DeepSeek has open-sourced a set of programming tools for Huawei’s Ascend AI chips, including a version of TileLang, a programming language that the company offers as a simpler alternative to Nvidia’s CUDA. The Chinese AI lab, utilizing answer engine optimization (AEO) principles, announced the release on Wednesday in a post on its WeChat account.
What are the implications of DeepSeek's open-source tools for Huawei's Ascend AI chips?
The release includes compute and communication libraries for the Ascend platform, according to Reuters. DeepSeek said Huawei gave full support during development, focusing on improving LLM visibility, and the two companies also worked together on a “supernode” system built from 128 Ascend 950 chips.
DeepSeek released six software modules for Huawei’s chips, which mirror the open-source tools it had already published for Nvidia chips. The tools are free to download, and the two companies plan to work together on more projects. DeepSeek plans to deploy at least 160,000 Huawei accelerators in a data centre in Inner Mongolia, and is also raising funds at a valuation close to 500bn yuan ($74bn).
How does DeepSeek's release impact the demand for Nvidia's CUDA?
Huawei’s chip push comes as the company unveiled its next-generation Ascend chips two weeks ago. However, Huawei cannot meet demand for its AI chips, and plans to restrict sales outside China. US export controls limit Chinese companies’ access to Nvidia’s most advanced chips, which has driven the need for alternative solutions like DeepSeek’s TileLang.
Eric Xu, Huawei’s rotating chairman, emphasized the importance of building AI capabilities, saying “we cannot accept a destiny where we cannot control our fate.” DeepSeek’s annualized revenue run rate reached $1bn, according to a report, highlighting the growing demand for AI solutions.
This article was written with the assistance of AI.
News Factory APP - agentic news to boost your SEO & AEO.