Carr argues that the shift from IT as a fragmented capital asset to a centralized utility service will create far more upheaval than the introduction of the PC and the Internet did in past decades. I think that's a bit of an overstatement, unless he is talking about the impact over time on companies that deliver IT solutions today. Broadband connectivity anytime, anyplace is a good example of a rudimentary utilty model in action, and, as Carr writes, it will be driven by continued innovations in virtualization, grid, Web services and other technologies:
Most of the broadly used components, from computers to operating systems to complex 'enterprise applications' that automate common business processes, will likely be purchased as cheap, generic commodities.
Carr suggests that IT utilities will pose a problem for companies such as Dell, Microsoft, SAP and Oracle, which thrive on selling direct to multitudes of corporate customers. The winners will come from the full-service vendors like IBM, HP and Sun; established hosting providers like VeriCenter in Houston; the Web utilities like Google, Yahoo and Amazon that have expertise in building large-scale infrastructure; and as yet unknown start-ups.
Carr uses the the same historical precedent as Sun’s president/science historian Jonathan Schwartz, citing the evolution of the electrical utilities as evidence of how IT will evolve into a services business. Carr asks:
Won't the private data center seem just as transitory a phenomenon – just as much a stop-gap measure – as the private dynamo? Won’t the rise of IT utilities seem both natural and necessary? And won’t the way corporate computing is practiced today appear fundamentally illogical -- and inherently doomed?
It does seem natural and necessary, but it will take decades to kill the way corporate computing is practiced today. I asked Carr whether the cost to build out IT utilities would be difficult to justify financially, especially given what happened to the ASPs (Application Service Providers) during the dotcom bust. Someone gets stuck with excess capacities, and as yet we don’t have efficient marketplaces, like commodity futures, for selling IT services. Nor is the standardized metering and billing infrastructure in place to enable IT utility marketplaces.
In an e-mail, Carr responded:
It seems to me that the expense of building out largely redundant and subscale infrastructures is exactly why we'll see a shift to more centralized, large-scale ones. But it will certainly take time to reach a mature utility model. As the utilities grow, they'll achieve ever greater economies and will slowly be able to take over ever larger chunks of existing corporate computing operations. If you're a big company like Wal-Mart, say, it's going to take a long time before an outside utility is able to match, much less exceed, the internal economies of your own vast computing operations (big manufacturers maintained their own electricity generators well into the 1930's and 1940's--50 years after utilities first emerged), but if you're a smaller company, the tipping point will come much sooner.
As Carr notes, "the biggest impediment to utility computing will not be technological but attitudinal." The first wave will come from the symbiosis between the early utility hardware providers and the early utility software-as-a-service providers. Carr pointed to Achieve, a provider of software for the healthcare industry, which uses a VeriCenter data center as its infrastructure for its solution.
Sun's Schwartz has said that five years from now most businesses will be buying computing, not computers. Like Carr, he also sees the migration barriers to the utility model as cultural rather than technical. Five years is optimistic, Sun's dream come true. The only convincing argument to drive a cultural or attitudinal shift is money. When purchasing computing from a utility costs a half or a third of what enterprises spend today to power their applications and data centers, the era of the IT utility will have arrived.