The hyperscale/cloud datacenter market is single-digit millions of CPUs shipped per year. So if the whole cloud went to Arm tomorrow it would be less than 1% of total Arm chip shipments.
My point is that unless there's a way to extract higher $ per chip in the datacenter then actually it doesn't make a difference for Arm.
Edit: By datacenter market I'm really referring to AWS/GCP/Azure... the rest of the market ain't going en-masse to Arm anytime soon.
That wouldn't make business sense.
But it also wouldn't make business sense for Amazon to hang their hat on ARM without assurances or contractual guarantees that they can keep selling ARM chips for a reasonable price into the future. Where I define "reasonable" as Arm taking a much, much smaller cut of the final price than what Intel can do.