His main point has always been that for significant breakthroughs, you start by searching for a problem that isn't even formulated yet. Solving the problems of your customers is the opposite of that. The impact of solving the "unformulated" problem may be so strong that it eliminates the need of solving the little particular problems. In that sense, office workers of the 1970s probably wanted better typewriters instead of GUIs.
Depends on your definition of "significant breakthrough". Technical breakthroughs are what Kay values, and that's OK. But to suggest they're the only kind of breakthroughs is myopic. Startups can, in their best incarnations, change human behavior and the way we interact with our world.
To call it purely technical is also myopic IMO... The GUI/typewriter change the previous post talks about is a paradigm shift much like in Kuhn's scientific revolution literature. Some neat aspects are paradigm shifts in like music production or other art, which is yes technical but achieves aesthetic end points and is about the human-tool relationship.
