There will be constant complaints about their excessive power but, other than China where Google has almost no sales, we live in neoliberalism (public policy run by business rather than business run by public policy) so nothing will likely actually happen that is highly consequential there.

I would suggest that even that view of the modern world of neoliberalism isn't necessary to feel optimistic about their resilience.
Even during the heyday of antitrust actions, when big firms were actually dismembered, the investors in those firms (keeping the pieces) almost invariably did very well. Standard Oil, Bell.
And certainly antitrust actions that end up as damp squibs are also just fine for investors. IBM, Microsoft, and so forth.

In one sense, a business that gets attacked for being too successful is...generally a successful one.

If you want to have a good investment return, you will require higher sales per present user.

I would add one small refinement to that: higher sales per present user on a per-share basis.
They do make oceans of profit. It's possible that a lot of this could be used to drop share count a lot. Lately it has been dropping maybe 1.4%/year.
I am not that hopeful about the magnitude of the benefit of that endeavour, as the average price paid will probably be high and more shares will keep getting created for compensation.
But revenue and earnings per share would be higher after a few years even with flat user base and revenue per user, and there could be some benefit.

You're right that the easy growth from more eyeballs will hit limits.
I do think there is still a considerable of scope for increasing ad revenue, and not-quite-ad revenue, on a few fronts.
For example, YouTube Premium has gone from ~30m to ~130m paying subscribers in the last 3 years, an example of the broad category of earnings from being (in effect) their own content rather than the advertising add-on to the content of others.

As an interesting aside, my table further up the thread can be updated with today's price, which is of course 9% higher.
The assumptions no doubt have many errors, but for whatever it's worth, the same arithmetic would now suggest one might expect a return of inflation + 6.3% to end of next year, half what it was in the post.

Jim