because on a per-inference level, it's _still_ cheaper to use GPU than it is to use CPU, even if you ignore latency and only consider offline batch applications.
(to be fair, you can do better than the OP's performance, but we _are_ talking several magnitudes of performance, not to mention energy efficiency, which matters at scale)
What? Who's doing 400 t/s @ 7b on a 4090?
batch size 1 is more like 200 tok/s, still a very large improvement. the reason why this is still an interesting comparison is that it's comparing consumer hardware vs scaled datacenter setups. datacenter setups are much more efficient in large part because they are designed to process many inputs at once, fully utilizing the hardware. there are advantages to scale in inference, for the gpu-rich.