Large language models (LLMs) and other neural networks draw substantial power when processing complex artificial-intelligence (AI) and machine-learning (ML) workloads. Designed for traditional server ...