TokenPost·
Former Google Employee Cites U.S. Power and Data Center Limits for AI GPUs

3TB
model loading size
The former employee said customers with strict latency needs require U.S. capacity and described an ideal 10-minute loading target for a three-terabyte model.