> 2 is still pretty good for a lot of things though.
It doesn't stop at 2. When I left Amazon about a decade ago there were at least half a dozen digital money services/products. Each vertical (retail, 3-p retail, Kindle, Gaming etc.,) all had their own stored-value service managing a slightly different version of the same digital money (gift-card, merchant balance, game coins). The feature sets overlapped ~95%. It may well turn out alright for the company but if you are an engineer or manager tasked with building yet another such service it's no fun at all.
"Two is one, and one is none" is a phrase I've heard for critical survival equipment in several contexts. For example two regulators when scuba diving,
two tie-offs when securing gear to a pack,
It's not like they had any systems competing, unless we're talking competition between Boeing and FAA, where one fights for profit by destroying evidence and fudging results, and the other tries to fight for safety (nominally).
P.S. It didn't start with McDonnell-Douglas acquisition - 737MAX is not the first 737 with deadly design failure - the first one was the original one. Except there Boeing also destroyed evidence from wreckages to hide the issue.
2 is still pretty good for a lot of things though. Specifically, the features and changes you need from the thing can be prioritized