From the angle I'm looking, AWS wants to sell Redis/Mongo/Elasticsearch because that's what the devs are using. The devs are using it because it solves one of their problems well (or, less charitably, because its a "standard part" of new-framework-de-jour).
I can't think of any piece of software that came out of a cloud provider into a place of being "the essential hot new dev framework".
I don't deny for a moment that Amazon/Google/Microsoft write a lot of software, but I haven't got an example I can point at of that moving out of their own ecosystem into the broader development community.
Things like Lambda, S3, Kinesis have become huge in the dev world in one form or another - Lambda as the serverless concept, S3 as the standard API interface, Kinesis as Kafka (not implying causation or copying, just that AWS independently developed a technology very similar to something that became huge in the dev world).
Not too sure about Microsoft.
Google has invented so many technologies that have shaped development that it would be exhausting to list. MapReduce, AngularJS, Kubernetes and TensorFlow each transformed the development world and that's just the tip of the iceberg.
And I'll give Amazon credit for Lambda and Kinesis, but at least in my head, S3 is just ludicrous-scale BerekleyDB (or any of the persist-to-disk forks of memcached including the one I wrote back when it was still written in Perl).
Amazon isn't in the business of creating and promoting new software to developers. Not in the way Mongo/Redis/Elastic are.
"Popular things" have to come from somewhere, and my optimistic side kinda hopes that _maybe_ these new "OSS unless you're a cloud provider" licenses will be a way to fund development of software too big and complex and ambitious that it won't ever get started as a "scratch your itch" project. I dunno if that'll pan out. But I'm glad someone is trying.
20 years ago, large chunks of "the web hosting industry" sold access to Linux/Apache/Sendmail/Postfix/BIND/vsftpd/Perl/PHP/MySQL/etc - often (at least in my part of the game) using paid-for WHM/CPanel or Plesk to manage all those.
While I sympathise with Mongo/Redis/Elastic et al, who're having their work packaged up and sold by Amazon/Microsoft/Google in their cloud offerings, I don't recall any outcry about all the open source projects local ISPs and CPanel/Plesk hosting companies bundled up and charged customers access for. I don't remember the authors of Linux, Apache/Sendmail et al complaining about "not getting paid" for their OSS.
I get it that the world has changed, and companies building the OSS they're built on wasn't so much of a thing back then (although MySQL is an example of that from 20+ years back...), but I'm not sure a choice by founders to invest significant resources writing OSS is a reason other companies should not be able to use software with he same licenses in the same way companies have been using it for decades.
I've super curious to see if these new category of licenses work out business-wise, I suspect if they do (which I'm not completely confident they will), that it'll allow development of software on a scale and complexity which the "scratch your own itch" OSS community couldn't tackle. There are almost certainly software project for which the initial coding curve is so steep and high that nobody is going to start working on them in their spare time or as an unofficial skunkworks project or 20% time project at work. It the _only_ people tackling large complex problems as "open source" (in a loose interpretation here) are PhD students, there'll be less amazing software created than if companies like Mongo/Redis/Elastic can pay teams of devs to work on software which while not "FSF certified open source", is still usable and understandable by others in ways that proprietary software is not. I could debug my not-technically-OSS Mongo problems, unlike my Oracle problems. That's not rms's vision of the future. But at least it's not Larry Ellison's either...