The idea is to use Apple Silicon to power these AI servers ... already allocated to NVIDIA could make things tough for Apple.