DeepSeek and Huawei Launch Open-Source Tools to Challenge Nvidia's AI Software Dominance
Nvidia's dominance in the AI industry has long been attributed to its powerful chips and advanced software development tools. However, Chinese AI startup DeepSeek and Huawei Technologies have now challenged this status quo with a suite of open-source programming tools built specifically for Huawei's Ascend AI chips.
The centerpiece of this package is TileLang, a high-level programming language designed as a simpler alternative to Nvidia's CUDA framework. This move aims to lower the switching cost for developers and make it easier for them to write code on Ascend hardware while still achieving full performance.
DeepSeek has already optimized its V4 AI model for Huawei's Ascend architecture, using the Ascend 950 chips in parts of its training process. The company is now formalizing this transition with tooling and infrastructure, including a new data center in Inner Mongolia that will deploy at least 160,000 Ascend 950DT chips.
The real contest here is over developer habits, as hardware can be swapped out easily but years of code and engineering muscle memory are more difficult to change. Nvidia CEO Jensen Huang has expressed concerns about the implications of this partnership for US interests, viewing it as a competitive threat.