This announcement to me is possible the strongest argument against building your own payments platform that I've ever seen.
Next to Walmart, Amazon is probably the largest single credit card payment recipient on planet earth right now, at least in volume of transactions. If Amazon went through the work as you mentioned, to build our their own payments solution for several years and eventually abandoned it in favor of using "off-the-shelf" Stripe. Now, I do recognize that I am sure Stripe is somewhat customized for Amazon's use case and they obviously aren't charging Amazon anywhere near 2.9% + 30¢ per transaction. So its a little different than the Stripe we use, but its still a fascinating argument in the "Buy vs Build" debate to favor buying over building.
I find it also interesting coming from Amazon. Since much of the work I am doing right now is moving internally-built technical systems to AWS' managed and built solutions. So I am deep in the transition of moving built technical systems to a buy model from AWS across many different technologies. To see Amazon dishing off their payments to another company instead of maintaining it in house just feels like the ultimate "circle of life" or rather, "circle of specialization".
Tech is getting so complicated now that its becoming harder and harder to justify in-house solutions to problems. It seems like anymore we are moving everything to be managed by a different provider with companies focusing on one or two things they are amazing at and everything else being passed to another company who specializes at that other thing.