Running PyTorch on TPU [Theoretical]

I have a purely theoretical question.
Recently read that it is possible to run pytorch code on TPU. In this regard, the question is whether the forces of pytorch and TF have become equal. That is, pytroch itself is much more convenient than TF, but now it can be as effective as TF on TPU clusters, and now, with any identical hardware or service (even Google cloud), they are equal in power. I’m right?