From Graphics to Optics
The company announced it will show off its technology at the National Fiber Optic Engineers Conference (NFOEC) in Baltimore next month (see Creo to Demo Crossconnect).
Transferring optical technology from other industries to data communications has been tried before -- and not always with success (see APA Optics Inc. (APAT)). Then again, there's always Corning Inc. (NYSE: GLW), which until lately was the ultimate rust-belt-to-datacom success story.
Creo says it has developed an alternative to 3D MEMS (micro-electro-mechanical system) technology, an approach to all-optical switches. The company says its all-optical switching technology will scale up to 4,000 ports, four times greater than announced MEMS-based systems.
“Our plan is to displace MEMS altogether,” says Doug Richardson, director of technology for Creo. “When we first started working on this a year ago, we looked at everyone else’s spec sheets. Then we did what we could to beat them.”
But analysts question the usefulness of Creo’s super-scaleable technology given the fact that large all-optical switch fabrics may not be in high demand.
MEMS technology, which consists of hundreds of tiny tilting mirrors that steer beams of light through a switch fabric, works well on a small scale, but larger matrices are difficult to maintain, say analysts. Nortel Networks Corp. (NYSE/Toronto: NT) and Integrated Micromachines Inc. (IMMI) have both announced that they are developing subsystems that scale to more than 1,000 ports. Others have also announced scaleable MEMS solutions (see MEMS Make It Big?). Corning says it is working on a 256x256 port matrix and Agere Systems (NYSE: AGR) says it has a 64x64 switch on the way. But so far none of these companies has shipped a fully populated fabric.
“It’s difficult to get a MEMS to scale over 16x16 ports,” says Scott Clavenna, president of PointEast Research LLC and director of research at Light Reading. “You can fabricate a design, but to get carrier-class reliability is extremely difficult. The technology behind how you actually change the angle of the reflection seems to be in its infancy.”
Creo’s technology doesn’t use mirrors to steer light beams -- it uses bendable fiber. Each fiber sits behind a lens, which focuses the light and aligns the beam with another lens on the other side of the switch. When a data beam comes in, the fiber is magnetically moved to line up with another fiber across the matrix. As light is transmitted, the lens directs the beam and focuses it on another lens on the other side, which focuses the traffic beam onto the appropriate fiber.
The switch scales modularly, with 12 fiber and lens pairs per module. At NFOEC in July, Creo plans to demonstrate a 32x32 optical switch (three modules on each side). The company says it will also show off a partially populated 1000x1000 switch fabric and will demonstrate fibers that can bend enough to support a 4000x4000 matrix.
The company is getting the word out about its new direction among equipment companies. Tellium Inc. (Nasdaq: TELM), which plans to build a large all-optical crossconnect, says it has already talked to Creo.
“It’s an interesting idea, and we’ve had conversations with them,” says Harry Carr, CEO of Tellium. “Right now we think that MEMS is the way to go, and we’ve already publicly announced that we’re partnering with Corning. But we’re always open to looking at new approaches.”
But the big problem may be with the market itself. All-optical switching technologies have yet to take off, and large all-optical switches may never be needed, say analysts like Tim Anderson from Salomon Smith Barney. He says the industry hasn’t yet decided how far matrices will need to scale.
“The issue is not whether or not MEMS scales,” says Anderson. “The issue is: What scale do we need and what will we need in the future? There is certainly no demand right now for a thousand-port all optical switch in high volume.”
Several equipment companies like Ciena Corp. (Nasdaq: CIEN) and Tellium are developing a hybrid approach that combines electrical switching at OC48 (2.5 Gbit/s) and OC192 (10 Gbit/s) speeds and optical switching to handle traffic at higher speeds like OC768 (40 Gbit/s). Hybrid switches will reduce the need for highly scaleable all-optical switches, because OC768 traffic, which will be switch optically, will likely be a much smaller proportion of the total traffic handled, says Clavenna.
Ross Turnball, an equities analyst specializing in industrial products for Odlum Ltd., covers Creo and wonders why the company would change its focus. He says Creo is a market leader in a sector expected to be worth about $100 billion in 2001.
“It doesn’t make much sense to me,” he says “It surprises me that they would enter a completely new market and spread themselves thin. They seem to have their hands full with the market they’re already in.”
So why is Creo taking on a new market? The answer could be in the numbers. Even though valuations of many optical-networking companies have fallen drastically over the past year, they still carry higher price-to-sales ratios than companies in other industries. For example, Creo currently has a market capitalization of about $944.2 million, with annual revenues running about $700 million, giving it a price-to-sales ratio of a little more than one. Compare this to the Lucent Technologies Inc. (NYSE: LU) components spinoff, Agere Systems, which has a market capitalization of $10.1 billion on revenue run rate of about $5 billion, giving it a price-to-sales ratio of roughly two-to-one, nearly twice that of Creo.
- Marguerite Reardon, Senior Editor, Light Reading
For more information on NFOEC, please visit the Light Reading NFOEC Site.