China is reportedly advancing towards 5nm AI GPUs, as a new report claims that domestic firms are set to tape out new chips that will bring new levels of performance.
China's Anfu Technology Partners Up With Xiangdi To Introduce Two New AI Chips, Featuring 5nm Technology
Beijing is focused on developing in-house solutions that reduce reliance on Western alternatives, and more importantly, ensure that the domestic industry relies on an internal tech stack. In a report by MyDrivers, it is revealed that China's Anfu Technology and Xiangdi are developing a new generation of 'Fuxi' GPUs, which are claimed to have 'excellent performance', especially in parallel computing workloads. Two chips are being planned for now, which will cater to both AI training and AI PC markets and integrate a 5nm process.
The report reveals that the next-gen Fuxi GPUs will feature up to 160 TFLOPS FP32 compute capabilities and a 5nm process. The integration of this process is rather interesting, given that China's domestic semiconductor firms haven't managed to scale up beyond 7nm, and reports about the presence of TSMC's 5nm chips in the region. Hence, Anfu Technology might have utilized the Taiwan giant's nodes, but this hasn't been confirmed yet. However, these advancements show that the Chinese AI industry is committed to a computational breakthrough.
One of the chips is claimed to be the Fuxi A0, which will be focused on rendering capabilities. The other is the Fuxi B0, which is an AI-focused option integrating an onboard NPU as well. The B0 chip will support mainstream AI models such as DeepSeek R1, which means that it can prove to be a viable end solution. The A0 chip is said to be targeted towards 'AI PC' workloads, while the B0 will focus on end-side model deployment.
Details around Anfu Technology and Xiangdi's new AI solutions are uncertain for now. However, it's interesting to see how rapidly China's computing capabilities are scaling up. These are mainly driven by the government's narrative to 'not rely' on NVIDIA's AI chips and instead opt for the likes of Huawei and Cambricon.
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.
