When you hire large number of engineers, you need to hire engineers to support those engineers. Then you need to hire people to support the engineers that are supporting the engineers and so on...
Basically as companies revenues grow, every cost center gets to put in a request to grow headcount. If you don't your colleagues will and you'll be looked down upon. then you take on shiny new non-essential projects and find ways to justify it. Launch it get promoted and leave. The next person comes in- decides it is crap- shuts it down and starts their own new project to reinvent the same thing . Org politics.
I dunno, MSFT products seem to have only gotten worse since 1995 - and their profits are doing just fine.
The quality of a product and its ability to make money seem to almost be inversely proportional.
In Google's case - what matters is - they're doubling headcount (largest expense) and revenue has barely budged.
It'd be nice if the products got better - but all they should really care about is if they're making more money.
I do get that there is a lot of extra complexity that goes from a startup solving a hard technical problems to a pre-IPO company that is solving a bunch of business problems as well as technical challenges. Yes companies gain less efficiency per hire as they grow, and many do grow for growth's sake these days, but it does make sense that some companies need a thousand people to do what a startup thinks they can do with 10.
But it certainly doesn't seem like Google is making great use of doubling headcount. On a product by product basis, at least for me, it feels like most Google products are worse than they were pre-pandemic.
If they had shown revenue growth highly correlated with headcount I would see strong evidence that my view is wrong, but the facts don't seem to point that way.