No, that is not true at all. I’ve been using GPUs for years and never heard of Pop OS. You need to install NVIDIA CUDA drivers or some other drivers based on the kind of GPU you have, and maybe some other stuff you don’t have. OK, I can imagine someone makes a Linux distribution with drivers preinstalled to make it easy, but you can always download and install them manually.
The usual way is to go backward. You might have Python and then TensorFlow, and you would start with Python and pull in the packages you need and then pull down what those need and so on.
The best plan is to just use Ubuntu and add what you need. Because you are likely using more than just an AI model on the PC. Perhaps you also need to run ROS2 or a web server or whatever, and those things expect some Debian-based OS like Ubuntu.
My latest setup is on macOS. This makes it easy because there is no choice of GPU on a Mac, so any software built for Mac “just works”.
If you happen to have a spare gaming PC, put it to use.