en.Wedoany.com Reported - COSMOPlat, in collaboration with Hygon DCU, has built an industrial AI agent system based on the synergy between a domestic computing power foundation and vertical industrial large models, achieving "autonomous computing power, specialized models, scenario-based deployment, and secure controllability." This accelerates the transition of industrial large models from technical validation to practical application.
The intelligent transformation of the manufacturing sector faces multiple constraints, including the disconnect between general-purpose large models and industrial realities, cybersecurity risks associated with overseas computing power, and low computing power utilization, all of which hinder the large-scale deployment of industrial AI.
As a domestic high-performance computing product, Hygon DCU is tailored for high-density AI computing scenarios in industry, supports seamless migration of CUDA code, and is compatible with mainstream AI frameworks and the vast majority of open-source large models. It meets the cybersecurity compliance and data security requirements of industries such as energy and heavy machinery. Its full-coverage computing power across endpoints, edges, and clouds provides a domestic computing power foundation for the large-scale deployment of industrial AI, supporting the application of COSMOPlat's Tianzhi industrial large model in more industrial scenarios.
Leveraging Haier's over 40 years of manufacturing experience, the COSMOPlat Tianzhi industrial large model has accumulated extensive manufacturing data and process know-how, incorporating over 200 expert algorithms and more than 110 agent development tools. It has deployed over 40 high-value industrial agents, serving multiple leading enterprises. In terms of computing power efficiency, COSMOPlat natively adapts to Hygon's domestic computing power, achieving dynamic allocation through intelligent scheduling, increasing overall computing power utilization by 30% and reducing resource idle rates to below 10%. In scenario deployment, leveraging Hygon DCU's low-latency, high-parallel computing power, the optimization cycle for vertical industrial models has been compressed from weeks to days, improving response efficiency in scenarios such as equipment diagnostics and intelligent recipe generation. In terms of security and compliance, the entire solution achieves full-chain domestic production of computing chips, model platforms, and industry applications.
COSMOPlat will continue to leverage its industrial internet platform capabilities and industrial large model expertise, partnering with domestic computing power providers like Hygon DCU to build a scalable, highly reliable, and fully autonomous domestic industrial AI application system, driving the effective deployment of industrial AI across more industries and scenarios.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
