I don't quite follow, but I totally believe that solutions to these problems can be made and maybe your solution really does work well. The issue is that solutions here are not standard yet and thus there will not be much collective knowledge to draw upon. I'd much rather ask engineers to fix age-old RDBMS issues than novel ones that are unsearchable with solutions whose repercussions are not yet known.

There's also issues with complexity even if the solutions get figured out. What are you getting in return? Serverless is great IMO when it's the simplest solution to a problem, ex: "I only have a front end but I really need a simple API action that daisy-chains a few API calls".