I don't think that's true—there is a subset of the best and brightest who work on greenfield projects very decoupled from existing customer bases, and they get to blog / present / post a lot about what they're doing. There are quite a few "best and brightest" people who are in large companies or slow-moving industries. They're often constrained to existing tooling, because moving to fancy new tooling is a huge risk and time sink for limited reward. They might be using cool personal tools—fancy editors and keyboards and window managers—but the stack they work on is generally "legacy".
And usually the problem of "How do we make this work slightly better for millions of end users" ends up being a bigger problem than "How do we do something really cool as a demo."
You can handle a lot of restrictions if the domain/problem is interesting enough and the constraints put on you don't feel too taxing, e.g. because they aren't enforced for your role or team very much. I feel like company size just isn't a good indicator for personal enjoyment/growth/$whatever, lots of research oriented divisions in larger corporations will let you work on interesting topics and hand off the engineering part to teams with people that enjoy that particular aspect of our world, both working for the same company.
Two, they have their pick of workplaces and focus on finding the biggest problem or perhaps the biggest paycheck, not the most freedom in tools. I have perfect freedom in tools hacking on OSS by myself; I don't look for that in a job.
Why do you assume it's impossible to have a fulfilling career writing Java 8 CRUD apps for Windows Server 2000?
I work with Java 8 on a greenfield high-frequency transaction platform for a very large company. It is extremely satisfying to build the "world's largest" of something, and no amount of shiny features in a cute new language would deter me from this work.