Working for a frozen water company must have seemed like a sensible career move in New England during the 1850s. During the age of steam, America's ice traders carved 10 million tons of the stuff out of local rivers and lakes every year. They kept it cool in specially constructed icehouses, wrapped it in hay and then exported it around the world.

In London, during the mid-1800s, no elite dinner party was complete without a sparkling mountain of ice cut from Wenham Lake in Massachusetts. Frozen water from New England also found its way to destinations as diverse as Martinique, India and Singapore. With it came recipes for ice cream and cocktails dreamed up by promoters to generate demand.

In the end, it was the electricity generators that killed off this thriving business. Equipped with fridges, householders began to make their own ice. By 1920, the frozen water trade had all but disappeared in North America. This story of this mighty industry that melted away crops up in Nick Carr's latest book, The Big Switch.

According to Carr, cloud computing threatens the traditional IT department just as surely as electricity generation once threatened the ice traders. In retrospect, he argues, today's IT departments will come to be viewed as an evolutionary dead-end -- a temporary aberration necessitated by client-server computing but wiped out by The Cloud, which is emerging as the dominant mode for corporate computing.

If that argument sounds familiar, that's because it refines the thesis that Carr first propounded five years ago in a paper for the Harvard Business Review called "IT Doesn't Matter". In that article and in the subsequent book, Carr lunged aggressively at the shibboleths worshipped by IT professionals and at the industry that generates US$2 trillion in revenues annually by selling hardware, software, services and bandwidth.

IT, Carr proposed, had become a mere cost of doing business. Increasingly commoditized, it was "essential to competition" but "inconsequential to strategy".

Carr argued that CEOs needed to aggressively restrict the ambitions of IT departments. As for CIOs, he pleaded that their "ultimate professional goal" should be to "render themselves obsolete".

This would happen, he said, once the infrastructure of IT became "so stable and robust, so taken for granted, that it no longer requires active high-level management".