By Eric W. Pfeiffer
Small Times Correspondent
More than 18 months ago, Nortel Networks announced it would spend $3.25 billion for Xros, a startup that was building a MEMS-based optical switch that promised to supplant its costly and inefficient electronic counterpart. The mega deal prompted venture capitalists to dump money into similar startups, to the tune of $1.2 billion last year.
In the meantime, the telecom sector swooned
 |
| MEMS assembly in OMM’s class 100 clean room. Courtesy OMM |
“The fourth quarter will really be telling,” says Marlene Bourne, senior analyst with Cahners In-Stat Group. “The shakeout will start as carriers decide they are ready to buy.” Those startups that don’t sell products, don’t generate revenue and starve to death.
Two companies are shipping switches to customers, although in very limited volume, and an estimated 40 other companies are waiting in the wings — most of them young R&D operations that have yet to enter the Darwinian world of commercial markets and consumer demand.
This doesn’t mean, however, that the telecom meltdown hasn’t had some impact on the MEMS optical-switch space. “The industry was way ahead of itself,” says Rajiv Ramaswami, vice president of systems architecture for Xros and one of the leading scientists in the field. Difficult, more realistic times have taken the spotlight off the technology. Companies are now focusing on developing the switches and gearing up for manufacturing, says Bourne.
The shakeout has done something else: It injected a little reality into the analysts’ projections of how big the MEMS optical switch space will become. The word “billions” is no longer as freely bandied about. “The original estimations were exaggerated,” says analyst Lawrence Gasman, president of Communications Industry Researchers. Gasman estimates the market for MEMS-based optical switches to be $120 million in 2002, growing to $595 million by 2005. Other analysts, such as Bourne at In-Stat, place this 2005 figure as high as $1.8 billion.
Granted, that’s a big point spread. But even if the figure is somewhere in between, the revenue generated by the MEMS optical switch will be a boost to the MEMS industry in general, considering one of the largest and most mature MEMS markets, air bag accelerometers in cars, has been around 10 years and generated $1.26 billion in 2000, according to Roger Grace Associates.
Lucent Technologies and OMM Inc. are already shipping to customers in limited volume. Xros’ switch, which uses tiny mirrors to bounce (or switch) light as it travels down fiber optic cables, remains in trials but will begin shipping in the second quarter of 2002. Other big-name companies such as Corning Inc., Siemens AG and JDS Uniphase Corp. continue to work in the space, according to Gasman. Agilent Technologies Inc.’s product, which uses tiny reflective bubbles instead of mirrors to switch light, is currently in trials. While Agilent’s product does not have moving parts (one of the common definitions of MEMS), its technology is manufactured using traditional semiconductor fabrication technology.
But Lucent and OMM are the two leaders, at least in terms of the first-to-market advantage. In a fast-moving market, where demand is right around the corner, this is a good position to be in. But in a market that moves much slower, a company like OMM, which has built up its organization for volume production, could burn through valuable cash as it waits for demand to kick in.
While Lucent and OMM are both shipping all-optical MEMS-based switches, this is where the similarity ends. Lucent is shipping what are known as 3-D switches. OMM abandoned its 3-D endeavor and is now shipping 2-D switches, which are less complicated to build and control than 3-D switches. In the 3-D switch, the tiny mirrors are capable of moving in a complete circle — on two axes as opposed to one. What this means is that 3-D switches can handle very large switch fabrics. Or to put it another way, each MEMS mirror in a 3-D switch can do more work, requiring fewer of them.
In the end, the telecom slowdown could doom the entire MEMS-based switch industry. The longer optics remains in the doldrums and the longer it takes for the carriers to open their purse strings, the easier it will be for another technology to develop and compete head-to-head with MEMS. What kind of technology? One possibility is liquid crystals, which basically act like an old pair of Polaroid sunglasses, blocking out certain light waves and letting others pass through. Corning is the big player in this space, says Gasman, and has built a switch using the technology. Gasman, however, feels the drawback to liquid crystals is its inability to scale to large switches.
Bourne likes a technology called planar waveguides, which sends light down a polymer substance heated to change its refractive index. The leader in this space is a company called Lynx Photonic Networks, and the technology “will give MEMS a run for its money,” says Bourne.
The wild card in all this is Corvis Corp., which says it has developed an all-optical switch but refuses to say what technology it’s based on — although analysts say it’s not MEMS.
Other cutting-edge technologies are farther out on the horizon, such as using a solid state optical switch. In essence, an optical chip made from a substance like indium phosphide that has good optical properties. “The technology is very primitive,” says Gasman. For now, analysts say, MEMS is likely to remain the best bet in switching the light fantastic.