i suspect its more likely that there are pools for c5a.large, xlarge, etc. and that they just aren't prioritizing as much buffer/liquidity in the large instance

i suspect its more likely that there are pools for c5a.large, xlarge, etc. and that they just aren't prioritizing as much buffer/liquidity in the large instance type pools. with chip shortages it may be worth it to them to prioritize more common workloads or those which they can get a consistent percentage of a host rented vs. all-or-nothing.