Six months ago, there was a lot of focus on the idea that the there would be a massive glut of unused computing power which would could a recession as AI use plateaued. The "compute bubble" belief was absolutely everywhere. The degree to which this was wrong deserves some notice