Transpectrum Flunks Out

The heyday of the "research-product turned optical-startup" may be coming to an end. Los Angeles startup Transpectrum is the latest such operation to shut its doors, shelving its ambitious plans for 10- and 40-Gbit/s transceivers.

Founder Behzad Razavi confirmed that the company has shut down but wouldn't say why -- although a good guess might be a lack of faith from investors, considering the dim short-term prospects for new optical networking technologies. Razavi wouldn't comment on who had funded Transpectrum, and the company's Website doesn't mention any venture backing.

Transpectrum was born of Razavi's research at UCLA, where he is a professor of electrical engineering, specializing in high-speed analog circuits.

"It was not a good experience, that's for sure. I don't think I would want to do it again," he tells Light Reading.

A little late to the party, Transpectrum launched in March 2002 with promises of a 40-Gbit/s transceiver -- sporting four lanes of 10 Gbit/s apiece -- by the third quarter of this year (see Transpectrum Demos Transceiver). The company's specialty was in semiconductor process technology -- the ability to coax fast and clean electrical signals through complementary metal-oxide silicon (CMOS).

That's important because CMOS is an inexpensive and well-understood semiconductor technology. Many high-speed components have to be built of more exotic materials such as indium phosphide or silicon germanium. In those cases, manufacturing is more difficult, harder to find, and more expensive. Moreover, CMOS devices can be merged onto a single chip, offering the promise of future cost and power savings.

But ambitious high-end technology has a tough time finding a home in this market. The short-term prospects for such products are weak at best. And by the time demand recovers, any product may have lost its usefulness, either because market requirements have changed or because a competing technology has advanced.

Vitesse Semiconductor Corp. (Nasdaq: VTSS) used that logic earlier this year to axe several high-end projects, including 10-Gbit/s packet-processing chips (see Vitesse Drops Some Packets). Vitesse CEO Louis Tomasetta reiterated his reasons on yesterday's earnings call with analysts, noting that by the time the market recovers, any high-end products designed today would likely be obsolete or irrelevant.

Could Transpectrum have pulled it off? We'll never know. Plenty of ideas -- both deserving and undeserving -- are being dragged down in the telecom collapse, and even those that resurface may find themselves strangers in a brutally changed industry. In fact, the end of the bubble appears to be driving more scientists away from Wall Street and back to the lab -- which may not be such a bad thing (see ECOC: Back to the Lab)

— Craig Matsumoto, Senior Editor, Light Reading
www.lightreading.com