This preprint presents an industrially feasible production scheme for supercomputing AI chips based on known Earth technologies, addressing key challenges in scaling high-performance computing hardware. The proposed approach aims to advance the development of next-generation AI and supercomputing systems, offering a practical pathway for large-scale deployment.
Wanjin Huang (Mon,) studied this question.