Two Sides Go To War

The great 3G saga is rolling towards its next phase: network building. This requires a commitment to a particular technology standard, a controversial move that jabs at all 3G's sore spots.

By Jasper Moiseiwitsch

Sunday Communications is in a crunch. But Bruce Hicks, group managing director of the mid-tier mobile service provider in Hong Kong's ultra-competitive market, won't admit it. He serenely spells out his plans to take Sunday to 3G. He is rational. He makes sense. But he is troubled. Or he should be. Because like many worldwide, Sunday is a GSM operator locked into rolling out 3G through W-CDMA, a 3G technology that has so far proven to be expensive and risky unlike its more practical and profitable twin, CDMA2000.

Watching operators and vendors work with W-CDMA to date has been like watching a slow-motion train crash, and Sunday's Hicks reaches deep to find a positive spin for it all. "I'm very confident of the W-CDMA standard. There are just too many vendors, too many operators, too many billions of dollars being committed to it [for it to fail]," he says, bravely.

He's not the only one being brave. The CFOs of similarly pressured W-CDMA operators are fighting a war on three fronts. They need to wait for vendors to sort out their technology problems, but they need to roll out their networks quickly to start earning a return on their license investments, and they also need to find funding to build an entirely new network. There's no point in them going to the markets because investors don't want to hear about 3G. And they can forget about expecting fast, fat returns on the investment because the new networks will likely be full of bugs, there won't be enough handsets and, for now, they won't even be able to migrate existing subscribers onto the new 3G service because handsets don't work on both systems.

Code Division

W-CDMA and CDMA2000 are the two dominant 3G standards. The technologies specify how wireless signals are coded, what frequency they get beamed out on, how the base stations are built, et cetera. They are the wireless equivalent of the operating systems that run in desktop PCs.

Japan offers a real-time, real-world view of how both standards compare. In one corner, there is NTT DoCoMo, a global mobile leader that innovated i-mode, one of the most successful wireless products so far. DoCoMo also launched the world's first commercial 3G network, in October 2001. In the other corner, KDDI. From its position as the market's second-ranked mobile player, KDDI launched its 3G network half a year later, but has vacuumed up 3.3 million subscribers to its 3G service, compared to DoCoMo's paltry 142,000. On a daily basis, KDDI signs up 100 customers to its high-speed service for every one that DoCoMo gets.

KDDI's service is simply better and cheaper than DoCoMo's. KDDI's 3G handsets cost about half of DoCoMo's. Evan Erlanson, a telecoms analyst with Bear Stearns, says that DoCoMo's handsets cost wholesale about US$500-600, while the KDDI high-speed phones cost about US$240-400. KDDI's 3G phones work on the operator's 2G network; DoCoMo's don't, and have been recalled numerous times. KDDI's 3G service covers 90 percent of Japan, while DoCoMo's service is limited to Tokyo and parts of Yokohama, where coverage is spotty.

Not surprisingly, this has had an impact on the two carriers' bottom lines. In November 2002, KDDI reported a group net profit for April-September of 20.45 billion yen (US$169 million) exceeding a forecast of a 14.5 billion yen profit by corporate research firm Toyo Keizai. A few days earlier, DoCoMo reported a 95 percent drop in its net profit for the same period.

What has KDDI – historically less innovative, profitable and technologically adept than DoCoMo – done right? It launched its 3G service on CDMA2000. DoCoMo, meanwhile, languishes on the W-CDMA platform.

A Big Mistake

The real CDMA2000 success story is in Korea, where LG Telecom, SK Telecom and KT Freetel have collectively signed on about 7.35 million subscribers to high-speed CDMA2000 1X networks (1X is the first step of the technological migration from second to third generation mobile telephony standards). CDMA2000's popularity in Korea and Japan has prompted many to comment that it is a better technology than W-CDMA. The truth is that CDMA2000 is the better technology for operators that already use CDMA, a second generation (2G) standard. But most of the world operates on GSM, and GSM operators can't upgrade to CDMA2000 because they don't have networks or spectrum that are compatible with CDMA. The wisest 3G route for these operators would seem to be W-CDMA.

Apart from being financially risky, there is nothing wrong with W-CDMA's technology per se. It has the bugs associated with heavily engineered technology produced by multiple vendors, which needs to meet myriad international regulatory requirements. But W-CDMA's technology issues are part of a set of problems – part financial, part "historical" – and each exacerbates the other.

The W-CDMA story began in the late 1990s, when a working committee at the European Union level mandated that new spectrum licensing require the W-CDMA standard. The committee, which involved all the key European telco vendors (Nokia, Ericsson, Alcatel and Siemens), mandated a single standard to harmonize manufacturing efforts and to facilitate global roaming. The vendors decided on W-CDMA because its CDMA core utilized spectrum more efficiently than any GSM-derived standard could at that time.

These were all sensible decisions. Nevertheless, they set telcos up for a world of financial and engineering pain. The most obvious place to begin is the disaster that was the European 3G spectrum auctions. The Europeans started the auctions first, and saw the most exuberant and inflated bidding for 3G spectrum (about US$100 billion was spent on licenses in Europe alone). Nevertheless, the excessive prices (and related debt and share price decimation) are linked to W-CDMA alone. Tellingly, there is no CDMA2000 gear yet available on 2.1GHz spectrum, or the bandwidth that the Europeans auctioned for use in 3G.

The European licenses typically had tight deadlines for rollout of these W-CDMA networks. In any case, the steep fees paid for these licenses comprised a ticking clock in themselves: the net present value (NPV) of these assets depreciates with each day's delay in launching the networks.
This has created a pressing need in the W-CDMA world to roll out as soon as possible. Meanwhile, the operators and vendors are struggling to overcome massive engineering obstacles, many unique to the W-CDMA standard. Some vendors, usually an optimistic lot, are even talking of a meaningful W-CDMA deployment as late as 2005-06.

To begin with, there is the issue of handset shortages, which greatly upset DoCoMo's 3G launch in 2001. Because W-CDMA is an entirely new technology, chip manufacturers have not been able to finalize standards for the handsets that work on these networks. The key vendors of W-CDMA handset chips, such as digital wireless product manufacturer Qualcomm, won't go to high volume commercial production of these chips until standards have been finalized. "We're moving towards having finalized chip design," says Marshall Towe, Qualcomm's managing director for Southeast Asia, "but we don't want to get into full production making millions and millions of chips, then have the ITU [International Telecoms Union] or another standards body saying, 'we're going to change that a little bit'."

Because CDMA2000 is an extension of CDMA, it is a more mature technology and vendors have more easily managed their production needs. CDMA2000 vendors are using the same core standards as they have long-used with CDMA: same spectrum, same channel spacing, same basic technology. CDMA2000 chip makers are already working with established standards, and making their chips at full commercial volume.

Because most of the W-CDMA world is coming to the technology from a GSM standard, there have been very difficult compatibility problems to sort out. GSM is a fundamentally different standard than W-CDMA, and this creates problems. For example, W-CDMA handsets currently do not work on GSM networks, which is why DoCoMo's 3G phones cannot "roam" outside the limited W-CDMA service area.

W-CDMA is also a radical break from GSM. Without getting too technical, GSM is TDMA based, which is a fundamentally different way of coding data to CDMA. Accordingly, operators making the move from GSM to W-CDMA must build entirely new networks, which presents immense funding challenges. It is an extremely expensive and risky proposition, especially if a company has no idea what the consumer interest in the new service will be, or even how well the technology might work.

"With W-CDMA you are provisioning the whole network to an extremely high capacity. If you don't have demand until two to three years out, that becomes an NPV-negative investment for you," says Erlanson of Bear Stearns.

Because Asia came late to the auction party and because operators here paid less for their spectrum, regional W-CDMA aspirants are feeling far less pressure. Nevertheless, the Singapore licensees (SingTel, StarHub and MobileOne) paid US$55.5 million for their licenses, with no rollout imminent. Sunday in Hong Kong is paying US$6.4 million a year for its license – about one quarter of its annual profit – and it only plans to build its W-CDMA network in the middle of 2004. And then there's Hong Kong's Hutchison Whampoa, which came in big and early to the European auctions, spending about US$9 billion globally on eight spectrum licenses, with returns still hazy (see box "High Wireless Act").

Interestingly, Hutchison Whampoa chairman Li Ka-Shing's faith in 3G is not shared by his son, Richard Li, the majority shareholder and chairman of US$2.8 billion Hong Kong telecoms firm PCCW. As reported in CFO Asia last month (see November 2002, "Playing For Keeps"), the company recently sold its mobile phone arm CSL to reduce its debt burden. PCCW CFO Alex Arena says that even without the need to reduce debt, PCCW would still have hived off its mobile assets. "The future of 3G technology isn't clear, so we'd rather wait on the sidelines and see what develops," he explains.

Meanwhile....

CDMA2000 operators don't have quite the same problems, although there is concern that Qualcomm, which invented CDMA, controls too much of the intellectual property rights (IPR) contained in the standard. But most importantly, the CDMA2000 camp has largely avoided buying new spectrum and hasn't had to pay for new licenses. This is essentially a historical fluke. Most of the new 3G spectrum licenses already awarded have specified the W-CDMA standard. Almost all the European regulators required this standard and, in Asia, Malaysia and Singapore did too.

With no new spectrum available the CDMA2000 vendors focused on ways to squeeze more capacity out of existing spectrum, and have been very successful at it. Korea's SK Telecom says that it wrung 50 percent more voice transmission out of existing spectrum when it upgraded to CDMA2000 1X. 1X promises to transmit data at rates of 307 kilobits per second (kbps). In reality, subscribers to Korea's high-speed CDMA2000 networks are getting data at about 40-66kbps – slow by 3G standards, but comparable to a household's dial-up internet connection.

The next upgrade, called 1X EV-DO, promises even greater payoff. This system can realistically transmit at one megabit per second (mbps), which is more capacity than anyone has been able to use to date.

The CDMA2000 operators also have fewer compatibility problems. They are coming to the technology from CDMA, which forms the core of CDMA2000. Unlike the transition from GSM to W-CDMA, CDMA2000 offers a clearer, cleaner upgrade path. CDMA2000 works on the same spectrum as CDMA. Much of the CDMA infrastructure can be retained during each upgrade. And, because operators are using the same spectrum and the same core technology, CDMA2000 handsets are backwards compatible – they work on 2G CDMA networks.

The upshot has been that CDMA2000 operators have been able to migrate to 3G more cheaply, with fewer technology hassles and with a more risk managed approach than that of the W-CDMA camp.

CLSA research estimates that it costs about US$100-200 per subscriber to upgrade to CDMA2000 1X. Given that KDDI reported in June that it earned about US$20 per month more from the subscribers its upgraded to 1X services, it seems that 1X offers a fairly quick and painless return on investment.

More importantly, capacity on a CDMA2000 network can be deployed incrementally and with discretion. For example, if the operator knows that there are certain parts of a city that have higher capacity needs than others (like the central business district) it can focus bandwidth in these areas. From a service perspective, it does wonders to be able to focus capacity when and where you need it. Crucially, from a financing perspective, it lets the operator control spending on upgrades as and when demand arises.

Nokia, a key W-CDMA vendor, argues that W-CDMA also has an evolutionary path. Many companies have upgraded their GSM networks to GPRS and Edge, which are the "2.5G" standards of the GSM world. The difference is that these investments are largely thrown out the window when the operator goes to W-CDMA. W-CDMA is not GSM based, does not work on GSM bandwidth and very little of the investment in GSM 2.5G is accruable to W-CDMA.

The Rebuttal

Nevertheless, those committed to W-CDMA still find reason for hope. Given that GSM operators have about 66 percent of global market share, and that W-CDMA is the chosen 3G standard for GSM operators, many expect that W-CDMA will wind up with a much larger market share than CDMA2000. More market share means more vendors and production, which should mean more competition, greater product diversity and lower prices.

"Although CDMA2000 has a much bigger presence than we anticipated and it will continue to grow, there are literally billions of dollars being invested in W-CDMA and there are many more operators committed to W-CDMA throughout the world," says Hicks. "That ultimately translates into lower cost of equipment and lower cost of terminal products [handsets]," he adds.

If there is one downside to CDMA2000, it is that the standard's inventor Qualcomm controls too much of the IPR, and retains many of the essential patents. "Generally there is a feeling that Qualcomm is heavy handed [in its control and licensing of CDMA-related technology] since it is the only one that owns the patents on CDMA. It becomes a monopoly. And any criticism that you would hear about any monopoly is valid here too," says Rohit Sobti, a Hong Kong-based telecoms analyst with Salomon Smith Barney.

Sunday's Hicks agrees, adding that Qualcomm's near monopoly over CDMA2000 encourages operators to go with W-CDMA, where the IPR is spread out among different vendors and control is more diluted. "It [Qualcomm] has a bigger ability to sell with licensing, royalties [for CDMA2000] than it does out of W-CDMA," he says. "There is a bigger concern that [Qualcomm could exact monopoly pricing] because you have a more predominant factor with Qualcomm in CDMA2000."

Code Dependent

Qualcomm's response to this is that its greater interest is to promote the CDMA standard. It can only do this if its licensing fees are reasonable, and if it can win the trust of vendors. Furthermore, the company claims that it gets equal royalties and licensing income from both W-CDMA and CDMA2000, and has no agenda to promote one over the other (this point is disputed by others).
"We love all our children equally, we like all flavors equally, as long as it's CDMA," says Qualcomm's Towe.

Beyond all this, there is an ongoing raging debate about the technical superiority of each standard. Each side presents complicated arguments for both, but these quickly dissolve into engineering detail. Analysts and other independent commentators generally shrug these discussions off, noting that both standards work quite well and to more or less equal levels of efficiency.

But the relative engineering strengths of both standards is not the point. The problem is not W-CDMA's technology per se, it is W-CDMA as a whole: the spectrum fees, the compatibility problems, the requirement for a wholesale upgrade of networks, and how these problems interact and compound one another.

Some day, when the handset technology develops, consumer tastes for high-speed mobile data emerge and everyone has been weaned off their legacy 2G phones, W-CDMA-based services will prove as sound a business proposition as anything offered by CDMA2000. In the meantime, however, the W-CDMA world will twist on the winds of half-formed technology and uncertain funding.