Frankly, not a single one.

You'd need:

* Significant improvements in 3D printing

* the ability to construct circuits as part of 3D printing

* a much wider range of materials available.

Essentially, you'd need to be able to ship hardware as a digital good. Which raises the huge question of DRM. The recent years of software startups side-stepped that by going the SaaS approach. (You'll notice desktop software is pretty much dead. There are no major new projects in that area since probably ~2008 or so.)

HaaS is going to be much more difficult. (Not necessarily undoable. Minds on too much coffee have suggested going a biological route, where devices just decompose after a certain time)