AI Agents for AI Infrastructure: Rethinking Operations at Scale

AI workloads are changing the operating conditions of the data center. As GPU clusters scale and workloads become more dynamic, maintaining reliable, high-performance compute requires infrastructure that can respond just as dynamically. In this session, Phaidra will share how CoreWeave is leveraging Phaidra’s AI agents to optimize AI infrastructure. Phaidra’s Liquid Cooling Agent uses real-time rack and CDU data to continuously optimize cooling as GPU workloads change, improving thermal stability and reducing the risk of GPU throttling. We’ll also show how Phaidra Prism gives operators intelligence and observability across their fleet of CDUs.