Mellanox Technologies is a typical Israeli-based firm in the datacom/telecom space in that it excels at technology development, but not necessarily from a marketing/sales perspective. While the vendor has expanded internationally and has its second headquarters in the US (including several of the top CXOs), which is potentially helpful to corporate communications, it appears that much of the culture still reflects Yokneam rather than Sunnyvale. Nonetheless, although getting its fair share of appreciation for its technical strengths may at times be problematic for Mellanox, its rather consistently positive financial performance as well as its impressive installed base of customers speak volumes – and when the supplier happens to place an emphasis in its messaging in a potentially questionable way, insufficient outside attention may actually be quite beneficial.
A good illustration of the lack of impact by Mellanox’s publicity efforts is regarding being the principal player in the huge jump in sales of 40GbE NICs in 2015 with continuing growth for an indefinite period of time. The common and false narrative in the marketplace that has taken place for several years, especially promoted by a good number of industry analysts, is that 40-gig has been on its way out, particularly in public networks. While there is no detrimental effect to Mellanox, it is illustrative of the lack of resonance regarding its activity because logically, if there would be widespread acknowledgement of the spike-up in 40G user interfaces, the effect on the line side at that date rate cannot be ignored.
An ordinarily more problematic case would concern Mellanox’s insistence in going against the grain with its attempts to continue to rather heavily promote its development in silicon photonics (SP). The firm has even recently appointed a VP of SP. Mellanox is pushing this agenda despite the overwhelming amount of criticism of these products that we have cited in the past. (Moreover, Google, which probably does more exhaustive cost analysis backed by practically unlimited resources to study every conceivable technical possibility for its network, ultimately decided against going in an SP direction.)
While Mellanox acknowledges the shift in the public characterization of SP that also now involves mainstream engineering publications taking on a more cynical view, it maintains that it is confident about its future success with its solution and that its photonics chip is unique in using a Franz-Keldysh modulator, stating it is very clean as well as 10 to 100 times smaller than competitive solutions. The vendor also says that the modulation is at a very low cost because an older node that is close to 100nm is employed, which has been depreciated, allowing for both 25G and 50G capability. In addition, Mellanox argues that it would not benefit from an integrated approach that would include its sophisticated electronics because it could have taken as long as five years of development with a very small payback at best – or even a loss.
It has becoming increasingly clear that the vision of having one pure SP chip has evolved based on the economics for just about all of the players in the market to a point that the original intent is meaningless and the future use of the term in the foreseeable future will be at least somewhat disingenuous. It becomes much easier to see how Mellanox can make an “SP” solution work that just concentrates on the following three core capabilities: 1) detecting and modulating the light; 2) splitting it into four parallel strands of light in a passive waveguide; and 3) diffracting the light that has been received and then recombining it on the transmit side for WDM. Compared to doing these functions discretely, the use of a photonic chip is a no-brainer – and the vast majority of observers in the space will undoubtedly buy the inevitable conclusion that Mellanox triumphed again, despite the overwhelming challenges associated with “silicon photonics.”
At the same time, despite numerous presentations and technical publications on the subject, Mellanox is not identified much with the concept of SP – another example of its lack of prowess on the marketing side. Yet, it might lead to the best of all worlds, in that the supplier will avoid a lot of the negative prejudice presently associated with the overall lack of consensus on the technology.
[written by Mark Lutkowitz]
(To read InnoLight IPO: Google’s Vertical Integration Without a Mess, please click here.)