As with fibeReality, both Finisar and Oclaro have problems foreseeing 400GbE in any meaningful way in the anticipatable time ahead. While each is more than glad to mention development efforts in this direction, their powerful statements included in a couple of recent articles on Gazettabyte cannot be dismissed. An executive from Finisar said, “There is probably more technical risk in 400 gigabits than any of the historical data-rate jumps we have seen.” The Chief Commercial Officer from Oclaro said, “The industry really wants four channels. When you use more lasers, you are adding more cost." In addition, with the expectation of NG-PON2 becoming the de facto standard for at least many large portions of the globe, 25G will even have longer legs. Any vendor that can produce a low-cost 25G laser will be a big winner in the future optical access space. All in all, it is not even that easy to accurately predict the arrival of just 200GbE.
Regarding the comment from Oclaro, it would eliminate any product standard or MSA except for 4x100G. We have discussed in the past that one is starting to push the envelope with technological challenges even at serial 50G, and continual price drops on 25G will inevitably stall its usage at least to some extent. So, the arrival of a single-lane 100G, which is commercially viable, and cost-effective, is hardly around the corner.
The declaration from Finisar is even more compelling and extraordinary. Very large investments were required (with barely any venture capital involvement) by both optical component and system firms to get to the current situation with 100G. It was scarcely a cakewalk, with the development of several form factors, etc., especially in comparison to the bare-bones technical requirements of 10G -- just short-reach and long-reach devices. Moreover, there has obviously been a significant amount of spending to produce 40G componentry as well in which, as with the other data rates, a full return on all investments would have been presumed, but not necessarily fully accomplished with any of them.
It was not surprising that Lumentum Holdings expressed more optimism in the Gazettabyte article than its two large rivals about getting to 400G. The supplier is being heavily evaluated these days on the basis of its perceived leadership position in providing VCSEL bare dies for Apple’s iPhone X. However, in our opinion, there is a growing amount of evidence beyond what we have expressed previously in that it has been a struggle for Lumentum to adequately produce the very large quantities demanded by the smartphone vendor, and goes much further than its characterization in its last quarterly earnings report (“bottlenecks in equipment capacity which have since then resolved.”)
NeoPhotonics, which has been certainly under a great amount of pressure given its recent performance in the market, also seems to provide a less pessimistic outlook than Finisar and Oclaro on 400G, including to us at ECOC.
Yet, the judgement offered by a director from Vivavi Solutions in the Gazettabyte piece, which as a test supplier, is based on a more neutral point of view, is quite instructive: “He describes as significant the challenges involved in developing a four-wavelength 400-gigabit design. These include signal integrity issues, the optics for 100-gigabit single wavelengths, the PAM-4 DSP, the connectors and the ‘insanely hot and hard’ thermal issues.”
Despite Lumentum’s supposed optimism about PAM-4 in getting to 400G on the client side, according to Vivavi: “The use of PAM-4 requires a slight increase in bandwidth…and introduces a loss that requires compensation using forward error correction…. 'Four-hundred-gigabits is the first Ethernet technology where you always have FEC on,'” We have expressed similar concerns about PAM-4 in the past. In general, the situation really has not changed very much from back in April 2015, when we wrote: “400GbE Spectacular Fantasy.”
[written by Mark Lutkowitz]