A group of Chinese semiconductor and AI companies has formed an alliance to push the adoption of locally developed processors for artificial intelligence (AI) projects, as Chinese regulators question Nvidia about any “back door” risks in its H20 chips. Initial members of the Model-Chips Ecosystem Innovation Alliance include Tencent Holdings-backed AI start-up StepFun; Infinigence AI; SiliconFlow; Huawei Technologies’ Ascend computing business unit MetaX; Biren Technology; Enflame; Iluvatar Corex; Cambricon Technologies and Moore Threads. The alliance was announced by Shanghai-based StepFun at the World Artificial Intelligence Conference (WAIC). The alliance reflected the growing efforts in the country’s semiconductor industry and nascent AI sector to push forward China’s tech self-sufficiency agenda – covering computing hardware, software and services – amid U.S. export restrictions on China and rising geopolitical tensions.
The group’s establishment comes amid an inquiry of the Cyberspace Administration of China (CAC) in Nvidia, which was recently given the green light by the Trump administration to resume shipments of its China-specific H20 AI processors. The CAC said its inquiry was based on recent reports of “serious security issues” involving Nvidia’s processors and demands from U.S. lawmakers to add tracking features to advanced chips. Moreover, U.S. AI experts indicated that remote-control technologies related to Nvidia’s chips have matured.
The Model-Chips Ecosystem Innovation Alliance also comes at a time when Chinese AI infrastructure collaborations are increasing. Infinigence AI – a computing infrastructure platform provider backed by talent from Tsinghua University and funding from domestic Big Tech companies – said it was working with Huawei to launch an updated version of its “compute ecosystem platform” in the Shanghai Foundation Model Innovation Center, a renowned AI incubator in the city.
Hong Kong-headquartered AI firm SenseTime unveiled the “compute mall” initiative with more than 10 domestic chip design companies, including Huawei and Moore Threads. This initiative would allow AI developers to freely combine and allocate a variety of computing resources, platform tools and AI model services just like “purchasing goods” at a supermarket.
During his visit to Beijing earlier this month, Nvidia co-founder and CEO Jensen Huang told Chinese media that the company was “working at full speed to restore the H20’s production capacity”. He said it currently took about nine months between placing wafer orders with the contract manufacturer and delivery of the chips, the South China Morning Post reports. AI spending by China’s major cloud services providers was set to reach CNY380 billion this year, up 60% from 2024.
