Chinese artificial intelligence company DeepSeek has delayed the launch of its new AI model, R2, due to technical difficulties encountered while training the model on Huawei's Ascend chips. The delay, which pushed the launch from May to the end of August 2025, comes after Chinese authorities encouraged DeepSeek to adopt Huawei processors instead of Nvidia's systems. Despite efforts, including on-site support from Huawei engineers, DeepSeek faced persistent issues related to chip stability, interconnect speed, and software stack maturity, which are currently more advanced in Nvidia's technology. As a result, DeepSeek resorted to using Nvidia chips for training the AI model while planning to use Huawei's chips for inference. This setback highlights the challenges China faces in replacing U.S. technology in its AI development sector, underscoring the limits of Beijing's push for domestic chip reliance. The Financial Times and multiple industry sources reported on the delay and technical struggles, noting the implications for China's ambitions in AI and semiconductor independence.