I can't speak for anyone else, but here's my thoughts on it; I have Solaris and Oracle experience in large enterprise environments, but it's not something I would consider at a Startup stage. And I'm not anti-Oracle by any means, it just comes down to maximizing my resources. For Oracle, you have these huge up front costs and buy in, and then you consider that there's newer (aka cooler) technology out there that has been proven to work (aka scale) and its way cheaper if not free. And then you consider the shift towards cloud based service delivery and the concept of on demand resources and pay per use. For the "startup mentality" it just seems like a no-brainer to start out with the cheap/free stuff that works, pay per use, and minimal buy in. Best bang for the buck.

So then you build it, it's running, and it grows. And then like any IT solution you build, you have this "lock-in" factor where you've built something on a specific platform in the cloud (or whatever) and it's painful to move to something different. This is compounded as the Userbase (and Dataset) grows bigger and bigger. So you just keep it going as long as you can and keep scaling, and I would guess this is why nobody in the Startup world uses Oracle.

If Oracle wants to power more Startups, Oracle needs to provide an affordable product that Startups choose to use from the start. They would probably kill it if they developed their own version of AWS. Seems that for most of the people I talk with, Oracle is synonymous with Enterprise, not Startups or new innovative technologies. So in some ways, the perception of Oracle and who uses their product might also play a factor. The knowledge and skill requirements are somewhat of a gray area for me, as any approach requires knowledge and skill to implement. I guess I wouldn't describe one approach as requiring more knowledge/skill than another (they all require knowledge/skill to implement, maintain, and scale).