DigestAI news desk

Cut through the AI noise.

Hardware & Computeupdated 3 min read

Deepseek releases open-source TileLang for Huawei Ascend chips

Deepseek and Huawei announced a collaboration to develop open-source software for Huawei’s Ascend AI chips, including TileLang, a programming language designed as an alternative to Nvidia’s CUDA. The release includes libraries for computation and data transfer between chips, all made available under an open-source license. Deepseek claims TileLang offers a simpler programming model than CUDA and…

4 sources

Key points

  • Deepseek and Huawei release open-source TileLang for Ascend chips, targeting Nvidia’s CUDA dominance
  • TileLang includes computation libraries and data transfer tools, tested on Nvidia hardware before AGI use
  • Huawei’s new AI processors face domestic demand shortages, reducing exports due to US restrictions

Research firm SemiAnalysis noted that CUDA’s moat is weakening, citing OpenAI’s Jalapeño chip outperforming Nvidia’s Blackwell in some scenarios. However, Nvidia still leads in agent workloads, and SemiAnalysis cautioned that its advantage lies in software integration rather than silicon alone. Earlier analysis highlighted Huawei’s CANN software stack as one of the few, alongside CUDA, to support DeepSeek V4 on day one.

Full story from The Decoder · by Maximilian SchreinerOpen source ↗

China's AI industry closes ranks as Deepseek ships open-source software for Huawei's Ascend chips

The Decoder · 30 September 2026

Loading the full article…

This text was published by The Decoder and written by Maximilian Schreiner. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Coverage and discussion

4sources
Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Hardware & Compute

All →

Related stories