Where're All The AI Chips?
In a long newsletter post, Ed Zitron argues that despite a trillion dollars in hyperscaler capital expenditures and NVIDIA's hundreds of billions in GPU sales, AI data center capacity is coming online very slowly, with an estimated $234 billion in GPUs and TPUs sitting in warehouses or unpowered data centers, and he calls Microsoft's claims of adding gigawatts of "capacity" deliberate deception. He predicts brutal write-downs, dead data center debt, and a collapse worse than the dot-com bubble.
