« Death by Venice | Main | IT doesn't matter, part 5 »

IT doesn't matter, part 4

January 08, 2007

This is the fourth installment of the article "IT Doesn't Matter." Part 1 is here.

The Commoditization of IT

Although more complex and malleable than its predecessors, IT has all the hallmarks of an infrastructural technology. In fact, its mix of characteristics guarantees particularly rapid commoditization. IT is, first of all, a transport mechanism – it carries digital information just as railroads carry goods and power grids carry electricity. And like any transport mechanism, it is far more valuable when shared than when used in isolation. The history of IT in business has been a history of increased interconnectivity and interoperability, from mainframe time-sharing to minicomputer-based local area networks to broader Ethernet networks and on to the Internet. Each stage in that progression has involved greater standardization of the technology and, at least recently, greater homogenization of its functionality. For most business applications today, the benefits of customization would be overwhelmed by the costs of isolation.

IT is also highly replicable. Indeed, it is hard to imagine a more perfect commodity than a byte of data – endlessly and perfectly reproducible at virtually no cost. The near-infinite scalability of many IT functions, when combined with technical standardization, dooms most proprietary applications to economic obsolescence. Why write your own application for word processing or e-mail or, for that matter, supply-chain management when you can buy a ready-made, state-of-the-art application for a fraction of the cost? But it’s not just the software that is replicable. Because most business activities and processes have come to be embedded in software, they become replicable, too. When companies buy a generic application, they buy a generic process as well. Both the cost savings and the interoperability benefits make the sacrifice of distinctiveness unavoidable.

The arrival of the Internet has accelerated the commoditization of IT by providing a perfect delivery channel for generic applications. More and more, companies will fulfill their IT requirements simply by purchasing fee-based “Web services” from third parties – similar to the way they currently buy electric power or telecommunications services. Most of the major business-technology vendors, from Microsoft to IBM, are trying to position themselves as IT utilities, companies that will control the provision of a diverse range of business applications over what is now called, tellingly, “the grid.” Again, the upshot is ever greater homogenization of IT capabilities, as more companies replace customized applications with generic ones.

Finally, and for all the reasons already discussed, IT is subject to rapid price deflation. When Gordon Moore made his famously prescient assertion that the density of circuits on a computer chip would double every two years, he was making a prediction about the coming explosion in processing power. But he was also making a prediction about the coming free fall in the price of computer functionality. The cost of processing power has dropped relentlessly, from $480 per million instructions per second (MIPS) in 1978 to $50 per MIPS in 1985 to $4 per MIPS in 1995, a trend that continues unabated. Similar declines have occurred in the cost of data storage and transmission. The rapidly increasing affordability of IT functionality has not only democratized the computer revolution, it has destroyed one of the most important potential barriers to competitors. Even the most cutting-edge IT capabilities quickly become available to all.

It’s no surprise, given these characteristics, that IT’s evolution has closely mirrored that of earlier infrastructural technologies. Its buildout has been every bit as breathtaking as that of the railroads (albeit with considerably fewer fatalities). Consider some statistics. During the last quarter of the twentieth century, the computational power of a microprocessor increased by a factor of 66,000. In the dozen years from 1989 to 2001, the number of host computers connected to the Internet grew from 80,000 to more than 125 million. Over the last ten years, the number of sites on the World Wide Web has grown from zero to nearly 40 million. And since the 1980s, more than 280 million miles of fiber-optic cable have been installed – enough, as Business Week recently noted, to “circle the earth 11,320 times.”

Part 5

Posted by nick at January 8, 2007 09:00 AM

Advertisement: Are you ready for "The Big Switch"? Nicholas Carr's new book "is the best read so far about the significance of the shift to cloud computing," says the Financial Times. The Independent says it's "lucid and mind-boggling." Order now from Amazon.com.

Comments

The costs of processing, data storage and transmission have dropped relentlessly since the 1970s. And companies from Google on down have dramatically and tellingly abandoned customized processors, storage devices and networks in favor of blades, RAIDs, and Ethernet.

Software is different.

Yes, you see application services providers doing increasing business for customer relationship management, supply chain management, billing, receivables management, etc. What is interesting is that for every generic application doing well on the market today, I guarantee there's an outsourcing company that's doing well too. They don't just provide a generic application - they also provide generic users to interface with it.

If I may be so bold: Every application is a process. If a business thinks that their core competence is not in optimizing that process, they will either outsource and genericize the hell out of it. Witness email, word processing, spreadsheet, and the browser. The more a company believes a process is core to their success, the more custom that process will be, and the less a generic application will make sense.

Not long ago, I worked for McMaster-Carr, a distributor of industrial supplies. Their document handling process was both idiosyncratic and core to their business, and if you think document handling was a generic process or that management wanted to make it so, then you have another thing coming. No one wants to commoditize the business process that makes them their money. That would mean that anyone could buy the process and compete with them for their lunch money.

Who's that stupid?

So there's a lot of parts of IT that will be and is commoditized and commoditizable. Everything, in fact, that's not part of the business's "special sauce." But the parts that are will always be customized. In this sense business IT will always be like the stock cars that race in NASCAR - off the rack but still playing on a whole nother level.

Posted by: Steve at January 8, 2007 04:32 PM

IT doesn't matter, part 4

January 08, 2007

Comments

Post a comment

Greatest hits

Other writing

Rough Type is: