Network engineers were salivating. Their vendors were boasting of all-optical (or “photonic”) switches capable of catapulting up to a staggering 1000 wavelengths from destination to destination without the need for electrical interference. They touted the use of 3D MEMS (micro-electro mechanical systems), liquid crystal display (LCD) or even bubble-jet technology to send traffic back and forth, and said that with the electronics gone, the associated headaches and high operational costs would disappear as well.

But something went wrong. A combination of overbuilt long-haul networks and shredded capex budgets has taken its toll on all-optical switching. Despite spending exorbitant amounts of money to develop or acquire all-optical switches, most large vendors have now bailed out of the all-optical biz and put the technologies on hold.

The challenge now is to pinpoint exactly how so many vendors went so far off course — and determine what's right when and if service providers feel the desire and have the luxury to spend again.

“We see absolutely no growth [in all-optical switching],” said Kathy Szelag, vice president of marketing for Lucent Technologies. “Carriers aren't spending money there. No one can afford to keep a core network build going.”

For Lucent, sales of its all-optical switch, the MEMS-based LambdaRouter, were so poor that in August the company stopped production. Around 100 employees were also shed.

Lucent had gained little traction with the product before its demise. Its only customers were Global Crossing and Japan Telecom, and in both cases, few switches were actually purchased. Japan Telecom, for example, had equipped only three nodes with the switches.

Yet Lucent isn't alone. Although Nortel Networks paid $3.25 billion in stock for all-optical switch start-up Xros Communications back in March 2000, the vendor stopped the bleeding and halted production of its OPTeraConnect PX switch earlier this year, a move that was also pinned on changing market conditions and limited sales prospects.

What was wrong with those big switches? Like the Lucent system, the Nortel switch was MEMS-based, which limits providers to extremely large long-haul switches. Alcatel and Ciena had both been working on developing MEMS-based switches as well, but didn't get in as deeply as their competitors.

“The market dynamics to drive all-optical never materialized,” said Tom Mock, Ciena's vice president of portfolio management. “The idea of switching wavelengths all-optically was predicated on a large number of cases sufficient to manage on the wavelength level.”

Even though vendors like Lucent and Nortel are pointing fingers at lack of interest and the market downturn, others believe the root of the problem lies in the technologies themselves.

“A number of companies are backing away from the all-optical market, and the primary reason is that they built the wrong type of switch,” said Dave Smith, vice president of engineering at Corvis.

The benefits of all-optical switching vs. the use of electronics are heavily weighted toward all-optical, according to Smith — factors such as the cost of the switching infrastructure, operational issues like power and size, and functionality are all superior with all-optical implementations. “Lucent took away the cost benefits and they took away the functionality benefits [by surrounding the switch with electronics],” Smith said. “And MEMS turned out not as mature and capable as was thought.”

Corvis' LCD-based switches are designed to handle far smaller numbers of wavelengths than the 1000×1000 port switch Nortel had.

That said, Corvis hasn't seen its sales explode, either. It made a significant win with Broadwing Communications, which based its national network on Corvis equipment, and notched wins with others such as Williams Communications and Qwest Communications. But spending with those providers and all others in the industry has halted.

Still, Smith maintains a market for all-optical switches does exist.

“Telcos will realize they have to show they can turn a profit,” he said. “Most are stuck with business as usual and will begin to look at making a profitable business out of long-haul.”

Charles Corbalis, CEO of Calient Networks, agrees. “There are absolutely no service providers saying they don't want all-optical switches. They say they aren't buying capital equipment.”

Calient is likely hoping to take up the leads left cold by other developers of the technology. Japan Telecom, for example, had deployed Lucent's LambdaRouter and is now trialing the Calient switch.

“Companies like Lucent and Nortel have to focus on their core competencies so they can maximize sales and minimize loss,” Corbalis said. “It's coming back to where new technologies are introduced by start-ups that don't have Wall Street and shareholders to please.”

Lucent and Nortel are now directing customers to other products. “Some carriers may start with OADMs (optical add/drop multiplexers), but that will be slow,” Lucent's Szelag said. “Most will stick with point-to-point.”

And given that market conditions will eventually change, and providers will again resume buying equipment, the future may not be as bleak as it seems.

“Service providers aren't saying never,” said Van Steen. “They are just saying, ‘In a couple years.’”