There's two broad use cases for an "Apple Cloud".
1) Those that specifically want/need to program in MacOS on Mac hardware, and need more hardware than what they own themselves. They're already served by companies like macstadium (
https://www.macstadium.com/usecases), though with the limitation that they each have to rent out the entire piece, or pieces, of equipment.
2) The general server market, which mostly wants to program in Linux, and only cares about cost/computation. This is the far bigger market, and thus I assume the one the OP is talking about Apple entering. Here the idea would be that energy consumption is a big part of the cost, and if Apple's architecture could deliver a server chip that's much more efficient than what's currently available, then Apple could offer server services more cheaply than the other big players.
For the latter, there's at least two problems:
1) Server chips are different from desktop chips, and AWS has put a lot of research into obtaining higher efficiencies, with their ARM-based Graviton chips. So Apple would need to redesign its desktop architecture for server use, and see whether it would be more efficient than the best AWS has come up with. That's a major research effort with no guarantee of success.
2) Producing chips to run Linux rather than MacOS goes against Apple's corporate philosophy.