I’m biased, as I am guessing you are too, but there is at least some innovation in infrastructure at Google outside of search and ads. Spanner, Borg/Kubernetes/GKE, PAAS, BigQuery and the successive internal query systems just to name a few. Sure many of these were funded by ads or motivated by their problems/use case originally - you’d expect that of a company that started with search and exclusively made money from ads for a long stretch - but they’ve all been developed to solve for many for workloads now, and even made available for external consumption as infrastructure products. For some they have made huge impacts on the entire software industry.

That said, yes there are a lot of SWEs and projects that are wasting engineering talent.

I’m guessing you were closer to the hardware than me, and also left at least several years ago, but I’m under the impression that the rest of industry has mostly caught up to where Google was in DC tech. They’re still at the tip of the spear with things like ML and unmatched in sheer scale, it’s just that the rest of the world has had a lot of time to play catch up.