I'm not aware of any serious GPU-type server hardware that has WiFi :)
You can run AI models slowly even streaming weights off a hard disk. The GPUs make inference fast enough to be pleasant to work with but do not determine the capability of the AI, only the speed.