Will native GPU support under Windows be perpetually supported?

I have a question because I have had some issues with API changes in Tensorflow.

Indeed, Tensorflow has decided to stop native GPU support on windows and I would like to know if PyTorch wants to follow the same way.