China's AI ecosystem is undergoing a major transformation as domestic chipmakers and cloud providers rapidly adopt DeepSeek's latest models. According to Reuters, GPU designer Moore Threads and AI chipmaker Hygon announced that their accelerators now support DeepSeek's R1 and V3 models, reinforcing China's efforts to reduce reliance on U.S. technology. Meanwhile, South China Morning Post reported that Huawei's Ascend cloud service, in collaboration with AI infrastructure start-up SiliconFlow, has benchmarked DeepSeek's models against high-end global GPUs, claiming comparable performance—a milestone in China's AI self-sufficiency.
Beyond chips, cloud giants Alibaba, Tencent, and Baidu have integrated DeepSeek's models into their platforms, per Reuters. DeepSeek's success proves China's "good enough" chips can sustain large-scale AI deployments, potentially reducing inference costs by over 50%. With domestic AI hardware and software aligning at an unprecedented pace, China is rapidly closing the gap with its Western rivals.
Commenting on DeepSeek’s rise, Chris Hung, Director General of the Taipei-based Market Intelligence & Consulting Institute (MIC), noted that lightweight AI and its continued expansion across industries will remain a key trend. As AI models become more advanced, lower costs and broader integration into everyday applications will be beneficial in the long run. While there are concerns about DeepSeek’s impact on industry players, Hung pointed out that DeepSeek and Nvidia maintain an upstream-downstream relationship, particularly in computing power and chip development, offering opportunities for collaboration.