This is the key idea.

Companies have economy of scale (Google, for instance, is running dozens to hundreds of web apps off of one well-maintained fabric) and the ability to force consolidation of labor behind a few ideas by controlling salaries so that the technically hard, detailed, or boring problems actually get solved. Open source volunteer projects rarely have either of those benefits.

In theory, you could compete with Google via

- Well-defined protocols

- That a handful of projects implement (because if it's too many, you split the available talent pool and end up with e.g. seven mediocre photo storage apps that are thin wrappers around a folder instead of one Google Photos with AI image search capability).

- Which solve very technically hard, detailed, or boring technical problems (AI image search is an actual game-changer feature; the difference between "Where is that one photo I took of my dog? I think it was Christmas. Which Christmas, hell I don't know" and "Show me every photo of my dog, no not that dog, the other dog").

I'd even risk putting up bullet point four: "And be willing to provide solutions for problems other people don't want solved without those other people working to torpedo your volunteer project" (there are lots of folks who think AI image detection is de-facto evil and nobody should be working on it, and any open source photo app they can control the fate of will fall short of Google's offering for end-users).