AI Agents Are Thirsty for Power

What the source reports

WIRED reports that explains why agent workloads break old single-query energy comparisons: agents may run for hours, spawn many internal prompts, and scale by task complexity rather than user count. The Signal Loss read: Explains why agent workloads break old single-query energy comparisons: agents may run for hours, spawn many internal prompts, and scale by task complexity rather than user count. Uses examples such as OpenAI's math swarm and Meta's Muse to frame AI data-center buildout as a consequence of agentic workloads, making energy accounting part of agent governance.