Not all open source projects achieve the famed positive flywheel; it takes decades, and most will fail in the process. The ones that reach exit velocity, though, become part of the fabric of civilization. At that point, it makes more sense to build on top of them rather than recreate the wheel.
Open source projects are by their nature more resilient, long-lasting and neutral than closed ‘service’ centric software. The incentives that influence their roadmap are very different from more commercial software. The outcomes they optimise for are often more narrow and more long-term than those built by tech companies.
And so open source projects that cross some sort of threshold (longevity? contributors? users? I’m not sure) end up being embedded in the machinery of the internet.