HOSSEIN ESLAMBOLCHI, AT&T

The irony of AT&T is that the most protean company in telecom is the one that hasn't changed its name in more than 100 years. Drawn and quartered more than once, it has seeped from long-distance into wireless, local exchange and broadband cable (and back). And its most dramatic shape-shifting trick may be yet to come.

These days, while AT&T CEO David Dorman pulls hard to port on the wheel of his Queen Mary, chief technology and information officer Hossein Eslambolchi toils away in the engine room, trying to turn the whole thing inside out. When Eslambolchi — who is also president of AT&T Labs — is done, AT&T will no longer be described as a telco, he said, but as a “software-based, IP-based company.”

AT&T's technological course is aligned with the constellations of Eslambolchi's two most ardent beliefs: “Software is the battleground for the 21st century,” he said, and “IP will eat everything — like Pac-Man.”

Though Eslambolchi's pop culture references are a bit rusty (Pac-Man?), his network is state-of-the-art. AT&T already has MPLS-enabled its ATM core and currently boasts one of the world's biggest IP networks. To equip it for the future, Eslambolchi instituted what he calls a “skill-set transformation” at AT&T Labs over the past two years, stacking its ranks with IP and software experts. And he upended the group's priorities, from an 80% focus on the core of the network to an 80% focus on the network's edge. He also shortened the group's calendar: 80% of its research was previously devoted to long-term strategies (those for which payoff comes five to 10 years down the road), and 20% was focused on the next three years. Today that 20% is still focused on the next three years, but the other 80% is devoted to the next 12 to 18 months.

Despite that present-tense thinking, AT&T may not be moving fast enough to fend off its most immediate threats. It hasn't moved as aggressively into the local exchange as the RBOCs have moved into long-distance, and it hasn't yet launched a competitive voice/DSL bundle. The resulting pressure on its consumer markets has Eslambolchi looking for any technology that will allow AT&T to bypass the Bell local loop and the billions of dollars in access charges that come with it.

“If I find a pigeon that can carry bits between A and B, I'm interested,” he said.

By the time this story is published, AT&T will have picked a vendor for multiservice access devices, said Eslambolchi. In next year's second quarter, it will deploy them. The company will pick a multiservice edge platform before this year's end and deploy it en masse early next year. And this winter, AT&T will deploy a next-generation packet-based digital cross-connect that it has already selected but hasn't yet announced.

The question remains whether technological changes can heal the cracks in the carrier's business, which has seen declining revenues and mounting competition. Rumors circulated this spring that AT&T might look to merge with an RBOC, one potential transformation Eslambolchi won't comment on — except to say, “It's like the law of gravity: The biggest planet wins. We are now like Jupiter. Jupiter always wins.”