I think autonomous cars are years or decades away from being mainstream, if ever. Intel creating a division for this is a future bet along with VR, drones and wearables. More demo-ware than big profit generating product lines.
BK makes a fundamental assumption about autonomous cars - that they will be continuously connected via cellular and generate large data streams that Intel can manage and evaluate at the servers. Does anyone want their car to depend on a strong cell signal to drive itself and operate safely? I can see downloading traffic information and maps, but a car needs to work even if the cell signal fades or you go in a tunnel. Intel doesn't have a play in sensors, and trying to apply x86 to in car computers to do the collision avoidance calculations and routing, which are embedded solutions, seems to be a stretch.
In any case, Intel has major growth problems in the next few years, and self driving cars won't save it from having to make severe cuts in 2017. RUN BK RUN!!!! (Run Intel employees too before you're looking for a job with 1000 other people).