Optical Communication Products Inc. (OCP) (Nasdaq: OCPI) has closed its Broomfield, Colo., facility, backing off from the market for long-wavelength Vertical Cavity Surface Emitting Lasers (VCSELs).

OCP presaged the move in its 10-K report filed with the SEC in December. "We expect to cease the operation of our Colorado facility as of January 31, 2006, because the cost of Fabry-Perot lasers has been decreasing, making 1300nm VCSEL technology less attractive as a cost-effective replacement, and because of the delay in development of the next generation of optical modules," the report reads.

The company got a headstart on the closure by trimming down the site earlier this month. "Right now, it's just a skeleton operation that's shutting down that Colorado facility," says Kirk Bovill, OCP vice president of marketing.OCP is retaining the Cielo intellectual property "for future development," according to the 10-K.

The Colorado closing isn't necessarily a bad sign for the rest of OCP. The company has reported two consecutive quarters of profits and holds a whopping $148 million in cash. For its fourth quarter, which ended Sept. 30, OCP reported profits of $1.6 million, or 1 cent per share, on revenues of $14.8 million.

Considered a cheap, efficient alternative to lasers commonly used in telecom, VCSELs were a hot item in the bubble years around 2000. Making them tunable, making them run at 10-Gbit/s, making them operate at longer wavelengths -- all were startup plans that earned VC money for the likes of Bandwidth9, Cielo Communications, Nova Crystals, and Siros Technology (all defunct).

The VCSEL market also attracted big companies including Alcatel (NYSE: ALA; Paris: CGEP:PA), Honeywell International Inc. (NYSE: HON), and Infineon Technologies AG (NYSE/Frankfurt: IFX).

OCP got into long-wavelength VCSELs with the acquisition of Cielo's assets in October 2002 for $5 million. At the time, several companies had produced 850nm VCSELs for short-range uses, but the output power on VCSELs in the 1310nm range still wasn't strong enough, creating a perceived opportunity for startups. Moreover, Cielo packaged its VCSELs into an array that reportedly sold well for OCP at first. (See OCPI Acquires Cielo, Cielo Pushes the Limits on Lasers, and OCP Achieves VCSEL Milestones.)

Today, several companies offer 1310nm VCSELs, but they haven't found the success that the 850nm generation enjoys.

"I think people make the assumption that what happened at 850nm will happen again," says Tom Hausken, an analyst with Strategies Unlimited . "It doesn't necessarily happen. The opportunity has gone past us."

OCP, which continues to hold an 850nm VCSEL line acquired from Gore in 2003, isn't the only company to step away from long-wavelength VCSELs. Last year, Infineon sold its 1310nm VCSEL line to Alight Technologies A/S . (See OCP Buys Gore's VCSEL Biz and Alight Buys Infineon VCSEL Platform.)

The VCSEL's luster hasn't completely faded, as the parts can still hold a price advantage against distributed feedback (DFB) lasers. Picolight Corp. recently introduced a 4-Gbit/s, 1310nm VCSEL aimed at storage networks. (See Picolight Ships VCSEL.)

In 2004, Finisar Corp. (Nasdaq: FNSR) bought up a Honeywell division that was building long-wavelength VCSELs, and JDSU (Nasdaq: JDSU; Toronto: JDU) got into the game with the 2004 purchase of startup E2O, a transceiver vendor that made its own VCSELs. (See Finisar Takes On Agilent and JDSU Buys E2O.)

Even the array idea is still alive, with VertiLas GmbH announcing plans last year for a 1550nm VCSEL array. (See VertiLas Demos 1550nm VCSEL Array.)

— Craig Matsumoto, Senior Editor, Light Reading