Do you lose more money than you're saving due to decreased productivity? Maybe. But that's hard to prove, and "This office proposal is $X million cheaper because we can lease three fewer floors" is very straightforward.
More and more, I'm glad I never took Google up on their interest in me. They absolutely have the money for real private offices, even in their expensive location.
Of course, the follow-on argument is typically "but we can't find enough sq ft of office space in one location!" to which I say: hundreds of businesses do this every day, what makes you special? Stop making excuses and invest in your assets.
Still, a lot of places at least are consistent about it all the way up to the CEO.
It would certainly explain the preponderance of open offices...
The impact on productivity also depends on the type of work. I suspect it's a net loss in the long run for things that require or benefit from deep work.
Personally I think it's a model that scales better (ie supports more developers being applied to a problem) but requires more brute force modularity to contain the lack of system design that comes from lack of deep thinking (and thus we get microservices).