Call it Lucent Start-Up, or Lucent Launchpad, where the new motto might be "Keep it going and keep it flowing."
Here in the Merrimack Valley north of Boston, in the plant where the old AT&T used to make bobbins and tool sheet metal, lies the future of troubled network equipment giant Lucent Technologies. Its here where the optical equipment is designed, assembled and tested, and where the comeback will happen, if it happens at all.
Gone are the separate repair shops, the far-flung testing labs, the segregated strategy sessions and the nightmarish inventory system.
In their place are assembly cells where a dozen people put together, repair and test high-speed optical products. Theyre places where everybody knows each others name, and where a glitch in a switch can be spotted, repaired and retested by the time the shift is over.
"We get them out fast," David J. Dunn, factory director of Lucents Optical Networking Group, said last month during a visit to the 40-year-old plant. "The smart guys are the ones that continue to evolve and move and change."
Thats exactly what Lucent hadnt been doing for most of the past decade. As the 1990s grew old, other legacy companies such as Nortel Networks and younger competitors such as Avici Systems, Ciena, Cisco Systems and Juniper Networks jumped into the OC-192 market — building products to carry traffic over optical fiber at 10 gigabits per second. Lucent, meanwhile, was content to supply its reliable long-distance voice customers with slower switches for aging networks.
Late last month, when Lucent Chief Executive Henry Schacht announced a restructuring and 16,000 job cuts, he blamed the companys $1.02 billion quarterly loss on too much vendor financing, too many last-minute discounts and overflowing, out-of-control inventory. Lucent restated fiscal 2000 earnings in December, after investigating its own accounting policies. Now the Securities and Exchange Commission has launched a probe of its own.
Executives eager to preserve their legacy lines would offer 90 percent vendor financing or, worse, offer deep discounts to customers if they would buy the products one quarter early. That way, it looked to the accountants that the old stuff was still selling, so more components would be ordered and gather dust in the warehouse until a last-minute flurry of discounted orders at the end of the next quarter got some of them moved.
All of that changed a few months ago at Lucents North Andover plant.
Key to the culture change was shrinking the assembly process. Its a "bank teller" approach, in which each team takes turns filling the next order in line. It keeps products from piling up in the post-assembly warehouse. And it keeps expensive components from piling up in the preassembly warehouse during slowdowns. Only when assembly begins is a new component reordered. Using multiple vendors for each component ensures against a sudden shortage that can stall production.
The group that ramps a new product up to high-volume is the "launchpad," and the goal isrocket-ship speed. "Its a challenge to reinvent ourselves," says Christopher B. Little, Lucents senior manager of high-speed optical parts manufacturing.
"Theres always been a criticism that Lucent is too big and takes too long to do things," says Wayne Barrett, director of manufacturing and engineering for optical networking products. "But here, were launching the entire manufacturing into volume manufacturing inside the same building."
Lucent was late to the gate when 10-Gbps products raced into the marketplace. But now its 10-Gbps line runs three shifts per day and the company is determined not to be late again, when the market shifts to even faster OC-768 (40-Gbps) optical products.
The shift from 10 Gbps to 40 Gbps will require the substitution of a few components, but almost no change in testing, repair, the assembly cells or the procurement process.
"In the past, we climbed a conservative ramp, introducing a product and growing it gradually," Barrett says. "The shifting demands of the marketplace today dictated to us that we need to climb a much more aggressive ramp." The plant is producing 12 times the number of 10-Gbps products as it was nine months ago.
Assembly workers now are part of daily evaluation teams with managers, and theyre chosen for their outspokenness, not their acquiescence.
The cells that make up the launchpads are 1,200 square feet, about the size of a two-bedroom condo. "Everybody knows whats happening," says John M. Hudak, technical manager for Lucents WaveStar LambdaRouter. "The feedback is immediate. That reduces inventory, and thats critical for cash flow. "We slow the supply chain down. Expensive items like lasers and receivers are on consignment. The longer we delay pulling those from the storeroom, the better we are doing."
Each new shift is given an immediate status report, so it doesnt spend the first 15 minutes wondering where the torque driver went.
Each cell on an eight-hour shift is producing about 13 boxes, though the goal is 15. When the materials are available, the plant can make 600 units per week. A materials shortage is measured not in hours of idleness, but in lost revenue opportunity.
That all sounds good, and Schachts downsizing plan sounds good, too, but this is all stuff Lucent should have done two or three years ago, says Amiya Chakravarty, a Tulane University management professor whose new book, Market Driven Enterprise, includes an examination of Lucent.
Lucent has good products and great research and development, but has been slower than Cisco and others to adjust to the new world, where manufacturing isnt necessarily important, Chakravarty says. "Cisco hardly manufactures anything themselves. The reason Cisco is getting ahead is theyve found a way of managing the chain of people from beginning to end, from the raw material suppliers to the customers."
Lucent has acted more like a conglomerate, doing everything from research to manufacturing, he says. "Lucent has had trouble collaborating with others, always afraid it would be giving up too much," he says. Because Lucent has its hands in so many fires, it has had to be as slow as the old telephony world and as quick-footed as management software and optical components.
Lucents workhorse product, the 5e telephone switch, is still making money, but with growth being in data traffic, networks are opting for systems that can switch everything — voice, data and video.
Lucent has "a tremendous amount of assets," but "do you look at the assets or the market capitalization?" Chakravarty asks. If Lucent spins off its microelectronics division, Agere Systems, at the reported $12 per share, that leaves the leftovers of the once-proud giant worth about $5 per share. Its stock was trading at about $77.50 early last year.
"Splitting up the company may help them get closer to the customer," he says. "But the question remains: Are they going to find a niche for themselves or try to be all things for everybody?"
Lucents rigid central purchasing office treated Agere worse than any competitor — insisting that it be as competitive with prices and services as any other vendor, but not letting it sell its products to anyone else, Chakravarty says.
Bill Lesieur, industry watcher at Technology Business Research, agrees that Lucents restructuring move is sound, but says the key question is whether Lucent can line up optical products quickly enough to make inroads against Cisco, Nortel and others. Lesieur asks: "Will the strong get stronger and the weak weaker, which often happens during a potential downturn?"
Lucents stock may have taken a beating for several quarters, but its still proud of its Bell Labs legacy and of its hard science. Little says Lucent is ahead of the industry in wavelength routers.
Lucent has been getting good reviews — and customers such as Time Warner Telecom — for its WaveStar LambdaRouter, which is actually an all-optical switch, capable of moving traffic onto as many as 256 different channels on a fiber.
The WaveStar LambdaRouter uses a pair of boxes, each holding 256 tiny mirrors. By magnetically moving the mirrors from zero to 3 degrees, traffic can come off any of 256 wavelengths, cross 4 inches of free air and beam onto any of the mirrors in the second box, rerouted onto any of 256 wavelengths. That means a video destined for Chicago from Miami can take a left turn at the New York switching station, or a bandwidth-gobbling videoconference can switch to an emptier wavelength just in time to prevent a collision of bits and bytes.
In six months, Lucent will pack two pieces of software with its WaveStar LambdaRouter — an optical systems manager, and an optical network navigator that makes the platform smarter and able to allocate bandwidth. And Lucent says it will be first to market this year with faster wavelength routers, one with a pair of 1,024-mirror boxes.
Lucent will survive because it still has a strong product development team, says Chris Nicoll, who follows the optical market at Current Analysis. "A shift to new technologies gives them a chance for redemption, and they have a very strong product. I wouldnt say Lucent is anywhere close to shutting its doors," he says.