Capacity planning in a cloud environment

In her post "Cloud computing killed the capacity star", Ivanka Menken brings up some good points. Just think what changes this could bring to the government budgeting process. The trends that Ivanka addresses could result in the following:

• Traditionally, demand management is used as an input into a capacity plan that is then used for capital expenditure budgeting. In the new world, this will no longer be sufficient, because the ability to rent capacity on "pay as you go" basis will now be a much more important input into the operational side of the budget.
• In economically unstable times (such as today) the focus is on low fixed cost. The accompanying budget strategy is to purchase infrastructure with excess capacity and sit on it for as long as possible. While this may not be the lowest initial cost approach, it does minimize the risk of mission failure over the longer term. In the new budget world, agencies can purchase the minimum infrastructure needed to meet current steady state requirements. This would typically result in an IT platform with minimal excess capacity. If more is needed later, one would use operational funds for cloud based services. This new strategy would conceivably result in lower initial cost and lower lifecycle cost as well.
• With IT services as a utility, budgeting for it will become more akin to budgeting for office electricity. Do agencies do a annual capacity plan for electricity? No. A flat rate is set based on another related factor ( number of employees, number of offices, etc). This approach could actually result in absolutely zero capital budget for IT !!

What do you bean counters think about that !!