LinkedIn actually chose Ember well before React had won, and definitely before it had become the new “no one gets fired for buying IBM.” That all preceded my time there, but the evaluation was late 2014 and the LinkedIn.com web app rebuild in Ember was kicked off in 2015 if I recall correctly. React only came out in 2013, and while it was definitely on a rocket ship it was way less obvious in late 2014 that it was going to be the de facto standard that it was by late 2016–early 2017.

Bonus note: One factor that was really not obvious at the point they were doing the evaluation was how much TypeScript and TypeScript-powered tooling was going to matter. TS shipped support for JSX/TSX in TS 1.6, in late 2015, and I contend that while JSX had a lot of pros and cons, the one absolutely unambiguous win it had over everything else from that point forward was that whether you were writing JS or TS, you got outstanding editor support in basically every editor in the world courtesy of the TS language server backing your JS or TS development. Just absolutely massive.

The fact that the TS team baked support for JSX into TS itself and that it's not pluggable is (a) totally understandable and reasonable from a maintenance effort POV, especially given that Microsoft uses React extensively, and (b) extremely frustrating for everyone not using JSX. Vue, Angular, Svelte, and Ember all had to build the exact same infrastructure to get that support, and that was a huge productivity hit for devs using those frameworks compared to React in the meantime.