It's not necessarily antithetical. They already operate a small fleet for the purpose of StreetView mapping. Imagine having an entire fleet of sensors all over

It's not necessarily antithetical. They already operate a small fleet for the purpose of StreetView mapping. Imagine having an entire fleet of sensors all over the world gathering data all the time. Traffic data, weather data, etc. The cars could also serve as a way to show people local advertisements. If I pay for a trip from A to B, Google can advertise for all the small businesses along the route from A to B. It would be like having billboards in the car tailored to each user. Furthermore, it would be super interesting if Google also took interest in Liquid Robotics as well if they see value in a global sensor network play. Also, what if cars could be used as a last mile wireless play? None of this would be trivial, but they wouldn't have the cost of acquiring plots of land for antennas like telcos do. Pairing of a cellphone with a mobile car antenna could be done by comparing the directional vector of the vehicle with the the directional vector of people needing to connect to an antenna. i.e. if I'm driving along side a Google car, that's the natural choice of antenna to connect to. You need to look at each car in the fleet as an asset with potential instead of purely as a capital expenditure limited to known revenue streams.