Thirty years ago, Intel founder Gordon Moore noticed transistor density on semiconductors doubled every 18 to 24 months, a phenomenon now known as Moore's Law.

Moore's Law drives the market for every major piece of computing infrastructure, including storage devices, RAM, and network adapters. It's begun to move switch, router, and optical network equipment companies, and will soon revolutionize the service business. As Sentient Networks CEO Nimish Shah puts it "Moore's Law very much applies to this market, and those that can't keep up with it are going to face some challenges."

Fortunately, a growing number of new carriers (such as Qwest Communications Inc.), CLECs (such as Covad Communications Co. of Santa Clara, Calif.), and research projects (such as Internet2) say they will abide by Moore's law. In other words, they will double their performance every 18 months and improve by an order of magnitude (10X) every 5 years while holding prices constant.

"Our objective is to improve services in line with Moore's Law," says the ubiquitous Dick Liebhaber, who serves as a member of Qwest's board of directors. "It's hard in the service business because it is more complex than technology price/performance curve - you've got to dig holes in the ground."Dan Lynch, one of Covad's investors, notes that CLECs are well aware of the need to improve services continuously. "Covad will find ways to keep the price the same and deliver more and more capacity, as opposed to telcos who say 'I'm giving you more so I can charge you more,'' he says.

Moore's Law is what separates the old carriers from the new, and it works because it provides a common tempo for the high dance - if you make disks, you know I/O has to improve to match CPU and RAM speeds. When you are in a market driven by Moore's Law, it doesn't pay to be out of step.

The other key to Moore's Law has been that demand for CPU cycles is elastic: the more there are, the more people will buy. In other words, customers prefer to buy the higher performance versions rather than spend less money. For the carriers willing to seize the opportunity, it's apparent that demand for data bandwidth is elastic too - in fact, it's outpacing Moore's Law. "In the last 13 years, prices for long distance services are down 60 percent, and Qwest just cut them in half again, but during the same time period, revenue for IXCs is up 2.3 times," says Liebhaber, who is also managing director of Veronis, Suhler & Associates, former CTO of MCI, and member of boards of Alcatel and Avici. "Demand is very elastic -- we have yet to scratch the surface of demand, because humans are wired visually, not aurally."

If Moore's Law for services is inevitable, who are the losers in all this? How about some of the traditional telcos, the Uncompetitive Local Exchange Carriers, who say their prices will remain stable? Demand is high, and they hope to maintain profit margins without dramatically improving service. That's led some, such as venture capitalist Roger McNamee, to suggest they are governed by "Moron's Law."

Morons or not, the exponential improvement that drives the computer industry is a radical departure for traditional telecom equipment and service providers, which have lived by slow, predictable demand growth, 20 year equipment write-downs, and cozy, 100 year relationships between with equipment companies like Nortel, Alcatel, Siemens, Ericsson and Lucent. Some argue that the Bells have no need to feel out of step, that connections from the home to the Internet are fifty times faster today than they were a decade ago. But that's not because phone service has improved. It's because modem manufacturers abide by Moore's Law.The teleopolies are stuck in a culture built for single digit growth. "The incumbent service providers carry a tremendous amount of legacy stuff, and that is what causes the mischief," Liebhaber says. "One piece of legacy they carry is customers - even if they figure out how to offer service at half price, it's hard for them to make a decision to cut revenue in half."

Living by Moore's Law made a winner of Intel. It also made a winner out of Seagate. It's rewarded the new service providers and equipment companies willing to live by it with astronomical market capitalizations.

For investors, the lesson is obvious: buy companies double price/performance every 18 months. Because it isn't just a good idea. It's the law.