I've heard it rumored that one of Peter Thiel's private beliefs is that tech bubbles are necessary to fund non-trivial innovation in new industries (when expected value to investors is largely negative). This is a fascinating question either way.
I basically agree with him but I'd put it a bit differently: non-trivial innovation requires something able to burn money on it beyond the constraints of rational short to medium term profit motive.
That could be a bubble that throws off so much surplus money that some of it gets spent on things that end up "failing" but generating a return in the long term. It could also be large government R&D expenditures, and that's indeed what it was through the most innovative period of the 20th century. Lastly it could be a massive cash cow monopoly like Ma Bell and its Bell Labs, but that usually overlaps a bit with government since usually those kinds of monopolies exist because of some form of government action in the marketplace.
