Karen Liu
Opnext surprises market with 100G transport IC
The road to higher-speed optical transport has been broadened, thanks to Opnext's announcement of its first 100G integrated circuit (IC). Later this year, Opnext will follow this announcement with the remainder of the critical chipset, which will enable Opnext to supply merchant transponders, opening the 100G market beyond those few OEMs that have internal ASIC development. This move is a positive sign for costs and volumes in the 100G market and represents a new IC-OC business model.100G needed urgently but chipset is a key cost element and barrier to entry
Opnext announced it has developed a 100G mux chip for optical transport. It will follow this announcement with the rest of the signal-processing capability in another chip. It will sell a merchant transponder; it has no plans to sell the chipset.Although this chip is only a small part of the solution, we are excited about its implications for 100G availability, cost, and industry health.With 40G still in the early stages of commercial deployment, carriers are already clamoring for 100G. In its urgency, the industry has coalesced around one approach: coherent optics plus high-speed, electronic digital signal processing (DSP). System and component vendors alike expect 100G to proceed through trials to deployment faster than 40G did.With 100G, a disproportionate amount of new technology must be developed for a relatively modest 2.5x increase in speed. The cost target is correspondingly modest compared to 40G, while the implementation is significantly more complex. The technical difficulty and materials cost are centered on the high-speed chipset. There are no merchant chipsets or transponders available today. Electronic signal processing is not a core competency of the optical industry. At 100G, it does not exist in any other industry either.Competition in 100G will heat up with merchant transponder availability
Nortel was the only OEM to adopt the coherent plus high-speed DSP approach at 40G. Other vendors looked askance at Nortel's five-year investment in its 40G chipset. But Nortel had first availability in 40G and offers the only commercially available 100G system today. Huawei is currently trialing its product, which is also based on internal chips.Once transponders are available, more OEMs can enter the 100G transport market. Each will claim differentiation in the algorithms programmed into the DSP but the 100G market will not be as fragmented as the 40G market. Broad availability of similar systems will result in continuous price declines for 100G systems.Interest in 40G coherent technology is rising. The Opnext chipset can be run at multiple clock rates, supporting both 40G and 100G and varying levels of forward error correction (FEC) overhead.Combined IC-OC is a robust new business model for OC companies
We are closely watching Opnext's approach as an example of the combined optics and electronics model that we highlighted in our 'Telecoms in 2020: components' report (published in December 2009). The OC industry has been frustrated that its optical technologies have not shared in semiconductors' inexorable advances. Vendors have tried multiple times to tap into Moore's Law: integrated optics, silicon optics, and, historically, Vitesse's and Triquint's entries and subsequent exits. This time, two favorable conditions may help the outcome:- The chosen 100G scheme places the technical burden on the electronics. It can actually tolerate lower-performance optics than the dominant 40G schemes.
- The fabless semiconductor model allows companies to compete based on application and design expertise. Opnext acquired the required transmission, digital IC, analog IC, and RF packaging competencies with its StrataLight acquisition. The announced chip is SiGe, the follow-on chip is CMOS — both are produced in commercial fabs.
