That is, if Amazon currently needs 50 planes' worth of capacity on average from third parties with a standard deviation of 10 planes' worth, and they decide to acquire 30 planes of their own, then it would mean residual demand for third parties would be 20 planes on average but still a standard deviation of 10 planes' worth.
However, the carriers have a very difficult job, and contrary to popular belief, they are phenomenally cost efficient given the constraints placed on them. Amazon may try to undercut on price, but given the cash flows and margins of large carriers, any significant pricing discount would introduce negative cash flows that could cripple even a behemoth like Amazon. And by significant, I mean anything more than 2-3% off of the rates that UPS or FedEx charges.
Given the constraints and inherent difficulty of the business, I would imagine the impact would be no different than if DHL tried to re-enter the US market.
If Amazon assumes the risks of volume variation, they gain a lot of scheduling flexibility (flying later at night, etc.), and they can better level load their volumes across their carriers and themselves. This is a win for both Amazon and their carriers.