China apparently was unaware of how to utilize its AI clusters before the debut of DeepSeek's AI models, as computing centers reported only 20% of utilization.
DeepSeek's Models Have Made China Fully Utilize Its Domestic AI Clusters; Market Now Sees Increased Demand For Computing
The debut of DeepSeek not only brought advancements to China's AI inferencing capabilities but also made the nation realize that its pursuit of AI accelerators is going in vain since most of its AI clusters were being heavily under-utilized. According to Baidu's content platform (via @OedoSoldier), it is claimed that the deployment of DeepSeek's R1 models on Chinese AI platforms has seen a surge in demand for hardware power, given that domestic computing centers are being used more than ever as before DeepSeek, the usage rates were pretty low.
It is claimed that China's AI clusters were not optimally utilized because the nation jumped into the early "AI computing" bandwagon, which forced them to acquire hardware in large quantities, only to realize that the demand was too "optimistic." Moreover, despite having a robust hardware arsenal, it is of little use if there isn't an adequate software ecosystem in place, and this ultimately creates disparities in the utilization of hardware; however, now, the domestic industry is in a much better state.
Not only in China, but the global AI markets are becoming much more mature, and they are realizing that computational capabilities aren't the only aspect that drives the technology to newer heights; instead, other factors are involved. Interestingly, the report also claims that China has seen an attraction to clients running DeepSeek's AI models locally, which is why there is a surge in demand for hardware, mainly GPUs. The domestic industry is more inclined towards NVIDIA's Hopper generation AI accelerators and their flagship consumer GPUs like the GeForce RTX 5090.
China's advancement in the AI sector shows that technology has no boundaries. Despite the influence of the US and its allies, the nation is on track to progress significantly in improving the capabilities of in-house projects. Recently, we discussed how Huawei's Ascend 910C chip is reported to be closer to NVIDIA's H100 in performance, which is yet again a significant milestone, so the gap between rivals is definitely getting narrower for China.
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
