AT&T’s Wireless Debacle

Mr. Gilder and Mr. Vigilante produce the monthly Gilder Technology Report.
Everyone from Alan Greenspan to Al Gore understands that prosperity now feeds on the explosive advance of new technologies. But most fail to acknowledge that business success in the new era will require mastery of technological paradigms and performance. Ordained by the towering sage of Caltech, Carver Mead, the prime rule is: Listen to the technology, find out what it is telling you. Companies and nations that fail to observe this rule will pay a serious price.The new AT&T Wireless company, which just accomplished its initial public offering, is the epitome of an antitechnology high-tech company whose leaders want to tell the technology what to do. In 1993, AT&T paid Craig McCaw some $25 billion, including debt, for the most dynamic firm in the U.S. cellular industry, albeit one suffering from obsolescent technology. Now, without upgrading the technology, AT&T executives are sticking the company on the public for a market cap of some $65 billion (including $10 billion in new cash), having driven its U.S. market share down almost 30%. Since wireless was growing faster outside the U.S., AT&T’s real world-wide market share has dropped even further.

In the fastest-growing industry in the world, AT&T’s stock has run some 50% behind the Standard and Poor’s 500.

Meanwhile the company doggedly opposed a U.S. innovation called Code Division Multiple Access, which could restore America to wireless leadership in the Internet age. Unless the company changes radically now, AT&T Wireless will become a low-tech wasteland.

Not Poised to Compete

The next generation of the Internet will be wireless. But AT&T Wireless isn’t poised to compete. Led by lawyer John Zeglis, the company shows little or no grasp of the technology imperatives of the Internet era. The danger is that the undertow from bad technology at AT&T can undermine all U.S. cellular. Is it too much to ask that AT&T be some kind of technology leader, rather than an obtuse establishment that would rather go down with the ship than concede that it has been wrong for seven years?

The wireless company’s one chance is AT&T chairman Michael Armstrong, who has shown flashes of navigational brilliance while piloting the world’s most unwieldy nongovernmental barge. But he must use the $10 billion in cash raised last Thursday not to protract the death throes but to swiftly reverse a decade’s worth of technological missteps.

In the early 1980s Mr. McCaw, then owner of one of the nation’s larger cable TV providers, sold out of cable to enter the cellular phone market in its infancy. By 1993 he had created the nation’s largest wireless phone company. AT&T’s brand, marketing muscle and acquisition warchest were supposed to transform this lead into unshakable dominance. But none of these advantages could overcome AT&T’s tone-deafness to technology.

As early as 1990, analog cellular was already a technology in crisis, transfixed by the painful paradox of wireless economics: In an era in which the dominant force in the world economy is the explosion in bandwidth — communication speed and capacity — wireless systems still face severe bandwidth constraints. The way out was digital — allowing wireless networks to harness the power of then-emerging “digital signal processor” microchips to “change wireless from a radio business into a computer business,” in the words of Donald Steinbrecher of the Massachusetts Institute of Technology. Digital would both enable the compression of voice signals into relatively compact coded bit streams and allow multiple users to share single channels. By the time Mr. McCaw was ready to make the jump, two digital technologies beckoned. Like most of his colleagues and competitors at the time, Mr. McCaw made the wrong choice.

Seduced by alluringly concessionary terms from European mobile system maker Ericsson, Mr. McCaw and his chief technology officer, Nick Kauser, opted for a Time Division Multiple Access system. TDMA divides a single-standard 30-kilohertz channel into six time slots, each measured in milliseconds, allowing the interleaving of several voice streams onto the single channel. In combination with other efficiencies, TDMA promised to multiply by three to six times the number of simultaneous users who could occupy a single cell.

The rejected alternative, Code Division Multiple Access, pioneered by Qualcomm, seems by comparison a fantastically demanding and complex technology. Designed to allow every user in the cell to share the same frequency simultaneously without dividing channels into rigid time slots, CDMA works by combining an elaborate coding scheme with difficult dynamic power regulation of mobile units from the cell base station. Mr. Kauser denounced it as a flagrant violation of the laws of physics. But his respect for nature’s law concealed an underappreciation for Moore’s; with processing power doubling every 18 months, the once unimaginable gyrations of CDMA processing could be performed by silicon signal processors that never broke a sweat.

Even so, academic theorists long maintained that for voice CDMA is no more efficient than TDMA — within any given cell. That might have been true in theory, but in practice CDMA is three times as efficient as TDMA for voice while using only 1/20th as much power to transmit. TDMA has tried to close the gap, unsuccessfully, and often at the cost of compromising voice quality.

Stealing away AT&T’s market have been CDMA companies such as Sprint PCS and Vodaphone/Bell Atlantic, whose joint venture will become the market-share leader with double AT&T’s subscribers and revenues.

For AT&T, the worst is yet to come. Inferior as TDMA is for voice, it is essentially worthless for wireless data, the sine qua non of the coming third-generation wireless networks. Data, typically transmitted in brief, variably sized, bandwidth-hogging bursts, plays to CDMA’s dynamic strengths. By contrast, TDMA is incapable of carrying Internet data at any reasonable speed.

By the end of this year the U.S. CDMA companies will begin upgrading to a system that will not only double their voice capacity on the same spectrum — thus increasing their conventional superiority to AT&T’s network — but will also carry data at speeds approximating the best landline dialup connections today. AT&T will have no response as long as it remains a TDMA network.

The only coming option for TDMA networks wishing to deliver robust data is a bit of current vaporware called EDGE, which will not arrive in the U.S. until 2002 if ever. If it does, its data rate of 384 kilobits per second, impressive sounding today, will seem paltry compared to Qualcomm’s High Data Rate system, which is six times as fast.

Today AT&T is pushing not one, or two, but three obsolete and incompatible wireless technologies: mobile TDMA for cellular voice, EDGE TDMA for data, and Angel, its heralded fixed wireless solution. All are costly and unproven and drastically inferior to the coherent upgrade path of the CDMA competition. AT&T should not take solace in the fact that BellSouth and SBC, two of the remaining Baby Bells, seem set to follow their former parent into the TDMA wasteland. AT&T doesn’t have a monopoly anymore.

Ignore Sunk Costs

A cardinal rule of business is that sunk costs are sunk. Unless you ignore them, you sink your business. If AT&T were smart enough to ignore the sunk costs of a foolish and extravagant TDMA mistake, it could once again take the world lead in wireless that Bell Labs long ago conferred. AT&T’s mission should be to add new capacity while upgrading for the new Internet era, all without inconveniencing existing customers.

Just as the current TDMA digital was phased in as an overlay on an analog system, the new CDMA system called 1X multicarrier can be phased in step by step. CDMA would allow AT&T to increase its voice capacity nearly threefold and offer up to 150 kilobits per second of data. Any further available spectrum can be allocated to the High Data Rate system, which allows use of 1.25-megahertz CDMA channels to offer up to two megabits per second of Internet data. Thus customers on the road could plug their laptops into their cell phones and get better Internet connections than they received from their office T1 lines or their residential Digital Subscriber Lines. At the same time, AT&T could replace its moribund Angel Wireless Local Loop technology with the same essential CDMA system.

With AT&T turning to CDMA, the U.S. could command a critical mass of customers that assures the nation global leadership in wireless Internet. As Peter Drucker says, “Don’t solve problems. Pursue opportunities.”