Cisco announced today its intent to acquire silicon photonics startup Lightwire for approximately $271m in cash. Lightwire’s technology reduces power consumption for data center and metro optical links with a novel approach to both laser modulator and driver. The acquisition hits several key themes we have highlighted before: system houses vertically integrating to own differentiating chip technology, power consumption as a hot button, silicon photonics and photonic integration entering the mainstream, and innovators finding fertile ground at the intersection of photonics and electronics.
Cisco CEO John Chambers stated that “in the long run, Huawei is the company’s toughest competitor. Huawei will always compete on price.” Huawei also has internal optics development. We believe the Lightwire acquisition will allow Cisco to compete on performance parameters such as power consumption as well as price. In the telecom transport arena, other OEMs such as Alcatel-Lucent, Ciena, and of course Infinera increasingly vertically integrate for differentiation.
Cisco backs up its transceiver value proposition
The Lightwire organization will become part of Cisco’s Transceiver Modules Group Business Unit and Supply Chain Operations Group. This group resells merchant transceivers with Cisco-specific features. Along with its compatriots, Cisco drove MSAs (multisource agreements) to make transceivers from multiple suppliers interoperable and price-competitive. In striking contrast, Cisco will now have its own differentiated optical technology at 100G for both long-haul (CoreOptics acquisition in 2010) and datacom/metro (Lightwire). This change reinforces our view that 40G/100G has opened the door for dramatically new technologies to enter.
The price difference between Cisco-approved transceivers and vanilla pluggables has fueled customer disobedience, resulting in a sizable gray market for non-approved transceivers from low-cost suppliers. We believe with differentiating technology Cisco can now point to the obvious superiority of its transceivers and thus continue to control the supply of optics into its switch ports.
Cisco commitment speaks to importance of hardware technology
It once seemed that Cisco and other system vendors valued software and the intelligence it enabled above hardware. With system designs up against heat dissipation limits, low-latency networks bounded by physical distance, and datacom optics straining to hit 40G and 100G, we are now in an era when having leading hardware matters again. In telecom optical transport, Cisco – along with Alcatel-Lucent and Ciena – has its own coherent 100G ASIC.
Cisco is one of the few players with the scale in Ethernet switch ports to exploit vertical integration for cost reduction. However, Lightwire’s technology takes a holistic approach to the optical and associated electronic components, unlike today’s typical architecture which separates the two. Hence we see the greater value for Cisco is to differentiate on superior power consumption, density, and, in future, speed.
Fabless IC model makes vertical integration possible
Cisco’s CoreOptics and Lightwire “optical” investments are still a far cry from buying a laser fab or transceiver operation. Lightwire is fabless, relying on external foundries but differentiating on novel device design. This is the same model system houses use for digital electronics ASICs. Both CoreOptics and Lightwire rely on purchased lasers (Lightwire makes an optical modulator but not the laser). The transaction is neutral or favorable for merchant laser vendors, including CyOptics.
What happens to Cisco’s transceiver suppliers?
Cisco dominates the datacom transceiver landscape, dictating vendor roadmaps. Cisco could now take its Lightwire chips to contract manufacturers to make its own, bypassing its transceiver suppliers. But it could have done so before too. Huawei has continued to buy from its component suppliers even while its vertical integration gives it additional options. We believe Cisco will continue to work with its existing favored vendors as we see no compelling reason to replicate the effort it has already put into infrastructure and process development with them. Internal development of optics does give visibility into underlying cost structures for increased negotiating power, but Cisco is already very familiar with transceivers.
Validation of photonic and vertical integration for telecom and datacom
We believe that Cisco’s in-house capability can accelerate silicon photonics and integrated opto-electronics acceptance. Component vendors need to work with a strong lead customer for new – particularly radically new – products. Infinera leapt ahead of the industry with optical integration by being its own customer. A further direction for speculation is whether Cisco will eventually join computer vendors HP, Oracle, and IBM in photonics for chip-to-chip interconnect.
Cisco
