Must reading for tech stocks:
boston.com
HIGH TECHNOLOGY
Westford start-up promises method to vastly boost speed of cable lines
By Peter J. Howe, Globe Staff, 2/18/2001
WESTFORD - The magic of high-speed Internet access through a digital subscriber line - if you can get one installed and make it work - is that it exploits huge amounts of data-transmitting capacity that went unused for decades on copper phone lines.
Now Narad Networks, one of the big recent winners in the local telecommunications start-up funding
derby, is aiming to do for cable televison networks what DSL has done for the phone system: Open
them up for Net connections dozens of times faster than even today's 1.5-megabit broadband links.
Seven-month-old Narad earlier this month landed $41.6 million in first-round venture funding from
Polaris Venture Partners and a big-name list of networking pioneers, including 3Com founder Robert
Metcalfe, top executives from Sonus Networks and Juniper Networks, and Rouzbeh Yassini, an
Andover broadband investor known as ''the father of the cable modem.''
Narad founder and chief executive Dev Gupta, 48, who has already sold two start-ups to Cisco
Systems, said he is convinced that even after investing billions of dollars in upgrading their systems,
cable giants like AT&T Broadband and AOL Time Warner have only just begun to tap their potential
power.
Across the country, cable companies have spent tens of billions of dollars extending fiber-optic lines
into neighborhoods to boost the carrying capacity of the ''coaxial'' cable lines that pass along most US
streets.
In the last 18 months, those investments have begun to pay off in a big way. Over 3.5 million US
homes have flocked to cable modem services such as AT&T's RoadRunner, and millions more have
signed up for scores of channels of digital television and local phone service delivered over upgraded
TV cables.
Gupta is willing to predict that before the end of this year, 120-employee Narad will have a system
undergoing trials at cable broadband companies that ''will make your coax look like fiber.''
Conceivably it will deliver ''gigabit ethernet'' connections of potentially 1 billion bits per second - over
600 times faster than RoadRunner or ExciteAtHome on their best days.
For now, Gupta is keeping close to the vest just what kinds of software and devices will be needed to
achieve these speeds, and how much they might cost.
However, the system will work on a ''pay as you grow'' model, requiring upgrades only to serve new
customers asking for super-high-speed Net service, not the kinds of border-to-border franchise
upgrades that have cost AT&T over $2 billion in Massachusetts and New Hampshire alone.
''You only put up my equipment when you have customers,'' Gupta said, adding that the recent Wall
Street collapse of so many telecom carriers shows that ''you cannot any longer build railroads hoping
that passengers will come.''
The theory underlying Narad is that any given foot of coaxial cable can easily handle gigabits of data
passing through. However, because signals fade over distance and need to be periodically amplified,
cable systems normally limit their throughput to ensure that the home five miles from the cable office
can get the same services as the home 100 feet away.
Narad is developing an ethernet-based approach to managing cable wires that would ensure far more
of the system's capacity is used in every segment between pole-mounted amplifiers and ''drops'' to
each home or business.
While it may be a long time before many US homes want or can use more than 1.5 megabits of Net
access, many of the country's 8 million small- and medium-sized businesses are clamoring for even
faster access.
Many face months of waiting to get a ''T-1'' broadband line from Verizon Communications or another
carrier. With exceptions such as Time Warner in Maine, cable companies have generally been slow
to market data services to businesses, in some cases because their networks were limited to
residential rather than commercial areas.
By focusing on further upgrades of perhaps 5 percent of their network mileage, Gupta said, cable
companies could easily double their cash flow selling super-high-speed access to businesses.
Narad's approach will likely put it in head-to-head competition with another nearby start-up, Quantum
Bridge Communications of Andover, which plans a public offering soon.
Quantum Bridge sells so-called passive optical networking systems that would bring low-cost fiber
connections from homes and businesses to the ''fiber node'' that normally serves a cluster of 500 to
1,000 cable homes, through a last-mile coaxial connection.
Narad, in contrast, would not add new wiring, but vastly expand the carrying capacity of the coaxial
last mile so that, in the best case, it carried at least as much traffic as a Quantum Bridge fiber line.
''We believe that the cable network has a lot of capacity that has yet to be exploited,'' Gupta said.
''They can make bandwidth on these networks 1,000 times cheaper than on copper [phone]
networks.''
While it will be months before Narad has a technology on the market that can be evaluated, the
company's vision has attracted many well-known supporters.
Metcalfe, known as the inventor of ethernet, a widely used network system for dividing
data-transmitting capacity among computers, said he sees Narad as having potential for
''revolutionizing the Internet's plumbing.''
Other investors include Pradeep Sindhu, chief technology officer of Internet router maker Juniper
Networks, a former dorm-mate of Gupta at the India Institute of Technology; Kopin Technologies
chief executive John Fan; and Rubin Gruber and Hassan Ahmed, founder and president of Sonus.
Gupta, who began working on high-speed data networks at Bell Labs in the mid-1970s, including early
DSL systems, later played key roles at Dagaz Technologies in New Jersey and MaxComm
Technologies Inc. of Chelmsford, both bought by Cisco - MaxComm for $143 million in August 1999.
This time around, Gupta said, he plans to stick with Narad and make it ''a big company. We've been
there, done that - twice,'' he said of the Cisco-takeover route.
''It's time to move on to something different, and I think the scale of this company just doesn't allow it
to be bought.'' |
