[maym86]

Exactly this. The hype was all in the prototyping phase of the technology, going as far as you can get to make a demo and impress some investors. Progress looked like it was happening quickly because a controlled demo could be made with existing technology.

Now that it's time to build full production systems the hardest problems and edge cases need to be addressed, which could be ignored for demo systems, and the solutions aren't available without some new research developments.