Huawei and DeepSeek Challenge Nvidia's AI Software Dominance
Nvidia's dominance in the AI software market is facing a challenge from Chinese AI startup DeepSeek and Huawei Technologies. On September 30, 2026, they unveiled 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 to be a simpler alternative to Nvidia's CUDA framework.
The software package includes six modules that cover compute, communication, and programming capabilities for Ascend hardware. TileLang is built to extract full performance from the underlying hardware, making development easier than it is with CUDA. DeepSeek is preparing to deploy at least 160,000 Huawei Ascend 950DT chips in a new data center in Inner Mongolia.
This announcement builds on work already in motion, with DeepSeek previously optimizing its V4 AI model for Huawei's Ascend architecture. The company plans to move further away from Nvidia hardware for training its models. Nvidia CEO Jensen Huang has voiced strong concerns about the collaboration, viewing it as a threatening competitive move.
The real contest here is over developer habits, with hardware being relatively easy to swap out but years of code, tutorials, and engineering muscle memory being harder to change. Releasing the tools as open source is a deliberate choice, inviting outside developers to build on, test, and improve the stack.