Once upon a time, people kept their life savings on their person or at their homes, stored in physical material like gold and jewelry and property. Then money was invented as a medium of exchange, and people stored a surrogate of their wealth. Then banking was invented, and people centralized their holdings in a bank, and were paid interest for the privilege. Why were they paid? Because the banks could reuse their money by lending it out, at an even greater rate of interest. Money is fungible. I do not lose anything by storing it at the bank (and allowing them to lend it) except the privacy of keeping secret how much money I have, and risk that the bank will be unable to pay me back. The first is resolved through regulations, and the use of multiple banks, the latter by insurance. In any case, it is much safer than storing the money in a mattress at home.
Once upon a time, people kept their life's information on their person or on computers at their home or work, stored in physical material like floppy disk drives, hard disk drives, solid state drives, CDs, DVDs, and USB chips. Then the internet was invented, and centralized servers were made inexpensively and redundantly, and people could store their information in the "cloud". In many cases the cloud is free, or charges only a small fee. In exchange, the recipients agree to allow their personal information to be used to generate customized advertising targeted at them personally. But imagine their were a way for the cloud to earn interest on information much the same way banks earn interest on money, by synthesizing it and "lending it out". Since information is not rivalrous, this may prove viable with sufficient artificial intelligence aimed at developing ontologies and computer intelligence. The risk is the loss of privacy. Alternatively the customer pays the cloud for storage and computation, retaining privacy, in exchange being relieved of duties of backup, which when neglected lead to all too much data loss.
Once upon a time people kept their personal transportation near their person, parking cars and bikes at their homes, workplaces, or other destinations. This was the only way to guarantee point to point transportation in a timely way where densities were low, incomes high, and taxis scarce. Then "cloud commuting" was invented, cars from a giant pool operated by organizations in the cloud would dispatch a vehicle that drives to the customer on demand and in short order, and then deliver the customer to the destination. The vehicle would have the customers preferences pre-loaded (seat position, computing ability, audio environment). The customer benefits of course by not tying up capital in vehicles, nor having to worry about maintaining or fueling vehicles. The fleet is used more efficiently, each vehicle would operate 2 times or 3 times or more miles per year than current vehicles, so the fleet would turnover faster and be more modern. Fewer vehicles overall would be needed. It is likely customers would need to pay for this service (either as a subscription or a per-use basis), there is no obvious analogue to financial interest payments (and while advertising might offset some costs, surely it would not cover them). However stores might subsidize transportation, as might employers, as benefits for the customers or staff.
The tension between centralization and decentralization has been continuous through the history of technology, each has its advantages and disadvantages (and strangely, each also has religious zealots convinced there is one true way). This is ultimately a question of costs and benefits, and who bears the costs and benefits.
I am skeptical that cloud commuting can be made to work quite yet, there are still a few more technologies to perfect. Having tested Zipcar, their system lacks in several ways, much the ways the first banks failed frequently. Zipcars are still not local enough, they charge too much for lateness, the technology is still imperfect. But imagine we have cars that drive themselves. (and to PRT-advocates, these will be cars driving on streets, there are not enough resources to build a new infrastructure network for specialized vehicles). Smart cars solve the localness problem, since the cars come to you. In a way it also solves the lateness problem, because there is no need to reserve a specific car for a specific window, any unused fleet car can be dispatched. There would need to load balancing features, and maybe coordinated carpooling at peak times. (It also saves on parking, especially parking in high value areas).
* Technological change, part 2: Autonomous vehicles
* The Future of Cars