Is there a decoupling of AI technology between the US and China? Playground founder warns of the risk of a two-track split in the open-source ecosystem.

According to Beating's monitoring, Playground AI founder Suhail Doshi pointed out at X that China's progress in achieving self-sufficiency in computing power will lead to a dual-track split risk in the global AI open-source ecosystem. Doshi warned that as China gradually reduces its dependence on Western hardware and builds its own computing power stack, China's massive and active open-source contributions will shift to a technological ecosystem that the United States is unwilling or unable to use. Against the backdrop of increasingly closed-source research and infrastructure by major US AI giants, the split in the global open-source ecosystem will backfire on the US's own AI innovation capabilities. This dual-track split in the technology stack has sparked widespread discussion in the industry. Commentators such as GeekPark pointed out that computing power self-sufficiency is leading to a decoupling of inward parallel technologies. Chinese R&D teams not only face more physical barriers to integrating into the global research cycle but also find it difficult to directly use the latest US model products. Industry engineers mentioned that the promotion of open-source software stacks such as Huawei's CANN heterogeneous computing architecture is allowing more chip manufacturers to implement unified interfaces, thereby accelerating the formation of a parallel hardware ecosystem outside of Nvidia's CUDA. As China is forced to rebuild its open-source toolchain from scratch due to export controls, American developers may face a dilemma in the future: either be forced to adopt an open-source system led and optimized by China, or have to independently rebuild all the technological wheels. Although Guillaume Verdon, an advocate of the Silicon Valley open-source movement, believes that the US computing power stack may soon surpass the traditional paradigm centered on floating-point operations, the fragmentation of the software ecosystem driven by hardware isolation has become a systemic variable that cannot be ignored in the US-China AI competition in the short term.