Big tech companies keeping tens of thousands of engineers on staff despite their core product not seeing substantial changes for years are just wasting money.
In big tech, having services without anyone having a semblance of maintenance responsibilities tends to lead to relatively quick failures. Even in boring enterprise companies, handing out maintenance to a 3rd party is still going to cost you, and is likely going to lower quality. I've seen way too many companies that ended up rewriting things after leaving a system mostly unmaintained for 5 years made it sop being fit to purpose.
If there's a lesson of software from the last decade is that undermaintained services become zombified software, and can eat your company's brains. That's why a place like google often would rather shut something down than claim it's finished.
With software, once you find PMF, it's not the building that's hard, but keeping the users paying and using your product. The entry barrier to building a copycat is very low. So you have to pile on resources to keep your software be the best available.
This is all fine and expected, you sacrifice engineering quality for development speed in the early days. Just expect to pay it back by keeping a large engineering team for many years to come, or see your entire tech explode in flames.
Those 'years' are finite however, Twitch has been mature for like a decade already. So finally the tech debt is all paid off, and given failure to develop major successful features, time to cut the staff.
May be a bit different now (as others said, free money is gone, so competition cropping up is much harder to do and much less a concern), but the tech boom of the 2000's very much worried about becoming the next MySpace to some startup's Facebook. That's why tech salaries became so high to begin with.
