This is a great point. The general approach of thinking of the job-after-next that you want to get, and looking at how your potential next job does (or doesn't) position you for it, is extremely useful.
I've seen a lot of recent college graduates work at a startup for a year or two as a cto. I would still hire them as junior developers. Working with engineers with a lot more experience and tech stacks is an invaluable experience.
I agree with you. Not that there's any way to control this, but I think it's always best to be the 3rd or 4th most advanced programmer on a team. You want to be good enough to get decent projects, but it's always good to have at least two people better than you. I say two because if there's only one "mentor", you'll absorb her biases.
Startup and large company programming jobs are very different. In large companies, being a good programmer is about knowing the tech stack and knowing how to maintain and leverage the code. In startups, new development is more important. I think the better way to go is the reverse of what you described: start at a big company for the learning experience (and seeing, in legacy horrors, what not to do) and then do a startup having learned those lessons. There certainly are lessons about engineering that are hard to learn except in a mature, already-scaled company.
The next time around he/she is taken seriously as a CTO in the startup world or in the corporate world.
And that opportunity is worth a lot of money in my book.