> but with cloud computing becoming an increasingly winner-take-all market, it leaves little room for the software's authors to generate revenue via hosted offerings of their software.
There simply has been no time where a startup doing SaaS of open source software (whether it's own or someone else's) had a strong route to profitability; this isn't an emergent property of evolving market conditions.
It's perhaps being acutely recognized because there was a particular time when companies started around open source software and, either initially or after a period with no clear business model, adopted “we’ll do SaaS of our core open source offering” as the whole, or a central piece, of their business model, and pretty quickly realized why, despite hosted SaaS being a commercial reality longer than cloud computing has even been an idea, SaaS of a particular open source software product—while sometimes part of the business of a broader SaaS provider—has never been a common successful standalone business model supporting companies centered around the open source offering, and certainly not for company is looking for the kind of growth modern VC-backed startups hope for.
They like to spin stories about AWS being predatory, evolving conditions in the cloud market, and attacking open source, but the real issue isn't with AWS, or cloud computing except insofar as that happens to the modern place where hosted SaaS is found, but that they've adopted a combination of SaaS-centric business model and growth expectations that would never have been realistic with open source software (and which I suspect they will next discover isn't much more realistic with their new relatively liberal proprietary license model either.)
> If you think that the Cockroach Labs, Redis Labs, Mongos, Elastics, etc., of the world provide a net benefit to the software world, I think you have to contend with the fact that there is going to be fewer such companies in the future if AWS keeps eating their lunch.
Software that serves as underlying infrastructure for other software has value to end-users that is with reduced by risks associated with being proprietary, and but also has value to it't is reduced by the non-exclusivity that exists when it is not. That's fairly obvious, and an intrinsic problem for such software being efficiently provided in a basically capitalist system, but unless one’s ideological preconceptions make it difficult to accept that capitalism may not always be as perfectly efficient as it would seem to be in theory under the kind of ideal assumptions sometimes encountered in Econ 101 classes, or unless one has invested in firms without business models which ignore this dilemma, I don't see how that is anything to “contend” lwith.