Tech war: Top scientist urges China to develop alternative to Nvidia’s CUDA software
Chinese scientist Li Guojie emphasizes that while China's AI accelerator chips are catching up in hardware, Nvidia's CUDA ecosystem remains a significant advantage that China must overcome for AI self-sufficiency. He suggests that China needs to develop a controllable AI software tool system that surpasses CUDA to truly compete with the US in AI technology.
Li acknowledges DeepSeek's impact on the CUDA ecosystem but notes that it hasn't completely bypassed it, highlighting the remaining barriers. He likens this challenge to replacing Western hardware-software systems and stresses the need for careful planning and long-term efforts to build a competitive AI software ecosystem.
via SCMP Full Text Feed
Chinese scientist Li Guojie emphasizes that while China's AI accelerator chips are catching up in hardware, Nvidia's CUDA ecosystem remains a significant advantage that China must overcome for AI self-sufficiency. He suggests that China needs to develop a controllable AI software tool system that surpasses CUDA to truly compete with the US in AI technology.
Li acknowledges DeepSeek's impact on the CUDA ecosystem but notes that it hasn't completely bypassed it, highlighting the remaining barriers. He likens this challenge to replacing Western hardware-software systems and stresses the need for careful planning and long-term efforts to build a competitive AI software ecosystem.
via SCMP Full Text Feed
