The examples where this isn't true (lockheed Martin's skunkworks, Apple) are very rate. More common are large companies shifting from innovation to services (IBM an GE).
This makes sense. Will one company necessarily know how to commercialize a technology? This is one reason the Angel and VC models work. They separate the innovation from big company politics, and create an enviroent where killing old products is encouraged.
VCs have their place, but they fund ideas with shorter runways. It's the opposite model to research.
Smaller firms lack development dollars but have more flexibility in finding a market niche. For early stage innovation, being small also can force simplicity.
The mouse was never invented at PARC! http://en.wikipedia.org/wiki/Douglas_Engelbart
The soul of that company became Agilent when HP/Agilent split a decade ago.