I was at OSBC last week, manning the SpikeSource booth and was able to slip into Nick Carr's keynote. Nick Carr basically compared IT today to the state of electicity in factories in the early 1900s. During that time factories had their internal power plants that weren't added value to the outcome of the factory per se (ie not a "core competency"), but were a money sink.
Over the course of 20 years, technology and business saavy allowed people to move from having their internal power generators to take advantage of low cost centralized power generators. Nick said today that IT was a "mess of cables", but in the future IT would be like an electrical socket (ie you just plug in). Buzzwords of the future of IT include: grid, virtualization, SOA, on-demand, blades, etc...
Nick shared a principle that stated the supply of any business resource will gravitate towards it's most economically efficient model. Because IT is a general purpose technology it can take advantage of economies of scale. But this consolidation requires new technologies. Nick then reviewed the history of electrical consolidation in the United States. Basically it was fragmented (self-hosted), a combination of technology (via Thomas Edison) and business (Samuel Insull) allowed the consolidation to take place.
Some of the advances in technology include the emergence of AC, transmission lines, current converters, standardization of networks and AC electric motors. On the business side, Insull invented "modern utility". Insull thought that electricity could be sold as a utility, which apparently was a novel thought at the time. And the factories followed.
Over the next 20 years, the rate of utility suppply roughtly correlated with the inverse of the private supply. Now to apply this to IT. Lots of IT is redundant while many costs are fixed no matter how much they are utilized. He quoted that server, pc and storage utilization rates were 10-35%, 5% and 25-50% respectively. CIO's are currently squeezed. They are spending 70-90% of their budget on infrastructure that is undifferentiated. They would like to spend more on innovation and R&D. Technological advances that could allow for utility IT involved open source (allows for economies of scale and is very flexible), fiber optic networks, IT automation and "Current Converters" (virtualization, grid, web services).
Interesting point Nick made was that although IT was supposed to automate many tasks, IT itself is the last thing to be automated. Carr opined that what is now needed is the new "Insull". A pioneer who scales the IT utility model is lacking. Some mannner to introduce metering and pricing for utility is also required. And the last bullet Carr mentioned (which might be the hardest to tackle) was helping companies overcome the assumption that they should own IT (a huge marketing challenge). It was an interesting analogy. Certainly there are many IT areas where the analogy should hold.