Silicon photonics, one year later

By Stephen Hardy

One of the bigger stories going into last year's OFC/NFOEC was the commotion started when Jefferies & Co.'s James Kisner downgraded Finisar's stock in light of the impending impact of silicon photonics technology on the data-center interconnect market. Intel had recently revealed its silicon photonics efforts were about to reach the engineering sampling stage, and Cisco was expected to unveil the first fruits from its acquisition of silicon photonics developer Lightwire at the show (which, in fact, it did). Add these events to the fact that companies such as Luxtera and Kotura (now part of Mellanox) had already produced silicon photonics engines, and it seemed the era of silicon photonics had arrived.

Kisner argued the potential cost benefit of silicon photonics versus conventional approaches was so compelling that companies that didn't possess this technology yet -- like, say, Finisar -- would quickly become technological dinosaurs. So down went Finisar's rating.

One year later, Kisner has weighed in again on Finisar. This time he's raised his rating from Hold to Buy, based on his assessment of the opportunities for Finisar in the data-center market. Is that because Finisar finally jumped into silicon photonics? Well, no -- silicon photonics won't begin to play a meaningful role in shaping the price of optical interconnects until sometime next year "at least," Kisner now reasons. Meanwhile, there's enough momentum in transceiver demand at 10 Gbps and below -- not to mention opportunities for transceivers and other products in carrier networks where silicon photonics' role is murky at best -- to make Finisar stock an attractive buy.

As if to pour more water on last year's silicon photonics frenzy, fellow financial industry analyst Simon M. Leopold of Raymond James issued a note shortly after Kisner's missive that Leopold entitled "Silicon Photonics: Not Ready for Prime Time Yet." Leopold argues that silicon photonics likely will play its most significant role in very short reach data-center applications and may have an impact in long reach links but will need some time to mature. Meanwhile, companies like Finisar will have plenty of opportunity to acquire silicon photonics expertise if/when it becomes necessary.

The reduction in expectations for silicon photonics is a welcome occurrence (except for companies in the space looking for financing). Optical communications technology development long has suffered from too much hype, followed by disappointment -- and punishment -- when reality intervened.

Which is not to say that silicon photonics has lost any of its true promise. We should see at least some of that promise come closer to fruition at OFC/NFOEC this year. Conference presentations, of course, will provide the technical details of ongoing silicon photonics development activities and achievements. On the exhibit floor, we'll see some of these achievements realized in the form of optical engines and transceivers, both from established players like those mentioned above as well as the several startups hoping to become established themselves.

That means we can make conclusions about the technology as we see it in action. What a novel concept.

[email protected]