“Mark Zuckerberg said on a Meta earnings call earlier this week that the company is training Llama 4 models “on a cluster that is bigger than 100,000 H100 AI GPUs, or bigger than anything that I’ve seen reported for what others are doing.””
At 3.7MWh per GPU a year, the cluster burns at least 370GWh. When an analyst asked how Meta powers it, executives dodged. Meta calls Llama open while restricting commercial use and saying nothing about how it was trained.