I have to disagree with your premise. Entrepeneurs who are starting these sorts of companies were often engineers at larger employers and were on teams where originally built these platforms that had become open source and relatively successful while they still worked there. Then they realized they were able to raise funding to start their own company around the software they previously built and then decided to do so. This applies for most of the companies mentioned in the article, where open source software later released by the new companies were the mistakes they had made.
I don't disagree with your post; but I think we're talking about different things, because the article talks about multiple kinds of companies. You're right that the companies that are now relicensing are ones that were born from a tool they built first, before monetization was visible goal. I am addressing companies in a more general sense, when entrepreneurs search for market opportunities. That's where the article notes that some see "open source" as a market strategy on its own.
And really, the dichotomy contributes to the situation.
The ones who built useful tool and years later realized it may be monetizable are hemmed in difficult choices. Do they reneg on open source and go proprietary from this point forward, effectively forking their own product and leaving the gratis, libre one -- the one most people will have exposure to -- stuck on an old version? Do they split into an open core and proprietary enhancements? Do they write a novel license and hope consumers will self-select into tl;dr harmless amateurs and very handsomely paying corporations? And if SaaS providers take their old version, and that fork gains ground?
Meanwhile, the ones who used 'open source' chiefly as a customer acquisition lead will face the same set of challenges. In the end, any artifacts published under an open source license are forever -- as long as the interest is there, a sufficiently dedicated party can take it, use it, enhance it, try to build a business around it, and the like. And any past version is a potential competitor, so your business models must be tolerant of that fact. They rarely are.
