A major difference is that cinema owners, airlines, sports teams etc. aim to maximize their users, and each additional user doesn't diminish the experience of o

A major difference is that cinema owners, airlines, sports teams etc. aim to maximize their users, and each additional user doesn't diminish the experience of other users (well to some extent, but not enough to make a huge difference to the demand). A lot of spare capacity in server farms seems to be due to the need to cater for demand spikes, so that the experience doesn't degrade. Each additional user will, after a certain point, harm the experience of the previous users. A stronger analogy would be a power utility. It would be quite difficult, I think, to do what you propose without being able to predict demand spikes well in advance and with a high degree of accuracy. Even if we could do that there's another issue: a power utlility can sell power to certain types of consumers at off peak times to soak up spare capacity, but electricity has many uses. A web-app isn't likely to have many non-core uses that would appeal to specific users at off-peak times. Cloud providers can use this sort of strategy, but I don't think it would be useful for individual web apps, unless you are selling something very generic. The best way to deal with this issue, from the perspective of an individual web app, if it bothers one, would probably be simply to use a cloud solution, and let them deal with the excess capacity issues.