[deltaqueue]

That "TPU advantage" might be slowing Google down (though likely not as much as their internal bureaucracy).

Porting CUDA-based research, debugging, and overall experimentation speed is likely slower.

The GPU is still king for training.

[awongh]

But maybe the TPU advantage is in inference? That's what I assume because the number of compute cycles are going to be all in inference vs training. So they could train on GPUs if they want.

[anthonypasq]

lmao, you know all Anthropic models are trained on TPU right?

[zobzu]

thats funny because my company sells them nvidia gpu for training. but im happy for the billions, they prolly use them for counterstrike!