Maybe Google is favoring TPUv4 over whatever GPU runs on its platform? With Hopper 100 on the way, I wonder when TPUv5 will come out. I also wonder how Intel'

Maybe Google is favoring TPUv4 over whatever GPU runs on its platform? With Hopper 100 on the way, I wonder when TPUv5 will come out. I also wonder how Intel's Gaudi2 vs Ponte Vecchio will work together, looks like duplicate efforts for me. AMD has its MI300 on the way, but it seems still far behind Nvidia|TPU|Intel at this point.