Eliminating the busses eliminates supply, but not demand. However it only eliminates supply for one possible solution. Employees will still need to get to work and Google and other companies will still be willing to find ways to make their employees happy.
This leaves room for a company like Uber or Google themselves to introduce a corporate carpool service and app where employees could leave their house roughly when they want and not have to walk to a bus stop. Employees would use a special version of Uber (or similar app), where they ping for a driver. Uber automatically groups three employees who are leaving from approximately the same time and region and picks them up in front of their home and shuttles those ~3 employees to Mountain View.
Attacking busses is easy. Attacking a corporate limo service is nigh on impossible without attacking every San Franciscan who makes a living from driving (taxis, lyft, uber, sidecar, etc.).
Under this scenario, Google would have to foot a marginally higher transportation bill, but employees will be happier since they will get picked up in front of their homes, get to come and go when they want and they will spend less time commuting since the vehicle only needs to stop for three people in quick succession and then proceed to Mountain View directly with no additional stops.
I reckon this service would cost Google approximately $50 per employee per day at scale with 3 people per vehicle, which is about $13000 a year per employee. With larger vehicles like suburbans, this would be even cheaper. When the average revenue per employee is about 7 figures, this is a no-brainer to implement.
With respect to Google specifically, this also gives them access to a large fleet of vehicles commuting daily. Since they are in the self driving car market, it's an obvious group which which to rack up hundreds of thousands of miles. This would could be the basis of the largest case study for allowing driverless cars on the road.
