China claims domestically-designed 14nm logic chips can rival 4nm Nvidia silicon — architecture leverages 3D hybrid bonding techniques for claimed 120 TFLOPS of power

China Semiconductor Industry Association vice chairman Wei Shaojun claimed at the ICC Global CEO Summit that a new domestically designed AI processor using 14nm logic and 18nm DRAM nodes can rival Nvidia's 4nm chips. The architecture leverages 3D hybrid bonding for increased power, with a claimed throughput of 120 TFLOPS. The design aims to reduce China's reliance on Nvidia's CUDA ecosystem and establish a domestic supply chain for AI development. While the technical details were not fully disclosed, the chip is expected to be formally revealed later this year, potentially challenging traditional semiconductor development strategies.