Inside the headquarters of Corvis Corp., there are polished floors in the entranceway, brand-new carpeting lining the hallways, and rich wood furniture, such as the 12-foot-long conference table surrounded by beautifully upholstered chairs.

Then there are the things most visitors can't -- or aren't allowed to -- see: the millions of dollars worth of fiber-optic equipment, the labs and the eventual manufacturing facility that is expected to employ 1,500 people building optical switches that could, if successful, revolutionize the efficiency of the Internet.

Though only two years old, Corvis clearly is no basement start-up. It is the brainchild of David Huber, a soft-spoken physicist sitting on a good idea and tens of millions of dollars in venture capital, ready to enter a business in which billions of dollars can be claimed by the company that can reduce the costs of a data transmission.

The Columbia company is one of the best-funded start-ups in the telecommunications industry today. It's also one of the most talked about, though (and perhaps because) Corvis has been all but silent on what, exactly, it is doing.

The Second Time Around

Corvis is an encore venture for Huber, a founder of Linthicum-based Ciena Corp.

Huber left Ciena in 1997, two months after the fiber-optics company's initial public offering raised $3.4 billion, the best IPO performance that year. Huber, a leader in the field of fiber optics, had a falling out with the Ciena board of directors, which disagreed with his technological vision for the company. So he left to pursue that vision on his own, taking $300 million worth of Ciena stock with him.

After a two-month rest, he sank part of that stake in Corvis, a company he hoped would offer the first all-optical network in the world. Strictly speaking, he has succeeded, say Huber and the few analysts privy to the company's technology. As for leaving Ciena, he says, "I haven't regretted it for a minute."

Huber is backed by some of the most respected money in venture capital. Kleiner Perkins Caufield & Byers and New Enterprise Associates, which has investments in UUNet Technologies and 3Com Corp., among others, have invested at least $15 million combined. Cisco Systems Inc. has also taken about a 10 percent stake in Corvis. Huber will not disclose an exact figure, but he says that counting his own investment, the company has raised somewhere between $40 million and $80 million.

Sitting in his office recently, Huber explained the basic idea behind Corvis. "The issue is cost," he says, slowly bringing his hands together like a thoughtful preacher. "And scalability."

In English: Long-distance carriers such as MCI WorldCom and AT&T, which haul data traffic from one place to another, spend huge amounts of money converting those data from light to electricity when they pass through a switch. Switches are the intersections that make sure data get where they're supposed to go, so eliminating the switches isn't an option. The trick is to make them better.

This year data traffic overtook voice traffic. It's a troubling development for carriers because data isn't as profitable as voice. Cutting the cost of the transfer is that much more important. And the electrical conversion at the switch accounts for about 75 percent of the cost of hauling traffic on an Internet backbone.

Huber claims he can eliminate that cost. Corvis's all-optical network will be able to transmit stronger optical signals than those of competing companies. Typically, a signal can travel about 500 to 600 kilometers in optical form before degrading, but in Corvis's network, signals could travel up to 3,200 kilometers before degradation sets in. In addition, the network can carry far more data than existing networks, and Corvis has developed an optical switch, which eliminates the need for converting the light to electricity.

The distance that signals can travel on the Corvis network is a breakthrough, says Mat Steinberg, an analyst with the research firm RHK Inc. in South San Francisco.

"Without the distance, you can't do an optical network because you would lose the economic advantage," Huber explains. "If you lose that, you're down in the trenches with the rest of them. That's no fun."

Keeping a Low Profile

Because of the intense competition, Huber has operated in stealth mode during much of the past two years.

"There was a lot of interest, but we resisted it pretty systematically," Huber says. "There is nothing to be gained by letting other people, other competitors, know what we're doing."

Corvis had no Web site until April. And even now, the site contains so little technical information that an analyst recently sighed in frustration when he could not find an answer there to a simple technical question. All analysts who speak with Corvis do so only after signing nondisclosure agreements that cover much of the technology.

Huber compares his work to the arms-race atmosphere that characterized the computer chip industry. "The ability to be there with a lead time of just a few months ahead of competitors is essential," he says. Few people have toured the Corvis lab, for example, where engineers in royal blue lab coats work on the hardware and software that will power the Corvis network. The lab is one of Huber's favorite places, and on a recent visit he moved from one giant hardware cabinet to another, pointing proudly to transmitters and receivers as well as switches, and explaining how they will direct wavelengths -- broken down into 160 colors -- around the country. But while Huber says secrecy is necessary, at least one analyst -- who requested anonymity -- maintains the precautions border on "arrogance and paranoia."

Chris Nicoll, an analyst with Current Analysis in Sterling, says the secrecy does raise questions. "I think it's distracting from Corvis's main message, all the secrecy they're applying to their technology," he says. "Corvis appears to have some revolutionary applications, but it's getting lost in the secrecy -- you know, why are they being secretive?"

There is one possible reason, some inside the industry have speculated. It could be that close examination of the Corvis network would reveal that the "all-optical network" is merely mostly optical. Although Huber denies he is twisting words, some industry insiders suggest that while end-to-end the network is optical, adding data along the way could require a minor amount of optical-to-electrical-to-optical (OEO) conversion.

This would not be a reason to attack Corvis, says RHK's Steinberg. "At the edge of the network the data would need to be converted . . . in order to be carried by local networks."

Fahri Diner, founder of Qtera, a competing optical firm, says he has heard there is some OEO conversion involved in the Corvis network, and he says that at a recent conference there was some debate over Corvis's use of the term "all optical."

Huber dismisses the rumors. "We've been very direct," he says. "And I'm not a person who believes in sleight of words."

Calmly Passionate About Science

Quietly competitive, Huber has spent most of his life trying to stay out of the "trenches." He grew up in the low-tech town of Le Grande, Ore., a burg known for its railroad and lumber. His father, a county agent, headed up the local 4-H club, and his mother taught school until they both retired and moved to Utah.

Huber is not sure where his love of science came from, except that he has always been a person with a full measure of curiosity. When he got to Eastern Oregon University, he decided that physics seemed the most interesting, cutting-edge field, and so he tackled it. He went on to get a PhD in electrical engineering from Brigham Young University.

Lightwave technology "is very challenging and very exciting. A lot of scientists get a rush of enjoyment from being on the leading edge," Huber says, though his facial expression rarely shows anything but calm.

Huber says one of his heroes is David Payne, a scientist who in 1987 developed a way to amplify light signals by sending them through fiber soaked in a rare element called erbium. The discovery paved the way for companies to send information literally at the speed of light.

"The optical amplifier just changed everything," says Huber, who in 1989 received a call asking him to join General Instrument Corp. and help it commercialize the discovery.

At Corvis, Huber surrounds himself with people who love technology as much as he does. "I don't think we have any mercenaries here," he says, referring to those who join start-ups with the sole purpose of cashing in and then getting out. "I've met some," he says as if talking about another species altogether. A former colleague at Ciena, after the stock had run up dramatically after the IPO, once rushed into his office and asked if Huber thought the stock had any life left in it. "That individual was definitely on the mercenary side," he says.

There's Money to Be Made

Not that Huber doesn't want to make money. The optical networking market for long-distance transport and switches will reach $14 billion in 2003, according to a report by RHK. Huber will not release revenue projections, but says he plans to capture a significant chunk of that market. Analysts project his company could bring in revenues of anywhere from $10 million to $30 million in 2000.

But Corvis will face stiff competition from companies such as Nortel Networks Corp., Lucent Technologies and Huber's own creation, Ciena. All of them have been developing optical offerings, many of which can work with a carrier's existing network.

Persuading service providers to replace their entire networks with the Corvis network could prove challenging, says one industry insider and potential Corvis competitor.

"It's not always easy to go to a carrier and say scrap what you have," says the source, who asked not to be identified.

Corvis says its system can work with a carrier's "legacy," or existing, systems. But Glenn Falcao, executive vice president of Corvis, says the challenge will not be in persuading customers to adopt Corvis technology, but in convincing them Corvis can provide sufficient service and support.

This is where being a start-up could hurt Corvis, analyst Nicoll says.

"The challenge is not necessarily product, it's service and support," he says. "When you compete with a Lucent or a Nortel, products are only maybe half of what is actually being evaluated." When people hear about an MCI WorldCom outage, for example, "you immediately hear that the Lucent engineers, or whoever, have been working around the clock -- that is hard for a small company to compete with."

Falcao, who spent 23 years at Nortel, most recently as president of its Internet and service provider business, says Corvis is busy putting a support infrastructure in place internally and has also begun working with an outsourcing agency to train it on Corvis products.

The sales cycle for Corvis products is expected to last from six to eight months. Corvis is about to enter the testing stage with several potential customers, although it will not be ready to announce the names of those prospects for at least a couple of weeks. "It is not a fast sales cycle," Falcao says, because the customers are "betting their business on you."

Depending on the size of the network, Corvis customers could spend anywhere from $150 million to $500 million, Falcao says.

In addition to established companies, there are a handful of start-ups eager to compete with Corvis. Cisco has hedged its bets, buying Monterey Networks of Richardson, Tex., and investing in Optical Networks of San Jose.

Because of the long and involved sales cycle, Corvis's head start could prove crucial to its success. It is months ahead of competitors such as Qtera, a company in Boca Raton, Fla., with $43 million in equity and debt financing. Qtera is also working on an all-optical network, although "we like to call it the `purely photonic network,' " says Qtera's Diner.

For his part, Huber marvels at how far technology has come. As a religious man -- he is a Mormon -- he believes technology is "amoral," but that people have a moral obligation to use it for good. As a scientist, he feels yet another obligation. "Our job is to see if we can be at the forefront," he says.

This can be difficult. When he entered the industry 18 years ago, he points out, the fastest data transmission was taking place at 30 megabits per second. Today, data can travel more than 100 times faster. "The technology has a life of its own," Huber says.

A Look at

Corvis Corp.

Headquarters: Columbia, Md.

Business: Fiber-optic

telecommunications networks

CEO: David Huber

Founded: 1997

Funding: Between $40 million and $80 million in venture capital