Detailed Report: Huawei Chip Crisis Delays DeepSeek’s New AI Model Launch
Launch Postponed
DeepSeek, a Chinese AI company, has delayed the release of its highly anticipated R2 model due to failures in training using Huawei’s Ascend chips. Despite assistance from Huawei engineers, the chips did not provide the necessary performance or stability. As a result, DeepSeek resorted to using Nvidia chips for training and reserved Huawei’s chips solely for inference tasks.Broader Implications
This setback highlights the ongoing challenges in China’s push for technological self-reliance as domestic chips continue to trail behind U.S.-made alternatives like Nvidia in terms of software support, chip stability, and inter-chip connectivity.Leadership Response & Competitive Landscape
DeepSeek’s founder, Liang Wenfeng, expressed dissatisfaction with the current progress and is urging the development of a more advanced version. Meanwhile, rivals such as Alibaba’s Qwen‑3 model are moving forward more smoothly with similar AI architectures.
Summary
DeepSeek’s launch of its R2 AI model has been delayed due to training failures with Huawei Ascend chips, forcing a fallback to Nvidia hardware. This incident underscores the performance gap in China’s chip technology and the difficulty in reducing dependence on foreign alternatives.