The Revolution That Failed - Company Business and Marketing - Brief Article

AT&T spent millions hiring world-class experts to build a universal platform for the Web. Then it all fell apart.

IN THE LAST FEW YEARS, AT&T HAS tried a number of ways to remake itself as an Internet company. CEO Michael Armstrong has spent billions on a cable network to deliver broadband services and content. It even considered selling off its 100-year-old phone business to focus on Internet services.

But long before those efforts, AT&T launched a secret Internet project, wider in scope than anything the company has ever attempted. The project, known by the code name Geoplex, was designed to build a universal software platform for the Internet. At a time when research budgets were shrinking and science was increasingly limited by the corporation's bottom line, AT&T's aggressive and well-funded Geoplex was an anomaly.

But management never quite knew what to do with the project, and the grand Geoplex concept withered. Most of the researchers behind it scattered to other companies. AT&T is now starting to use elements of the Geoplex technology, but the scope, scale and ambition of the original project are gone forever. And with them perhaps went AT&T's best chance to transform itself into a powerhouse technology company for the Internet Economy.

Born out of academic and government research projects, conceived by scientists working independently, the Internet is imperfectly designed. Its very name encapsulates the problem: It's a collection of different networks, a "between-net," not a single seamless network.

Geoplex was meant to change that. Its goal was a software platform, or basic architecture, for making Net-based networks run more smoothly. With Geoplex, companies and other entities would be able to more easily exchange data and build businesses. Techies refer to the project as middleware, a nebulous term for software that connects systems.

Think of the United Nations. In the U.N.'s General Assembly, representatives of many different nations speak many different languages. Each one, however, can pick up a headset and listen in his or her own language. Geoplex was intended to work in a similar fashion.

In 1995, executives at AT&T, then still the home of Bell Labs -- the birthplace of the laser, communications satellite and transistor -- decided they needed to hire the best Internet researchers available. One of the first was a Croatian-born scientist at Sun Microsystems, Dado Vrsalavic, who worked on Sun's Solaris operating system, the software that powers Sun' ubiquitous Web servers.

An imposing former international water-polo player with a thick Croatian accent, Vrsalavic looks as much like an Olympic wrestler stuffed into a suit as he does a research scientist. He's also an inspiring leader who was able to get hundreds of scientists, as well as AT&T management, to support his vision for almost five years.

His first project was to help build WorldNet, AT&T's consumer Internet service. After that, Vrsalavic took the helm of a group known as the Internet Platforms Group. The group started with five people in Middleton, N.J., and eventually grew to roughly 250 scientists in the United States and 50 more in the United Kingdom.

But before they could start working, the earth moved beneath them. In 1996 Bell Labs was split from AT&T -- taking two-thirds of the company's staff -- to become part of the newly minted Lucent. The Internet Platforms Group stayed with AT&T, joining what was to be known as AT&T Labs. "That was the first of, let's say, unfortunate events that threw us off of our expected schedule," says Nino Vidovic, the former AT&T VP of Internet platform technology.

The platforms group soon found itself adrift within the phone giant. "When we joined AT&T, it was a technology company," says Vidovic. "After the divestiture, we were almost alone as a technology group within a services company. They didn't always know where we fit."

Another problem was that Geoplex was supposed to be a joint venture with IBM, not an independent research project funded solely by AT&T. But Big Blue instead turned its efforts toward building its Global Network Services Network -- which, ironically, AT&T wound up buying in December 1998 for $5 billion in cash.

Undaunted, the Geoplex group set about writing a set of software interfaces that would enable functions like authenticating and registering customers as well as billing.

With the beginnings of a product in hand, the group muddled through a series of business models. After the joint venture with IBM fell apart, they tried to sell the software as a sort of starter kit for Internet service providers, hoping the software would filter up through the industry. They tried to spin themselves off as a company called InterNOS, but that fell apart with the Lucent divestiture. Then they regrouped and began plans to spin off again, this time as GeoSphere Communications. The plan was to get big national telcos like Deutsche Telekom and Japan's NTT to adopt the platform, hoping it would filter down. Trials were run in Germany, Japan and Hong Kong.

The team had been gearing up for the GeoSphere spinoff when Michael Armstrong came on board as chief executive in 1997. The plan was canceled shortly thereafter. "It was quite a shock' says Vidovic. "We felt like we were weeks from that event, and then we were suddenly told it was not going to happen."

By then it had become clear that Geoplex would never fulfill its grand promise. The delays took their toll, and people began to leave, In October 1999, Vrsalavic took a job as VP and chief technology officer with Intel, and the floodgates opened. Soon other key researchers bailed out.

Vidovic and George Vanecek, AT&T's former chief scientist, founded an Internet startup. Geoplex seemed stillborn.

The perception within the industry is that the project failed. "Geoplex, boy that was a disaster," says Danny Lewin, co-founder and chief technical officer of Akamai. "The fundamental ideas were good, but they just didn't work very well in the real world."

One thing that hurt Geoplex was that the group was literally ahead of its time. Back in 1995, there was no name for what they were trying to do. Today, a bunch of terms exist for the technologies they created. In fact, smaller companies that offer only small pieces of technology similar to what Vrsalavic and company attempted have made billions on Wall Street.

For example, Akamai and Inktomi have multibillion-dollar market capitalizations based solely on caching technology similar to that developed at AT&T.

If the Geoplex group had become a spinoff, it could have dropped the bigger project of trying to fix the communication gaps between carriers and tried to build a more focused business. "If we had gone independent, we would have behaved like Akamai, or Inktomi, or Software.com," says Vrsalavic. "We could've focused on smaller issues that would be below the radar of AT&T, but pretty big for a startup."

The group never managed to become an independent company and never quite found its place within Ma Bell. "We were almost doomed to fail," adds Vanecek. "If we kept it inside AT&T, we are not a standard and are ignored by the rest of the industry. If we take it outside of AT&T, we lose the support of AT&T and the various business units."

Though Vrsalavic says he doesn't have any regrets, it's clear he and the other Geoplex leaders would do things differently if given another chance. "If I could do it all over again, I would go open source," says Vrsalavic. "Open source would give a lot of people a chance to work on it. Now that I have seen how much it'll cost, I doubt any one company could do what we were trying to do. Not even AT&T."

And even if Geoplex had worked, it's not clear it could have become the universal standard that AT&T envisioned. "The world wasn't ready to have a monolith dropped on it," says Shane Robison, Vrsalavic s successor. "It was a one-size-fits-all approach, but the world doesn't work that way."

"The original concept and point of view [of Geoplex] was largely right," says John Petrillo, AT&T's executive VP of strategy, the man who hired Vrsalavic. "What we didn't have was a good idea of how the market would evolve around us."

Nevertheless, several Geoplex fragments made it into the world. Elements of software developed by Vrsalavic's team were passed on to universities involved in the Internet 2 research project. At least one product, server software developed with Sun Microsystems for early trials, was talked about in a press release, though little has come of it.

Under Robison, the first product for AT&T, code-named OneComm, has finally been announced. The plan is to offer a broad swath of messaging options based around Geoplex technology. Using the platform, AT&T has created software that will give its applications a common engine for looking up information -- making it possible, for example, to access the same e-mail from the Internet and the company's wireless phones.

That sounds simple, but it's a deadly serious matter for AT&T, which has built its business future around offering bundles of applications on various platforms.

But even as parts of Geoplex are beginning to see the light, AT&T has announced a round of layoffs among scientists in the Internet platforms organization, with 50 more researchers given pink slips. And just last week, Robison left to become the chief technical officer at Compaq, leaving the rollout to others.

"It was an exciting time to be at AT&T," says Vidovic. "We thought we were truly creating something that would last for years and transform the industry. We were ahead of our time."

COPYRIGHT 2000 Standard Media International
COPYRIGHT 2000 Gale Group