Login to Continue Learning
China is reportedly making strides towards developing 5nm AI GPUs. A new report indicates that domestic firms are preparing to create new chips with enhanced performance capabilities.
**China’s Anfu Technology Partners Up With Xiangdi to Introduce Two New AI Chips Featuring 5nm Technology**
Beijing is focused on creating in-house solutions to reduce reliance on Western alternatives and ensure the domestic industry relies on an internal tech stack. According to a report by MyDrivers, China’s Anfu Technology and Xiangdi are developing a new generation of ‘Fuxi’ GPUs that are claimed to have “excellent performance,” particularly in parallel computing tasks. Two chips are being planned: one for AI training and the other for AI PCs. Both will incorporate 5nm technology.
The report details that the next-gen Fuxi GPUs will feature up to 160 TFLOPS of FP32 compute capabilities on a 5nm process. The integration of this advanced process is noteworthy, given that China’s domestic semiconductor firms have struggled to scale beyond 7nm and there are reports of TSMC’s 5nm chips in the region. While Anfu Technology might have utilized TSMC’s nodes, this remains speculation. However, these advancements demonstrate China’s commitment to a computational breakthrough.
One chip, the Fuxi A0, is designed for rendering capabilities, while the other, the Fuxi B0, integrates an onboard NPU and supports mainstream AI models such as DeepSeek R1. The B0 chip can serve as a viable end solution, whereas the A0 is targeted at ‘AI PC’ workloads.
Details about Anfu Technology and Xiangdi’s new AI solutions are still uncertain. Nevertheless, it is fascinating to observe how rapidly China’s computing capabilities are scaling up. These advancements are largely driven by the government’s narrative to reduce reliance on NVIDIA’s AI chips and opt for local alternatives like Huawei and Cambricon.
