The $1,000,000,000,000 Bet Someone's going to get rich delivering broadband. Someone's going home in tears.
By Stephanie N. Mehta

(FORTUNE Magazine) – There's an old saying at the poker table that if you don't know who the sucker is, it's you. Well, there are more than a few suckers out there betting on broadband, you can be sure. The world is just waiting to find out who they are.

Some of the largest companies anywhere are gambling that hyperfast, constant Internet access will be the next great staple of consumer life. From software firms pushing new services for super-juiced networks to entertainment conglomerates looking to stream out their products online, these businesses just can't seem to drop their cash fast enough. That's a gutsy play, but experience would suggest that if ever there were a time to hedge, this is it.

Of course, none of this potential would even exist without someone's delivering "broad bandwidth" to consumers, and among those with the most to gain--or lose--are the communications companies hoping to do just that. AT&T, for example, has spent $100 billion acquiring cable television lines in a bid to sell continuous streams of voice, video, and Net services. The Baby Bells, too, are spending billions to turn their aging copper phone lines into fat data pipes. In Britain, cellular companies have pledged $35 billion to license airwaves for a new kind of speedy wireless service. Carriers have committed another $46 billion for similar wireless spectrum in Germany, and they'll have to pony up billions more for licenses elsewhere in Europe. Even the $165 billion proposed merger between AOL and FORTUNE's parent, Time Warner, was partly driven by AOL's desire to acquire the broadband links controlled by the country's second-largest cable company. Indeed, our back-of-the-envelope calculation indicates that, starting with the first serious deployments of fast online access a couple of years ago and projecting to the completion of next-generation wireless networks, the investments stack up to over a trillion bucks. Easily.

One look at the characters behind the money makes clear how serious this thing is going to get. John Malone and Craig McCaw, for example, are backing startups that aim to deliver broadband services via satellite. Paul Allen owns a huge chunk of RCN, a relative newcomer that is building a new network that bypasses the incumbent phone and cable operators' lines. John Sidgmore is vice chairman of WorldCom: "Think about how much capital is going into broadband, both financial and intellectual," he says. "This is where all the action is."

But investing isn't winning. As with any new technology wherein lots of players chase a limited pool of would-be subscribers, someone's bound to go down. Hard. As we've seen in the wireless market, competition and pricing pressure may benefit consumers, but it's starting to take a toll on providers. That could be a bad omen for the broadband players.

And so the race is on to connect the world. As with most Internet gambits, being first appears to be a huge advantage: Once you've grabbed a customer, you're likely to hold on to him. Take Gary Chavez, a manager at a technology consulting firm. Earlier this year Chavez signed up for Pacific Bell's high-speed service--not out of any loyalty to the phone company but because it was the only choice in his San Francisco neighborhood. After waiting for several weeks to be set up, Chavez is happy enough with what he got, and he's even thinking about shelling out more money for a higher-speed connection from Pac Bell. "I wouldn't switch and, say, go with a cable modem," he says, referring to the local competition. Even when people aren't 100% satisfied with their superfast hookups, few, it seems, will go through the hassle of switching. For providers, that kind of loyalty--even by default--could mean all the difference.

It's getting those customers in the first place that will be tough. A scant 5% of residential customers now have some form of broadband, primarily through cable operators. No one, including us, knows which technologies will prevail, but the shakeout is going to make for some high drama. "It's so early," says John Zeglis, head of AT&T Wireless, which has just begun selling high-speed Internet access using wireless spectrum. "I don't think anybody falls terminally, fatally behind at this stage of garnering new customers." So at what stage do the fatalities begin? Stage two? Three? Nobody wants to be the first to find out.

The Paul Allens and Craig McCaws of the world are going to have to take the country a block at a time. They'll have to build their networks at mind-boggling expense--only to be confronted, in the end, by still another variable: Does anybody care? Much of this issue of FORTUNE is devoted to the various applications broadband Internet access might give rise to, and to the underlying technology. Businesses are certain to be lucrative customers. But it remains a very open question just how eager for broadband the rank-and-file consumer will be.

To help readers better understand the challenges these companies face--and just what, exactly, they're peddling--we've decided to survey the various broadband platforms under construction: cable, DSL, fixed wireless, and fiber-to-the-home. Keep your eyes peeled--and if you spot the sucker, please let us know.

Mike Armstrong is trying to reinvent the company that invented the telephone, using a rival industry's technology.

As chairman and CEO of AT&T, Armstrong has spent $100 billion to acquire two cable companies, giving Ma Bell direct connections to about 28 million homes. He'll use those connections to serve up a growing array of entertainment channels, local-phone services, and, of course, online access. These efforts are meant to offset declines in AT&T's long-distance voice business.

Most cable operators began experimenting with high-speed data services in the mid-1990s, but cable then was built to broadcast information in only one direction; connecting it to the Internet meant upgrading networks to accommodate two-way traffic. Today, broadband cable is actually a mix of coaxial cable and fiber (a.k.a. hybrid fiber-coax). You connect your computer (or, soon, your TV) to a special cable modem; data travel from your house along a coaxial cable that runs to a neighborhood node, where it connects with fiber. From there data travel to a "head end," where signals are processed for fast transmission (or, conversely, delivery).

There's no question that cable operators such as AT&T have a strong lead over their phone-company counterparts. By the end of the year, there will be about four million cable-modem subscribers, compared with 1.7 million residential users of the telco alternative, the digital subscriber line (DSL). Cable services boast of having faster speeds than the other mainstream residential solutions--DSL and fixed wireless--often telling customers to expect downloading speeds of about three million bits per second, which is about 50 times faster than a 56K dial-up modem. In real life, however, many homes in a neighborhood share the coaxial part of the network: If many customers are using their cable modems at the same time, the network slows down, especially in the outgoing direction, or uplink. Needless to say, AT&T's phone-company rivals point to cable's shared architecture in promoting their own services, and tout their "dedicated" connections as more secure and dependable. (AT&T counters that it can easily boost the capacity of its cable networks by adding more equipment at the node.)

No other broadband effort has received more scrutiny than Armstrong's. Critics have called his strategy both desperate and rash; many question whether cable networks can be effectively revamped to handle the large volumes of phone and data traffic AT&T would generate. But Dan Somers, head of broadband services at AT&T, says the company's cable acquisitions were anything but impetuous. "We did a significant amount of work looking at capability and reliability and costs," he says. "The conclusion was straightforward: Hybrid fiber-coax met the criteria we laid out better than any other form."

AT&T says it is spending an additional $2.5 billion to upgrade its cable lines for telephone and broadband services. When a customer signs up for one of these add-ons, AT&T then spends another $750 to equip each home with special gear. Eventually it hopes to drop those extra costs to about $400 a customer.

With customers paying about $60 a month for local phone service and $40 a month for broadband, analysts believe the company can get a rate of return of 25% to 30% on its cable properties by 2004. This calculation assumes AT&T will hook up about a third of the homes in its cable territories with local phone service and 25% of homes it serves with high-speed Internet access. Yet most experts expect that only 25% to 30% of homes nationwide will have broadband by then. That means, on average, that AT&T will have to capture nearly every broadband customer in its service areas, leaving scarcely a crumb for its competitors.

While the company maintains that it has the superior technology for broadband, AT&T and the other cable operators understand that they need all the help they can get. That's why television is such a big part of the cable operators' strategy: If they can push fast Internet services directly to TV sets (they say a full-blown set-top box could appear as early as next year), they'll have a big advantage marketing to Internet neophytes and technophobes.

Similarly, the cable crowd understands that many customers will buy in based purely on convenience. Dan Somers says this is his dream experience for consumers: "They walk into a store, buy a device--fully load-ed--bring it home, plug it in, and they're ready to rock & roll." Broadband isn't there today, Somers says, "but we're working on it."

Joseph Nacchio is, as usual, talking fast. But instead of chattering about Qwest's modern fiber-optic network (a favorite topic) or its market capitalization (another fave), the CEO is racing through a somewhat less glamorous subject: copper telephone wires.

This isn't a total shock. When it completed its acquisition of local telco US West earlier this year, Qwest inherited a network with millions of miles of copper connecting customers in 14 Western states. US West was one of the first Baby Bells to roll out DSL, which transforms the phone companies' traditional "twisted pair" of copper wires into a broadband pipe. Digital subscriber lines were developed by scientists a decade or so back, but in the pre-competition, pre-Internet days there wasn't much use for them. A few years ago, US West took the lead in resuscitating DSL as a consumer Internet service; the Bell was so aggressive that it actually convinced networking giant Cisco to buy a DSL gear maker to help speed deployment.

Virtually every home in the U.S. has a dedicated copper line that links it to a phone company central switching office. Using a special modem plugged into a regular phone jack, DSL ships data and voice traffic across your phone line to an ominous-sounding box at the central office called a DSLAM. The DSLAM--a relatively cheap piece of equipment--pumps up the capacity of the copper phone line to the customer and routes traffic between the Internet and the home. Voice traffic is stripped off and travels across the regular network, meaning customers can surf and talk at the same time.

At first blush, the ubiquity of the telephone in the U.S. would appear to give DSL a huge advantage over cable, which is actually used in only about 70% of homes. But DSL doesn't reach everybody. Customers who live more than three miles from the central office, for example, can't get the service; the DSL signal degrades if it has to travel long distances on copper. Some newer phone lines don't work well with DSL. All of which creates a bottleneck in ordering service because the carrier needs to figure out if the customer's line is even qualified for the installation.

The phone companies are trying to work out these kinks. Local phone giant SBC is spending $6 billion on DSL, and a big chunk of that money is being used to install fiber-optic cables and remote switches in far-flung neighborhoods, essentially shortening those copper loops. Under SBC's plan, it expects to reach most of its consumers with download speeds of 1.5 megabits per second; many customers will be able to access the Internet at speeds of six megabits per second--enough for really bandwidth-hungry applications such as downloading movies. Typically, though, customers in most parts of the country can expect much slower speeds--about 640 kilobits per second downstream and about half that speed upstream.

The big advantage of DSL is customer acceptance. Survey after survey shows consumers would rather buy their high-speed connection from the phone company than from the cable company. Meanwhile, the phone companies' upgrades and equipment can be used to sell to business customers that typically pay more than consumers. (Consumers generally pay about $40 a month for DSL service.) Phone companies now spend about $1,400 on average to add a customer--about double the cost to cable operators--according to research by Sanford C. Bernstein and McKinsey & Co. But with costs going down, plus the boost from business customers, the researchers expect phone companies to show a positive cash flow on DSL by 2001.

Back at Qwest, Nacchio isn't saying exactly how much he'll spend on DSL. During the merger process, US West's DSL sales slowed and some analysts wondered whether Nacchio, a swashbuckling sort, would halt the project in an effort to cut costs. But when, a few weeks later, he announced plans to slash 11,000 jobs, Nacchio also said he was slightly raising capital expenditures on a bucket of projects, including DSL. He vows that the company, which has a mere 175,000 subscribers today, will end the year with 250,000 DSL users and add another 250,000 in 2001. (In contrast, much larger SBC has promised to hook up a million DSL lines by the end of the year.) "I think DSL is the broadband solution that's going to dominate," Nacchio says, barely pausing for air. "There's more copper in the ground than anything else. You've got a natural highway in the ground to the consumer, and that's a big deal."

Qwest may ultimately have less riding on DSL than some of its Bell brethren. The company, which started out as a wholesale provider of long-distance services and later served mostly businesses before buying US West, has a much broader portfolio of operations. For example, it recently announced a new division aimed at storing and delivering broadband content; it also operates a unit that stores and delivers software applications. All these businesses push traffic along Qwest's core fiber-optic network. "We're broadband junkies, but we also like a fiber-rich diet," laughs Nacchio.

Bill Esrey, not surprisingly, is an unabashed fan of wireless. The Sprint CEO oversaw the carrier's construction of a new, nationwide cellular network that has become the phone company's most prized asset. But on a sweltering summer day in late August, Esrey is sitting in his Westwood, Kan., office enthusiastically describing a different wireless: fixed wireless.

Using a swath of spectrum previously allocated for broadcast entertainment, Sprint has begun selling two-way broadband services in Phoenix and Tucson and it plans to be in 12 cities by the end of the year. The technology, multipoint multichannel distribution system, or MMDS, allows customers to send and receive fairly robust streams of data between a central broadcast tower and small antennas affixed to their homes. Information from the Internet flows across fiber to Sprint's local head end; from there it is routed--sometimes via fiber, sometimes via microwave--to the big central tower, where data are transmitted to the intended home antennas. A cable links the home antenna to a special modem for the service.

Esrey, a former investment banker, seems particularly pleased when he describes the economics of fixed wireless. Sprint was able to quietly snap up MMDS licenses covering about 30% of the country for about $1.2 billion. And once the company installs a central tower and high-powered computers in each market, it needs to spend capital only when a customer signs up. "We looked at all the models, and this seemed to us to be the most effective way to get to customers in the end," he says. Sanford C. Bernstein estimates that companies such as Sprint currently spend about $1,000 to add a new MMDS customer.

Esrey is so fond of MMDS that he would like to be able to roll it out in other cities. The problem is that WorldCom--which had to abort a proposed acquisition of Sprint when regulators balked at the No. 2 and No. 3 long-distance carriers' combining--controls licenses covering another third of the country. So even as AT&T has developed a residential fixed-wireless service to deliver broadband to customers beyond reach of its cable operation, Sprint has been forced to improvise by leasing DSL lines when its wireless signals fall short. That's a lot less attractive to Esrey because Sprint doesn't control those copper networks.

Of course, MMDS has its own challenges. The signal between the tower and the home antennas can't travel through or around buildings; even trees and hills can flummox the system. That is why Sprint is rolling out the service in flat cities such as Phoenix. Analysts also worry that the uplink channel isn't very wide right now. The consumer service downloads at up to two megabits per second but upstreams at around 256 K. And, like the cable system, wireless airwaves are a shared resource, and multiple users on the system can slow performance.

Esrey doesn't seem too worried about these issues. MMDS is so new that it could be a good long time before Sprint's local networks are strained for capacity. And when it comes to assessing the competition, Esrey is equally sanguine. A true wireless believer, he doesn't think cable and DSL will ever be a threat down the road. Rather, he thinks fiber will remain the main conduit for serving big business customers and for moving traffic across countries and under oceans. Consumers and small businesses will get their broadband via the airwaves. "Twenty-five years from now, you'll be left with two things," he predicts. "Fiber and wireless."

European wireless executives such as Vodafone CEO Chris Gent aren't waiting 25 years. Gent believes many consumers' first experiences with broadband will involve neither wires nor computers. Instead they'll cruise through Web pages and e-mail on some sort of handheld device, no strings attached. It is this vision that helped drive Gent's $180 billion acquisition of Mannesmann, a deal that gave Vodafone the massive scale it will need to absorb the cost of building out a broadband wireless network in Europe. But Gent's got plenty of company. Rivals France Telecom and Deutsche Telekom also appear to be laying claim to pan-European wireless data networks. In fact, the European version of the broadband land grab involves almost no land at all. Some cable operators in Europe are rolling out high-speed services, but the big European telcos all seem to be betting on unplugged broadband.

At first the speeds will be well shy of those promised by their wired counterparts. But in just a few years a new era of networks, dubbed 3G, as in "third generation," promises to deliver the wireless Web at 2.4 million bits per second--about 42 times the speed of a 56K modem.

A nascent form of mobile broadband is already cropping up in the U.S. Companies such as Metricom and MobileStar have begun offering always-on, fast links via unlicensed wireless spectrum. Metricom, which sells its Ricochet wireless service through companies such as WorldCom and Juno, now offers 128K-plus wireless connections to the Internet in San Diego, Atlanta, and seven other cities. MobileStar, which requires a special wireless Ethernet card, installs its radios in busy locales such as hotels and frequent-flier lounges. MobileStar's service is meatier--the company says users download data at about 1.5 million bits a second--but it isn't 100% mobile; you can't run around town with your laptop tucked under your arm and maintain an uninterrupted connection to the Internet.

But with both services you can start to see why mobile broadband is so compelling. Some of the possibilities are pretty fun: With a broadband connection, you might download a movie before your airplane leaves, then watch it during the flight. On a more practical note, business people could tap into their corporate networks at any time and replicate anything they could do at the office. Already, wireless broadband is becoming a substitute for hard-wired Internet access. Mark Goode, CEO of MobileStar, says he has talked to Dallas consumers who arrive at the airport early just to take advantage of the broadband connection at the Admiral's Club (for more, see Personal Fortune). Timothy Dreisbach, CEO of Metricom, says he believes wireless broadband could easily eclipse cable modems or DSL based solely on convenience. With the Ricochet service, there's no need for a technician to install special gear, and you're not plugged into the wall.

Bob Metcalfe, Ethernet inventor and pundit, posed a simple question at a tech conference a couple of years ago in response to complaints about cable modems and DSL. Why, he asked, isn't anyone deploying fiber directly to the home?

At least one company is., a small Houston company, works with real estate developers to install fiber-optic strands directly in new homes. It is wiring about 60,000 houses in Texas and plans to expand to Phoenix and Las Vegas. "Fiber-to-the-home is the holy grail of broadband," declares founder and CEO Michael McClere. "It is definitely where the market is headed." ClearWorks says its service delivers a whopping 100 megabits of bandwidth. The company also uses the fiber to deliver local phone and cable television services.

Fiber-to-the-home is inspired by the networks used by many businesses. Fiber strands travel from each house in a neighborhood to a local distribution box, which consolidates them for delivery back to the phone-company central office for routing to the Internet. There's no question that fiber is the fastest, most reliable, and probably most future-proof pipeline around. But phone companies have struggled for years to figure out a low-cost way to deliver it.

It isn't the price of the actual glass fiber that's too high; it's the optical components and other equipment that make fiber-to-the-home prohibitively expensive. But McClere says he has developed proprietary technology that allows him to deliver hyperfast fiber service for $1,500 per customer. That's much higher than the cable-modem and MMDS services, but it's comparable with the average per-customer cost to the Bells of hooking up DSL.

Of course, ClearWorks is building new networks (in middle-class neighborhoods), and because the company works with developers, it doesn't have to tear down walls or dig up any rosebushes. The strategy is paying off: McClere says ClearWorks wins the business of 99% of customers in the communities it serves. That's a great penetration rate, obviously, but ClearWorks may have a hard time getting any economies of scale.

Even some phone companies are considering stringing fiber to homes when they have to deliver service to a new community. BellSouth, for example, has launched a trial of fiber-to-the-home in an Atlanta suburb. Carriers in Germany and Japan are performing similar experiments.

Some naysayers question whether any residential customer actually needs that much speed. McClere says consumers will think of uses, such as the Houston customer who is using his spare bandwidth to run a dial-up Internet service company out of his home. "The majority of telecommunications companies are just building for today," he says. "We're building for tomorrow."

That's the amazing thing: Carriers have just begun backing broadband, and already companies such as ClearWorks appear to be upping the ante. That could mean another round of investing even before carriers digest their first trillion in spending.

Vendors are already salivating over the possibility of more network upgrades. McClere says every equipment maker he meets is building components for costly fiber-to-the-home systems. Same goes for gear to jazz up DSL and cable lines. "In the labs, the guys who wear white coats have shown you can get up to 30 megabits out of DSL and cable," says WorldCom's John Sidgmore. "Twenty-five years from now, DSL will have broadband capacity that's 20 to 30 times what it is today."

In all this there is an unflagging assumption that consumers will want Internet access at speeds much, much faster than they can get with dial-up modems. Providers are undeterred by the fact that there aren't that many consumer applications yet that can sop up 30 megabits of bandwidth. "Let's kick back and think what the proposition is," says AT&T's Zeglis. "You can have access to all the information in the world and communicate with all the people in the world. There's no chance that this isn't going to change the way people live." The industry is betting a trillion dollars on it.