Mission Impossible For Broadband Providers?

Japan is witnessing a "rash" -- as one NTT exec puts it -- of broadband upstarts. But do any of these challengers stand a chance against NTT?

With the Nihon Keizai Shimbun recently predicting "a fully-fledged battle in the market for broadband communications," and some analysts predicting a telecommunications revolution in the next few years, broadband is something that clearly deserves our attention.

But before bowing down before this latest technological savior, we need to ask: Who are the broadband providers and what are they offering that we haven't seen before? And can they establish a beachhead in this closed system?

Take a look at Japan's broadband challenge.

BANDWIDTH
For years, governments, industry, and media pundits have been promising fast access to the information society from the comfort of your home. Unlimited streams of digitalized, interactive information through fat broadband telecommunications networks, allowing you to not just bank, shop, and surf on the Net, but watch movies and video-chat with friends and family on the other side of the world.

Although the promise of broadband was clear enough and helped to spark the imaginations of thousands of futurists, the details of how all this was going to happen were not. Who would hook us up? The telcos? Cable companies? Satellite? And the unanswered detail that loomed largest: Who was going to pay for the expensive high-bandwidth network to every door in the land?

In some parallel universe where technocrats run monopolies with unlimited budgets, the network is made of fiber-optic cable and built by the state telco. Japan is as close to that universe as anywhere. NTT controls over 95 percent of the local telecom market and has laid a vast infrastructure of high-capacity, fiber-optic backbone covering 70 percent of the country, according to Deputy Director General Kobo Inamura of the Communications Policy Bureau in the Ministry of Public Management, Home Affairs, Posts and Telecommuni-cations. The government's e-Japan strategy, announced this year, calls for the wiring of 30 million homes for high-speed Internet access by 2005, and it's clear that NTT plans to have a big hand in it.

Eric Gan of eAccess

Vernon Chamberlin of Jupiter Telecommunications

Keizo Kawamura of Tokyo Metallic

But in the real world, the state is drowning in debt, NTT has prioritized its corporate clients, and few think there is a chance of it providing door-to-door digital nirvana for Japan's 44 million households anytime soon. Getting fiber to what's known as "the last mile" will take years. Indeed some people, such as Shinji Moriuki, senior analyst at Daiwa Institute of Research, question whether it's possible to lay the entire country with fiber, given the costs.

As things stand, there are about 700,000 broadband users in Japan, the majority cable subscribers. The number is growing -- and fast -- but is still well below other advanced countries like the US and Sweden. What's more, most of Japan's 24 million home Internet users are stuck with a slow and expensive call-up service that is charged by the minute and can't be used when okaa-san is on the telephone. "Users are fast approaching the limits of their patience," claims Iwao Tojo, CEO of DSL provider Tokyo Metallic Communications Corp.

Tokyo Metallic is part of a motley crew of upstarts challenging NTT. A mix of cable TV companies, DSL providers, and private telcos, these challengers think they can take on the near-monopoly by offering Japan's long-suffering consumers something they've never had -- cheap, broadband Net access.

"These companies have the running for the time being," says Lincoln Owens, manager of the Digital Product Department at cableco Jupiter Telecommunications. "The window of opportunity for us is now."

Is NTT worried? Undoubtedly. But while some of the challengers bypass the NTT networks altogether and thus pose a pure threat, others represent a great revenue opportunity because they need the kind of telecom services that NTT is ideally positioned to offer.

"We've been slow in offering these kinds of services, so we can understand why there has been this rash of start-ups," says Tetsuya Kanada, executive manager of NTT-ME's broadband section. "But when you look at all the services needed to set up a broadband service, including the provider, applications, and the rest of the network, we very much welcome them. It's more business for us."

Still, with deregulation and increased consumer demand for faster Net access, there's undeniably unprecedented opportunity in Japan's mammoth telecom market, and for the past couple of years the new telcos and cablecos have been scrambling furiously to sign up people to their services. The projections look rosy. Seed planning Inc., a research company, predicts that by 2010, some 15 million people in Japan will be accessing the Net via cable and asymmetric digital subscriber lines. Jupiter Media Metrix estimates that about 22 million Japanese households will be connected to the Internet, with 8.6 million gaining access through broadband connections by 2005.

But what is a broadband connection? Before proceeding further, a clear definition is needed. The term "broadband" is used loosely, but it generally means a high-speed, high-volume connection to the Internet. Research firm Ovum defines business broadband as upward of 144 Kbps (kilobits per second), according to e-Marketer. So ISDN, which offers speeds of 64 Kbps (and is used by 30 percent of Japan's Net users), isn't really broadband. But ADSL, at 1.5 Mbps, is. Of course, it's all relative. What seems broadband today will seem pokey in 50 years. Or even in one year: A joint venture between TTNet, Tokyo Electron, Avex, and others will reportedly offer 40-Gbps service starting next spring -- about 16 times faster than conventional fiber lines and quick enough to transmit a 2-hour DVD movie in less than one second. The service will be aimed at businesses in central Tokyo (for upwards of ¥400,000 per month), but how long before such speeds reach consumers?

To make an educated guess on how broadband will develop, we need to see how things have evolved so far. It started with the old-fashioned copper telephone wires still in use today. The amount of information that can be transmitted via this conduit has traditionally been very limited (often as low as 8 kilobits per second). It also suffers from interference (the crackling sound on old phone lines), and the signals traveling over it fade over long distances, so they have to be boosted every 2 kilometers or so.

Think of copper wires as being at one end of a broadband spectrum. At the other end are fiber-optics. A fiber-optic "wire" is a flexible thread of glass that carries data as pulses of light. Depend-ing on the lasers used to send the data, each thread can carry many hundreds of times more information than copper wires. These threads (hundreds of which can be packed into a single cable) require less signal boosting (every 50 kilometers) than copper wires and don't suffer from electrical interference. Fiber has been used on the main arteries of the world's telecommunications system for some years and is gradually being extended to cities and towns. Japan's government has announced plans calling for high-speed fiber connections to the home (fiber to the home, or FTTH) with speeds from 30 to 100 Mbps. But it will be years before fiber can reach every home, and in the meantime the phone and cable companies that invested billions in the copper infrastructure are looking to recoup their investments.

The first solution to boosting copper's bandwidth was ISDN, or Integrated Services Digital Network (the "integrated" means that users can talk and surf at the same time). ISDN increases capacity to around 64 Kbps, which sounded like broadband when the technology got started. While most US companies balked at the capital needed to set up ISDN, NTT plunged right in. As a result, according to Inamura, today Japan has more ISDN users than anywhere in the world. Not a distinction to be proud of, it turns out.

Next came DSL, or digital subscriber line, which carries traffic much faster (around 1.5 Mbps) by employing different frequencies along the same copper wires. It also allows for simultaneous voice and data traffic, and, unlike dial-up ISDN, is always on. xDSL has taken off rapidly in the US and South Korea, but it's still in the early stages in Japan, with just over 100,000 users as of late April. (Not a bad number considering there were 19 subscribers in January 2000, but then neighboring South Korea, the world leader in this category, has well over 4 million DSL homes.)

For all the popularity of DSL in the United States, what's really been driving broadband there is the cable TV networks. US cable companies spent huge sums of money, often leveraged to the hilt, to build their networks. Using those networks to sell fast Net access is an important way to recoup that investment. These networks are made of copper coaxial "fatter cable" and backed up with fiber-optic backbone. Because they link directly to the home, they don't need the phone company's last mile -- a key advantage. "Cable led broadband in the US," says Kirk Boodry, senior telecom analyst at Dresdner Kleinwort Wasserstein in Tokyo. So far, it's the leading broadband technology in Japan, too.

Cable Bleeds the Way
Eight years ago, most telecommunications experts observing Japan would have laughed at the idea of cablecos bridging the last mile to the home and depriving NTT of its usual cut. Back then, the country's anemic cable industry -- fed for years on a thin gruel of small, low-profit, and scattered franchises -- hardly looked like it could support itself, let alone a capital-hungry new experiment like broadband. In the mid-1990s, cable in Japan had a household penetration rate of little over 5 percent, compared with 63 percent in the United States. (And much of that 5 percent consisted of "compensatory systems" designed to bounce existing programs over mountains and high buildings.)

But a lot can change in eight years, and cable has been fattened by injections of foreign capital and know-how. Despite evidence to the contrary, the government here has been slowly deregulating the communications industries. As early as 1993, the telecommunications ministry, anxious to attract US cable TV capital and know-how into the moribund Japanese industry, relaxed the restrictions on foreign ownership of cable systems to 33 percent from 20 percent and allowed cable operators to provide "full network services" -- meaning telephony and broadcasting. Crucially, they also allowed the formation of multi-system operators (MSOs), reversing restrictions that kept cable companies in their local areas. (A result of those regulations is that even today there are some 600 cablecos serving a country roughly the size of California.)

The changes were enough to start some consolidation and convince international players to take a shot at the Japanese market. Time Warner and US West teamed up with Itochu and Toshiba to form Titus Communications K.K., which later took Microsoft on board as a majority shareholder. Meanwhile, TeleCommunications International Inc. (TCI, now Liberty Media) hooked up with Sumitomo to form Jupiter Telecommunications Co., or J-Com. By 2000, Jupiter was already the largest MSO in the country. Then it got even bigger when it absorbed Titus late last year. As Hoovers Online puts it, "Other Japanese cable TV operators are mere asteroids compared to Jupiter Telecommunications."

With Jupiter leading the way, Japan's cable industry has been furiously laying its own network, bypassing NTT and signing up subscribers in densely populated urban areas. J-Com says that about 20 million homes in Japan are now "past cable" (meaning cable runs by the house) and that its own network accounts for about 6.2 million of them. The company has laid a fiber-optic backbone system around large chunks of the country, especially in the Kanto area, Osaka, Fukuoka, and Sapporo, dropping coaxial off the network and running it past houses. Tokyu Cable, in which Sony is an investor, is also a serious player, with the most subscribers for one system -- 180,000 paying subscribers for the one covering Meguro, Kawasaki, and Yokohama.

Other cable firms are getting in on the act. The telecommunications ministry says cable TV Net users numbered 625,000 at the end of 2000, quadrupling from 154,000 at the end of 1999.

Japan's cable players have the bandwidth, and they're going to use it. Jupiter intends to offer interactive services through its pipes and merge all its services into one bundled package. Plans include Internet over the television (instead of cable-to-the-PC Net access), t-commerce (e-commerce over TV), and even that old chestnut video on demand (VOD), which Jupiter says can finally be offered cheaply to Japanese consumers. Tomen Mediacom, which has shares in six cable TV stations and plans to take stakes in five more by the end of this year, is already offering an interactive TV service called Mediatti (see "Braving the Interactive TV Concept," page 56, February 2001). The company, backed by Pacific Century Cyberworks and operating in the Tokorozawa, Sayama, Kana-gawa, and Yokohama areas, is planning to roll out the service across the whole country by steadily acquiring controlling interests in Japan's hundreds of small operators.

The cable companies, then, are as close as anyone to delivering broadband bliss to Japan's consumers. But there are problems, of course.

Cable has massively expensive infrastructure costs, and to finance them cablecos must keep widening their TV service areas and signing on new subscribers. Jupiter says about 70 percent of their capital expenditure is on system construction, the result of high material and labor costs and the expense of negotiating and clearing regulations. In the US, the figure is 35 percent. As Ben Wedmore, software and Internet analyst at HSBC Securities, says, "It's all very well having cable going past 20 million homes, but you have to get customers in -- and keep existing ones on board."

To attract customers, cable needs content, and that can be a problem in a country with a small independent television sector. In other parts of the world, the content gap is often filled by foreign -- mainly US -- programming. But American TV simply isn't that popular in Japan. "One of our biggest sellers here, believe it or not, is NHK Sumo," says Jupiter's Owens.

Another problem is, naturally, competition. Fending off NTT, who the cablecos expect to come back strongly in the next couple of years, will not be easy. And the ASDL providers have entered the ring. Owens says Jupiter, for one, can take the heat. "We're in this for the long haul," he says. Jupiter president Vernon Chamberlin says the company will go capital-hunting this year with an IPO. Once that's happened, he says, things will get really interesting.

DSL: An Intermediate Technology?
Companies offering xDSL Internet access believe they hold the key to a broadband Japan. xDSL is the generic term for a range of technologies that soup up existing copper wire networks to provide high-speed digital communications. ADSL (asymmetric digital subscriber line) has downstreaming speeds of up to 1.5 Mbps -- ideally suited to Internet users who may be downloading a lot of information but sending little more than emails upstream. The DSL family includes asymmetric, symmetric, high-bit-rate, and very-high-bit-rate versions, operating at download speeds of 1.5 to 9 Mbps, 0.16 to 2 Mbps, 1.5 to 2 Mbps, and 13 to 52 Mbps, respectively.

The advantage of DSL for the new telcos who are pushing it is cost -- by using existing networks, they get more bang for their buck. No need to dig up roads and pavements, or negotiate with head-spinning construction regulations to lay thousands of kilometers of new cable. The disadvantage is that the telcos are dependent on NTT to lease their equipment, and that can be a frustrating business, even with recent deregulation forcing NTT to open up its central offices and let startup telcos make connections to the local switching system.

ASDL broadband technology, always on for a flat fee of about ¥6,000 a month, makes ISDN, a slow dial-up service that requires customers to have two phone lines, look dated indeed. But while it takes a mere two weeks to install an ISDN service, ASDL installation has been known to take a few months. The situation was bad enough for eAccess -- one of the DSL firms leading the charge in Japan -- to complain publicly that NTT hindered access to its central offices and took months to install equipment that should be done in a couple of days. Eventually the startup put Japan's Fair Trade Commission on NTT's case. "It's been a fight," says Eric Gan, chief operating officer of e-Access.

With $1 million of his own money in the company, the former telecom analyst's frustration is understandable. Established in 1999 with Gan and CEO Sachio Semmoto's seed money, and $45 million from Goldman, Sachs & Co., Morgan Stanley Dean Witter and others, eAccess saw enough during a free trial period from April to October 2000 to make them think they were on to something. The company started real service from November 2000 and claims to have hooked up more than 30,000 subscribers, mainly in the Kanto region. It's planning to reach out to Osaka and Nagoya this year.

eAccess has adopted a "stages" philosophy to providing Internet access: Instead of laying new infrastructure and then offering higher speeds to users years later, it first looks at what real-world users actually want today, and then proceeds from there. The formula works something like this: 1) reach the customers with access by using the most convenient technology available now; 2) provide content compelling enough to keep them loyal, such as a portal with networked games and movie clips; 3) construct a fiber network to cut down the access fees it pays NTT. The way Gan sees things, to begin with Step 3 is a strategy error.

"When people look at broadband, the first mistake they make is what technology to go for. When you come down to marketing, customers don't care whether the cable coming into their house is wireless, coaxial, fiber, DSL broadband, or anything else. What they care about is convenience and cost. Will I have to wait five years for this and pay through the nose? You can't tell the customer to wait five years for a fiber network."

The company's fastest-way-to-the-customer approach is in many ways opposite to that taken by the technocrats and national planners who want to lay the network first. For Inamura, DSL is "little more than an intermediate technology." The nationwide fiber network, he says, still has to be built.

"In the US," he says, "where the market is much freer, large parts of the country have few computers and little access to the Internet. Their market is dominated by a number of huge companies. Would it not be better for NTT to fight for more funding so that it can build a national fiber network, centrally funded with standardized technology?"

Gan says he formed his own company partly because he grew tired of waiting for NTT to open its network. And with Semmoto on board, the man who took on NTT once before and won -- he co-founded DDI Corp. in 1984 -- eAccess has a fighting chance. Semmoto knows very well how NTT works, having helped develop the same ISDN lines he's now competing against.

It was not eAccess, but Tokyo Metallic Communications Corp. that opened the DSL salvo in Japan. Metallic was the first DSL provider in Japan when it began service in late 1999. Led by CEO Go Hirano, a former NTT multimedia engineer also frustrated by NTT's glacial pace, the company started as a tiny venture with backing from Sanwa, JAFCO, and JP Morgan, among others. It was this venture that persuaded the then Ministry of Posts and Telecommunications to open up NTT's facilities to other operators in July 1999, after Chairman Hiroaki Kobayashi, who served on MPT's xDSL subcommittee in 1997, warned the bureaucrats that Japan was being left behind by the US and even (gasp!) South Korea. (See "Fast Country," page 32, December 2000.) It was a landmark decision that cleared the way for today's DSL challenge.

Metallic claims to have 20,000 subscribers, mainly in Tokyo. Hirano says the company will expand into Osaka (Osaka Metallic has just been founded) and Nagoya and is aiming for 1 million subscribers by April 2001. Like eAccess, Metallic says the advantage of its service is convenience and cost. Sub-scribers are sent an easy-to-install modem, so there's none of the labor cost of sending out a technician, the company claims.

According to Keizo Kawamura, a senior director at the company, Metallic faces two problems: "finance and NTT." The company is planning an IPO this year, but many analysts say that the money supply may well tighten up in response to the tech slowdown in the US economy. Indeed, Japan's fledgling broadband startups may want to avert their eyes from the bloodbath in the world's broadband capital, where DSL and cable providers have been falling like ninepins. Worse, the long-term winners are often proving to be the established telcos, who are snapping up the very companies that once challenged them.

NTT is not going to be left behind. Though it would rather not cannibalize its ISDN customer base, which it spent a lot of money acquiring, it won't hesitate to do so before others get the chance. NTT East and West started DSL service at the end of last year, after, according to Gan, "running negative campaigns" about the technology at first. Furthermore, NTT-ME started offering DSL shortly after Tokyo Metallic and eAccess -- much earlier than NTT East/West and "out of line" with NTT policy.

By the end of February, there were a little over 34,000 xDSL users in Japan. One month later, that figure had more than doubled to over 68,000 and was rocketing skyward. If such growth continues, Metallic and eAccess might have a shot, despite NTT muscling into the DSL market with its own offerings.

Usen: From Cable Radio to Internet Broadband
The fastest residential broadband service rolled out so far in Japan is not by a cable or DSL challenger, but by the Osaka-based cable radio company usen. The company, started in 1961, already has 220,000 kilometers of copper cable strung across the country's electricity poles, covering 98 percent of all municipalities. The firm can establish a nationwide fiber network by simply adding new lines along the copper ones, which it's already done in some parts of Tokyo.

In a fine example of Osaka chutzpah, which could yet backfire, usen laid its copper lines without asking permission from the electricity companies. What's more, it compounded the sin by illegally rebroadcasting radio programs over its network.

Not playing by the book has hardly endeared the company to NTT and the cable companies, and usen has faced problems getting partners willing to risk the wrath of the telco giant, though content providers Sony and Avex jumped right in. In a bid to placate the opposition before the company goes looking for money, current president Yasuhide Uno, who took over from his father in 1998, has agreed to pay for the use of the poles, and to compensate the utilities and broadcasters.

An NTT main distribution frame, the deregulation of which was indispensable to DSL challengers like Tokyo Metallic Photograph courtesy of NTT East Japan

usen is currently offering 100-Mbps fiber Net access ("the world's fastest Internet connection service") in Tokyo's Shibuya, Setagaya, Suginami, Ota, and Meguro wards. Plans call for expanding into all the major cities within the next three years, by which time the company hopes to have 1 million Net subscribers. These customers will be offered what usen hopes is another small stake in NTT's heart -- Internet telephony.

usen is the nearest thing NTT has to a fiber-to-the-home competitor and, to show it's not messing about, it's raising funds on Nasdaq Japan. (See "Grading the Exchanges," page 22.)

Like Jupiter, usen, which pulls in about ¥100 billion a year broadcasting music and other content over cable wires, plans to go head-to-head, or perhaps head-to-ankle, with NTT by bypassing the telecom giant's network and expanding its own. It's a risky, capital-hungry business that depends, again, on rapid expansion.

usen's high-speed claims have raised a few eyebrows, however, and some say that even if usen delivers fiber to customers' homes, the backbone network may not support higher speeds further upstream. Gan of eAccess is skeptical of the ambitious fiber plans. "There is so much existing copper in the ground already. When I hear of companies spending billions of yen to lay cable, I want to know, What are you going to charge the customer? If you charge ¥5,000, which is probably all this market will stand right now because people are paying less, not more, you need a hell of a lot of customers to make that capital back. We use the existing network and charge as little as possible."

usen, however, already has its own network in many places. There's no question that if the Osaka challenger acquires the necessary capital it will give NTT a run for its money.

NTT knows it: In mid-April, the telecom giant announced it would begin offering its own 100-Mbps fiber-optic service to certain areas starting in July.

The Power Companies
Power utilities are gearing up to make use of their own massive facilities for broadband. They've spent years building networks of poles and cables across the country and were forced to build hundreds of compensatory systems when their power lines interfered with TV signals. What was once a liability has become a broadband asset.

The power utilities have for some time been offering data communication services through their telecom affiliates. Collectively, the utilities have a nationwide fiber-optic network on par with that of NTT. If these utilities were to somehow consolidate their efforts, they'd be a credible challenger.

As it turns out, they're doing just that. In mid-April, Tokyo Electric Power (Tepco), Kansai Electric Power, and eight other electric power companies announced that they would integrate their data communications operations into PNJ-C -- a joint venture they set up in 1999 -- by this fall. PNJ-C will become the core of all the utilities' data communication services. The hope is that PNJ-C will become a major Japanese telco, on equal footing with NTT, KDDI, and Japan Telecom. Furthermore, Qwest Communications of the US will hold a 20 percent stake in PNJ-C, a relationship that will help the new telco offer more international services.

Tepco, which will be the biggest stakeholder in PNJ-C and has a network covering 1 million households in the Tokyo area, has been making some noise in the broadband space. In May, Tepco signed up to a joint venture with trading house Mitsubishi Corp. and affiliate Space Communications to distribute movies, music, and other content to PCs via satellite-based Internet connections. The venture will bypass NTT by using a satellite to bounce signals to cable TV operators and ISPs, who will in turn re-transmit to subscribers. Meanwhile, Microsoft, Tepco, and Softbank have announced the trial launch of 1.5-Mbps fiber optic-based home services in Saitama, with FTTH to come later.

The Race for the Last Mile
All these companies toiling mightily to bring broadband into Japanese homes is impressive, but success is not guaranteed for any of them. Theoretically, each could be out of business -- or out of the broadband business, or inside of NTT's belly -- within five years.

Furthermore, given the success of wireless technologies in Japan, there are questions about the merit of any fixed-line broadband solution. Will the Japanese, who have taken to mobile communications like fish to water, enjoy being chained to fixed lines in tiny homes? With 35 percent of Japan's 66.3 million Net users getting their daily fix on a mobile phone, according to IDC, there is room for doubt. Moreover, as hyperfast wireless Web surfing becomes a reality in the next couple of years, the rationale for fixed-line broadband becomes less clear. NTT DoCoMo and competitors J-Phone and KDDI are all planning to debut third-generation (3G) cellular networks this year, and industry experts claim it is only a matter of time before 2-Mbps wireless access becomes available, faster than anything currently on offer from DSL and cable. Meanwhile, fixed-line wireless networks, where the user taps into a broadband wireless LAN, is expected to grow rapidly over the next few years. Nomura Research Institute forecasts 720,000 such connections by 2005, for both the business and consumer sectors.

Perhaps wireless and fixed can coexist? "A fixed network does not have to be instead of wireless," says Andy Suzuki, an executive vice-president at Tomen Cyber-business Solutions. "Wireless can be used to supplement fixed and vice-versa. But for the time being at least, if you want media-rich content like video-on-demand and games, mobile networks are just too narrow."

Others wonder whether broadband is a technology looking for an application at this point. "Everyone has to ask, Where is the demand coming from?" says HBSC's Ben Wedmore. "Read the official reports and you'll hear that everyone wants video-on-demand and other bandwidth-hungry stuff like that. But I don't think there is much demand for it. What there is demand for is adult porn sites. You have to keep asking, What are we supposed to get from a computer that we don't get from the video store or from the movie theater or magazine stand? Why did the Internet take off? Email -- and you don't need 100 Kbps for that."

Andy Suzuki disagrees. "If you look at what is going to become popular in the next couple of years -- like interactive TV with people sending video-mail across the world -- you're going to need bandwidth for that. It's the next logical step from email, and bigger bandwidth is clearly better for the future."

The next couple of years will show who is right. Telecommunications is now the growth industry in Japan. After the bubble burst, from 1991 to 1996, for instance, the rate of increase for the number of telecommunications businesses in Japan was 153.0 percent, compared with 21.2 percent for construction, 8.7 for transport, and --1.5 for manufacturing. And that was before the current DSL, cable, and mobile boom.

With over a dozen new broadband companies already started, the market is becoming crowded. NTT is already fighting back, trimming the cost of its DSL lines and ISDN services, and moving forward with its own 100-Mbps fiber offerings. And if what BusinessWeek in April called the "telecom meltdown" in the US this year is anything to go by, things will get tougher for the broadband startups. Who will survive is anybody's guess, but most analysts say that although NTT may be worried about customers trickling away while it builds its network, it will still come out strong.

"NTT is going to dominate because on a functional level they still own the network," says Boodry. "The reason is not just access; you can get access to their facilities and have weak support mechanisms. And DSL is a niche product. NTT is leasing the lines, so they're still making revenue, and they're carrying all the backup traffic."

Some industry experts then favor the cable firms, which are building their own networks and offering broadband Internet access as only one of a range of multimedia services. But with its huge construction costs, cable eats up capital, and money may start to dry up.

So is it mission impossible for the broadband upstarts? Not impossible, but certainly an uphill struggle, says HSBC's Ben Wedmore. "I think the Internet will happen in a big way here," he says. "When you look at the venture capital pouring in, the level of technical infrastructure, and the government's backing, there's no question. But competition has to happen. We all cheer these small companies on. But will they have more than a tiny share of the market in a few years? That's just not clear."

Related Stories
Investor: "Braving the Interactive TV Concept," page 56, February 2001.
Feature: "NTT's Lowered Interconnect Fees: the Beginning of the Beginning," page 32, October 2000.
People: Fred Briggs, CTO of Worldcom, page 12, December 2000.
People: Hiroaki Kobayashi, president of Tokyo Metallic Communications, page 8, May 2000.
People: Hidehiko Yuzaki, founder of Acca Networks, page 12, May 2001.
People: Eric Gan (COO) and Sachio Sammoto (CEO) of eAccess, page8, June 2000.
People: Atsuki Ishida, CEO of Freebit.Com, page 18 (this issue).