Navigating Liquid Cooling Architectures for Data Centers with AI Workloads

Many AI servers with accelerators (e.g., GPUs) used for training LLMs (large language models) and inference workloads, generate enough heat to necessitate liquid cooling. These servers are equipped with input and output piping and require an ecosystem of manifolds, CDUs (cooling distribution) and outdoor heat rejection. There are six common heat rejection architectures for liquid cooling where we provide guidance on selecting the best one for your AI servers or cluster.