The industrial revolution of the nineteenth century led to the pervasive replacement of manual labor with steel-based machinery powered by coal technology. The visible icons of this revolution are Thomas Newcomen and James Watt with their improvements to the steam engine design.

One aspect that has received little attention is the role of the underlying industrial processes. Railway robber barons did not start from ground zero; they were able to build their empires without having to own coal or iron mines, or having deep knowledge about the extraction technologies. Different grades of steel with known properties became available to build locomotives and steam engines. Manufacturing became more efficient due to a number of standards. Standardized screw sizes in nuts and bolts made these parts interchangeable not only lowered the cost of building the railroad infrastructure, but also made possible the large scale production of firearms that the tycoons needed to defend their lairs.

A similar transformation is happening with the information technology industry. This transformation is being driven by the synergistic interaction of three technologies, virtualization, service orientation and grid computing. As in the industrial revolution, this trio of technologies allows an efficient division of labor. The payoff of this efficiency comes in reduced cost of the delivery of IT services and in their reach across market segments and across geographies. IT services will no longer be the exclusive privilege of large organizations that can afford a sizable in house IT organization; these services will be affordable to small businesses and even individual consumers, and not only in advanced economies but also in developing countries across the world.

There are three essential components to that drives an IT service: the application that defines the service, the data providing the user context, and the computing engines that power the application. Sixty years ago all the pieces were tightly integrated: software was custom built for a specific target machine, and data was essentially an appendage of the code. The industrial evolution analog would be a locomotive manufacturer having to mine the iron ore, doing materials research, making the different kinds of steel and even machining the bolts. This would be an expensive proposition. Since bolts would be unique, the user would be forced to purchase replacement bolts from the locomotive manufacturer. Industries in their initial stages tend to be vertically integrated in this manner, and their products are expensive, limiting their market reach.

An individual consumer can get connected to the world through e-mail in just 10 minutes through a Web mail provider. Fifteen years ago a similar user would have needed expertise to build a TCP/IP stack on top of Windows 3.1, and even with that expertise it would have taken a couple days to set up an ISP account and research and integrate the necessary components. Thirty years ago the user would have had to write a SMTP client or even purchase at least a PDP-11 computer and integrate a Unix stack. It would not have been that easy; it might have been necessary to start by compiling the source code and configuring it specifically for the target machine. A corporate or university research environment would have been necessary to start with a running system to run the compilation.

It is useful to draw an analogy with a mature industry to see this pattern at work. Let's look at the processes used by an automobile insurance company with national coverage to fix a fender bender for a client.

Unless the accident happened in a large city, the company may not even have a local office. The customer calls a toll-free number to file a claim. The insurance company assigns the case to a different company, a settlement company with local presence. An adjuster for the settlement company assesses the damage and refers the case to body shop to have the the repairs done. Meanwhile, the customer is given a temporary replacement vehicle from a car rental company while the repairs are made. We can trace the economic chain ten or twelve more steps until the point where raw materials are extracted. The insurance company can make a business, not because it has expertise in the myriad steps that it takes to deliver their automobile collision service, but because it can rely on a pre-existing infrastructure of services, each one with predictable time and cost. The level of predictability is such that the insurance company can come up with the cost of the insurance policy and reasonably predict what the profit margin will be.