COBO’s Scheme: Odd Life/Demise Likely

Avatar photoPublished on
Blog 1535639434 215

fibeReality expressed our point of view that if COBO ever happens, it would be in the form of Co-Packaged Optics (CPO). In our opinion, it is also instructive to closely examine both COBO’s historic relevance to the present, crumbled optical ecosystem, which we were anticipating right up to the precipice. We believe this situation was almost solely created by the detrimental mentality of Microsoft, as well as other hyperscale data center operators. These players might be shocked by the increasingly derogatory remarks we are hearing regarding even the intellect of their network planners and engineers from executives at major optical component companies. Brownie points are only given out at the procurement departments of the Microsofts for bringing in say, transceivers for ten cents less, enabling countless competitors with lower-cost solutions to enter the space, while seemingly ignoring that money is being taken out of the pockets of the big componentry suppliers. This capital would have funded new developments, such as for 400GbE, but instead there became self-fulfilling circumstances. As far back as March 2015, we wrote: “With both system vendors [increasingly encouraged by hyperscaler demand for open line systems] and little component firms…cherry picking the most lucrative opportunities, how does that prevent the major players in optical componentry from getting squeezed out with greater commoditization of hardware vital to their businesses? Who is going to do the bulk of the research for the next generation of optoelectronics devices?” Yet, at the same time, it was expected by Microsoft (along with Facebook now on CPO) that these same firms, which made up the core of the ecosystem should somehow be able to also finance pet projects, such as COBO. We will also discuss the peculiar nature of the COBO concept, which we would contend impossibly strived to provide the “Holy Grail” solution, similar to the overall view on solitons in the distant past.

Adding insult to injury, our recollection is that a fairly confiscatory $10,000 from companies were required (we understand that it was eventually dropped) to even be a member of COBO. From a public relations perspective alone, suppliers were between a rock and a hard place, if they refused to demonstrate at least a minimal amount of interest in a concept being pushed by a firm with a high profile, such as Microsoft. At least Oclaro had the courage to take a public stand (which undoubtedly had the quiet support of Finisar, Lumentum, and NeoPhotonics) that it would not be financially viable, especially after all of the investment in 100G, to be expected to make R&D expenditures with on-board optics.

Although early on, Microsoft was not publicly bullish about CPO being the next revolutionary step, we have had no doubt that it would ever internally be married to a single solution. From the beginning, we were not optimistic about COBO’s arrival, stating that it could be a many as a couple of decades into the future before it came to fruition. Moreover, there did not appear to be support for the COBO’s development of a specification from the other big hyperscalers, including Google and Amazon.

Yet, it is important to note that in highlighting COBO for the upcoming ECOC event next month, Microsoft refers to it in a “future-proofing,” context, a term, which should be eliminated from any practical analysis about technology. We have commented before that the corporation from a network point of view is “somewhat of an enterprise trapped in a traditional carrier’s body and mindset,” and a major characteristic of the latter is a desire to obtain as many requirements on a wish list as possible. We think that the exceptionally ambitious nature of COBO on size, reach – and on accommodating future data rates, reflects this attitude.

Of course, usually, there is an inevitability of smaller and far less complex technology resulting in lower costs. There was also always the strong possibility that indefinitely, there would be an insufficient business model to move from quite a cottage industry to one of mass production, and so the expectation of sales of transceivers being affected has always been in doubt.

As always, fibeReality does not recommend any securities, and this writer does not invest in any companies being analyzed by us.

To follow us on our totally separate, quick update company blog, which is exclusively on fibeReality’s LinkedIn page, please click here.

[written by Mark Lutkowitz]