Even without totally revolutioning the Bayer tech, there are ways for it to evolve.
Sigma's Foveon is the most known example. It has its downsides, true, but we should consider that it hasn't been developed by the biggest and wealthiest company out there. Invest some billions in it and it would be fine in no time.
Moreover, the biggest problem with Sigma cameras is the electronics around the sensor. Put a Foveon inside a 7D and I'm sure many people would buy that camera.
Indeed, Sigma might get the most benefits by licensing their technology more than trying to do everything on their own - something they don't have the resources to do well.