The Professor of Outputmaxxing — Anjney Midha, AMP
Frontier lab xAI may be running at sub-10% Model FLOPs Utilization, far below best-in-class 60-70%.
“The AI scaling debate always focuses on the question of "how do we get more GPUs?" but the better question may be: how do we make the most of ones we already have.”
An interview with Anjney Midha (AMP) reframes the AI scaling debate around Model FLOPs Utilization, noting xAI may run at sub-10% MFU versus best-in-class 60-70% and historical runs like PaLM at 46%. It matters because efficiency gains on existing GPUs could rival adding more compute, reshaping how labs think about scaling.