I disagree. It is meta, but not too meta. You're right, startups helping startups that help startups will fail. I think there are still enough startups for ONE round of meta. There's nothing that says it will go beyond that.

Sometimes you see posting on HN people saying they don't have a technical background but have a brilliant idea. You can create a company that creates prototype for startups. If you look at http://buglabs.net, they are giving you a tool to create your own product.

It's not scalable because the market isn't that big. Too far up the food chain.

Succesful startups: thousands, if not millions.

Startups for startups: B2B, or YCombinator, or Techcrunch.

Startups for startups for startups: can't think of any. These needs are already served by the B2B companies. At this stage, the ecosystem is small and covers itself.

The market just gets too small at a certain point, but going meta can work.

Startups are just an emerging niche. If you look at it that way, the metaness is irrelevant.

Also, such a startup would be guaranteed to have an intimate understanding of the problems they are trying to solve!