"We had to," says Mohd Ridzuan Mohd Nor. This man of few words is the wireless broadband manager at Jaring, Malaysia's second-largest Internet service provider, and I've just asked him why his company is creating a 1-megabit-per-second wireless service. We're on the roof of a 33-story office building in downtown Kuala Lumpur to look at the antennas and radios that make up one of the network's 10 base stations in this city.

"We had to," he repeats and starts walking ahead of me, down an open-air corridor cluttered with steam pipes and air ducts. A 3-meter-high wall blocks our view of the famous Petronas Twin Towers, for a few years the tallest office buildings in the world. Suddenly he stops and turns, his arms at his sides, palms up, a gesture of resigned explanation. "Ninety-eight percent." That's the portion of the broadband market owned by Jaring's competitor, Telekom Malaysia Berhad, the national carrier, also based in Kuala Lumpur. It's a tiny market at the moment, made up of only about 350 000 households in this nation of 24 million. But Telekom Malaysia has 98 percent of it. Right now, three out of every 10 Internet users in the country is a Jaring customer, but just about all of them use dial-up modems. It's only a matter of time until they demand faster connections. Jaring thus needs to do something before 98 percent of its customers jump ship. It needs a network of its own. And unlike its competitor, it's going wireless.

Malaysia has long been a telecommunications leader in Southeast Asia. It was the first country in the region to hook into the Internet. Its enlightened spectrum management policies encourage innovation. A so-called multimedia supercorridor—a miniature Silicon Valley constructed during the global dot-com bubble of the 1990s—still flourishes. So it's not surprising that a wireless broadband service like Jaring's is found here and only a few other places around the world. This type of broadband is not available in Taipei, home of the skyscraper that supplanted Petronas as the world's tallest, nor in Chicago, where the Sears Tower once held that record, nor in any other large city in the United States, which once led the world in Internet access but is now lagging.

Jaring's transformation from mere service provider to network operator comes with considerable risk. For one thing, the company had to choose from among at least half a dozen ways to provide wireless broadband. Most of them are relatively immature, yet with Moore's relentless law still at work, they could all soon be superseded by some "next big thing" in radio technology. Capital costs, although only about one-fourth those of building a wired network, are nonetheless considerable. Getting the right licensed radio spectrum, in terms of both geography and frequency bands, is a challenge. But the biggest area of uncertainty for Jaring may be pricing these fast wireless connections, when monthly rates for comparable wired services are still in flux.

Yet the potential rewards are great, while the risks of doing nothing are greater still—and not just for Jaring. All over the world, Internet service providers face similar pressures. For example, in the United States, where spectrum is more difficult to acquire than in Malaysia, the second-largest Internet provider, Earthlink Inc., will be building a wireless network for the city of Philadelphia using Wi-Fi. This technology, which takes advantage of unlicensed spectrum, was never designed for municipal broadband. Unlike some newer methods, Wi-Fi can't clear trees, punch through tall buildings, or reach a base station kilometers away.

If it works out—if Jaring can weave together a winning broadband network using the airwaves—its success could encourage other Internet service providers in the United States, Taiwan, and everywhere in between. All over the world, Goliath phone and cable companies have snagged customers using their preexisting copper phone lines and coaxial cables—a competition in which, until now, David had no slingshot of his own.

To mitigate the risks, Jaring chose equipment—base stations and end-user devices—that can operate on a variety of spectral bands and should be able to weather the storms of technological change. The maker of this gear, Soma Networks Inc., of San Francisco, claims that if its clients want to switch over to third-generation (3G) cellular technology, the needed modifications can be made at modest cost. Large cellular carriers such as Verizon, SK Telecom, and T-Mobile are bringing 3G, and even so-called 3.5G, services on line. And if, a few years from now, Jaring wants to move to the up-and-coming protocol known as WiMax, a long-distance alternative to Wi-Fi that is still being refined by an IEEE standards committee, again, the network can be easily changed over.

Besides this flexibility, Soma also claims to offer better phone service than its competitors, shielding the needed bandwidth from other applications being run over the data network. For an engineer, telephony is just one of many uses for a high-speed digital network, but to a marketer, it's the most important—the one that people are accustomed to paying for and the one that will cause them to stay with, or change to, a service provider like Jaring. By encoding voice as data packets and sending it over the Internet (voice over Internet Protocol, or VoIP) instead of sending it over a dedicated circuit, as Telekom Malaysia does, Jaring can charge customers as little as 10 Malaysian sen, less than 4 U.S. cents, per minute, for domestic and even some international calls. For a few years at least, telephony is the killer app of wireless broadband, and it's a key reason Jaring chose Soma's equipment.