2026, 14:44 UTC)
not affiliated with or endorsed by DeepSeek