ANAHEIM, Calif. – The optical recovery hasn't brought sweeping new product categories back to OFC/NFOEC yet, but at least the business side of the business is back.

The show remains relatively small, with 565 exhibitors this year, but the mood was worlds apart from the last few years. (See Smiles Abound at OFC/NFOEC.) More companies talked of being profitable, and a bigger count of analysts and investment bankers seemed to be prowling the floor. Their trip out here was rewarded by a stock spike just before the show, giving everyone plenty to talk about. (See Optical Stocks Climb Again.)

That includes talk of IPOs. Transceiver vendor Opnext Inc. (Nasdaq: OPXT), last year's favorite to go public, might have disappointed by staying quiet. Plenty of other candidates appear to be in the wings, though.

"I count, at the component level, five or six [private companies] that are well over $10 million a quarter," said Richard Craig, CEO of Santur Corp. , speaking at the The Optical Society (OSA) (OSA) Executive Forum 2006the day before OFC/NFOEC. "Two or three of those companies are already starting to talk IPO. You'll probably see it late this year or early next year."

Craig wouldn't make any picks, but he noted that a certain unnamed tunable laser company saw revenues jump 400 percent each of the last two years. (Did we mention Santur makes tunable lasers?)

Santur isn't leading the IPO buzz, though. Most sources at the show pegged Opnext or Optium Corp. (Nasdaq: OPTM) as the most likely candidates; neither company openly verified any plans, naturally. Opnext, a Hitachi Ltd. (NYSE: HIT; Paris: PHA) spinoff, was said to exceed $100 million in revenues last year. Optium Corp. (Nasdaq: OPTM) has quietly built a solid business with 10-Gbit/s transceivers and just this month made the move into reconfigurable optical add/drop multiplexers (ROADMs) with the acquisition of Engana Pty. Ltd. (See Optium to Acquire Engana.)

Further down the list was Fiberxon Inc. , a company, based mostly in China, with its eye on a Silicon Valley-like IPO. (See Fiberxon Mulls IPO.) Chief executive Li Hsu says Fiberxon has been profitable for 12 quarters in a row, with revenues hitting $39 million in 2005.

On the product side of OFC/NFOEC, the recovery isn't old enough for crazy new ideas to come forth yet, although Lambda OpticalSystems Corp. is reviving talk of the 256x256 all-optical switch. (See Lambda Optical intros OXC.)

Overall, this year's product parade looked a lot like last year's, maybe with smaller sizes and improved specs. Here's a look at some of the hot topics of discussion.

40 gig. Yes, again
Are you tired of this one yet? Vendors are continuing to report progress in OC768/STM256 networking -- 40-Gbit/s links -- becoming a real market. Sort of.

"There's interest, but let's be realistic -- it's not hundreds of millions of dollars. You're talking $10 million, maybe, in the next year and a half," says Todd Swanson, vice president of marketing for Finisar Corp. (Nasdaq: FNSR).

Nearly all the 40-Gbit/s work being done is in very short-reach links, connecting routers or switches to each other inside a data center, but the technology is seeping into the network. Lucent Technologies Inc. (NYSE: LU) announced (at CeBIT, technically) a field trial with Japan Telecom Co. Ltd. and says it's got two 40-Gbit/s contracts signed, although neither customer is ready to deploy the technology yet. (Lucent officials would neither confirm nor deny whether Japan Telecom is one of the two.) (See Lucent Spews CeBIT News.)

Bigger numbers are on the horizon. CyOptics Inc. is developing 80-Gbit/s links for a Cray supercomputing project. And Lucent reported it's run 100-Gbit/s Ethernet in the lab using 40-Gbit/s components and duobinary modulation.

To Page 2

10-Gig evolves
Talk of 40-Gbit/s is possible because 10-Gbit/s links are more widely adopted. But when it comes to 10-Gbit/s Ethernet, module vendors are still getting hit with criticism that prices haven't fallen fast enough.

"How do you get 10-Gbit/s volumes to ramp up similar to 1-Gbit/s?" Finisar's Swanson asks. "Some of the [telecom] standards out there are so rigorous, it's hard to have a low-cost part."

That's because telecom standards plan for worst-case scenarios. It might be worthwhile to back off on that in exchange for some cheaper parts, Swanson suggests: "The problem [otherwise] is you've just spent two to three times what you needed to for 99 percent of the applications."

Among the hot 10-Gbit/s product groups were transceivers for the 10GBase-LRM standard, which uses electronic dispersion compensation to send signals 220 meters down multimode fiber. Being crafted by the Institute of Electrical and Electronics Engineers Inc. (IEEE) 802.3aq task force, it's an alternative to the four-wavelength LX4 standard -- one difference being that LX4 works on 300 meters of fiber. (See Vendors Still Driving LX4.)

Shrinks
Here's an idea that never went away: the planar lightwave circuit (PLC). The idea is to pack optical components together onto a chip, usually made of indium phosphide (InP), to gain the integration advantages that normal semiconductors enjoy.

With recovery blooming, PLCs could get big again, particularly in the next wave of devices. "One of the key factors in buying Agility [a tunable laser vendor] was the monolithic technology," says Mike Ricci, senior vice president of the JDSU (Nasdaq: JDSU; Toronto: JDU) components and modules group. (See JDSU Tunes In Agility.)

(Agility CEO Ron Nelson will be leaving JDSU at the end of the month, by the way. His plan? Golf and surfing. Maybe he'll develop a monolithic integration of the two.)

When it comes to shrinking part sizes, most attention goes to transmitters, but amplifiers are getting smaller as well. Avanex Corp. (Nasdaq: AVNX) displayed an erbium-doped fiber amplifier (EDFA) packed into a Xenpak module -- a size and shape that was intended for 10-Gbit/s transceivers. Avanex claims it's the first pluggable EDFA to hit the market. (See Avanex Intros Xenpak EDFA.)

Aside from the novelty, Avanex thought it important to show the product as a sign that the company's R&D wasn't dormant during the company's restructuring. "I had customers really challenging me about that," says Yves LeMaitre, vice president and general manager of Avanex's optical components division.

Bend and flex
We're saving this one for last because it's got the most burdensome collection of links. All the talk of tunability taking over the network appears to be coming true -- it's just that it took several years of hardship and dozens of startup evaporations to happen.

Several executives, particularly the Avanex, Bookham, and JDSU crews, pounded the table for the concept, noting that transmitters are on their way to being tunable by default, and that ROADMs are in demand for new WDM buildouts. They describe the resulting optical network as a more flexible one, able to tune up wavelengths on demand.

OFC/NFOEC saw quite a few announcements related to tunable lasers, ROADMs, or the little piece-parts that go inside them. Here's a laser-related list:

...and the ROADM roster:

— Craig Matsumoto, Senior Editor, Light Reading