The way to do it is to find an open-source project which is strategically important to a company, and yet which can't be made proprietary without destroying its strategic value. For example, the reason Android is open-source is because it would be hard to convince a broad base of hardware partners to support it if were closed-source (and now that one particular hardware partner is starting to dominate, it's getting more closed-source). Webkit is open-source because it started as open-source, and yet it is strategically important to Apple, Google, and Opera (well, until the Blink fork, but Blink is also open-source because it was derived from Webkit). Linux is strategically important to IBM, RedHat, etc. because they sell enterprise consulting services; having the source code available gives businesses confidence that they can continue running even if the vendor goes out of business, and yet they still drop a lot of money on consulting services.
It generally does not work to start a company based on open-source principles without also thinking about the strategic implications of who else will use the technology and what incentive they have for keeping it open. For example, LiveJournal was based on open-source principles and developed MogileFS, memcached, Gearman, and LJ itself. The effect of this was that every time they did something wrong, someone would start a clone of LJ and a clump of users would move over. And then Facebook took memcached and used it as a core part of their architecture, and won the whole social networking space.