Ciena’s New Obsession with Vertical Integration

Avatar photoPublished on
blog 1486467768 74

Ciena’s purchase of several assets from TeraXion represents the first time that the system player has acquired just components as opposed to full systems from an outside vendor, symbolizing an evolving shift in philosophy over the last several years away from less of a reliance on merchant componentry suppliers to more internal development and construction. While large optical vendors in general have somewhat adopted vertical integration strategies as the market has moved to 100G and coherent technology, there seems to be a particularly higher degree of emphasis by Ciena on the practice in recent years, especially relating to added functionality on a chip set, which was apparently influenced by the purchase of the Nortel Networks division. However, a potential problem could be that the top leadership at the transmission equipment giant may be too lax about protecting its proprietary advantage with its WaveLogic 3 (WL3) device in favor of short-term concerns regarding keeping Wall Street investors happy.

In 2007, Ciena’s views against moving anywhere close to the Infinera model could not have been more pronounced. In the middle of that year, the VP of Strategy at the former stated the following at an investor’s conference: “…We’re so bullish on things like 100G because we see the convergence of telecom and datacom on a single type of interface as being a way to drive volume. If you have a proprietary vertically integrated solution, it is hard to drive that kind of volume.” In fact, before Ciena formerly introduced WL3 in March, 2012, it barely mentioned the first generation of the ASIC before that time, and it did not stress the WL concept until a year earlier.

In 2008, the President of Nortel’s Metro Ethernet Networks (ultimately to become a senior VP at Ciena) started touting its internal ASIC on the 6500 relating to 40G capability with the future promise of operating at the 100G rate. In 2010, the same year Ciena completed the acquisition of those assets, it hired Infinera’s VP of Product Marketing (who had been an executive with the former in the past), and he became SVP Marketing at Ciena, and he would later play an instrumental role in promoting the WL3. In advance of that effort, there was some foreshadowing of the dramatic change in messaging in September, 2011: “We have got a plan that we are rolling out over the next 18 months, which is basically to look at improved efficiencies across our complete supply chain, more strategic partnerships, and fewer vendors, basically, as we do that, a certain amount of vertical integration as well.” (On an interesting side note, despite the “Nortel” exec’s loyalty to the cause for over 25 years, he decided to take a COO position at a testing company last year, perhaps another sign of the historic tension between the higher-level engineering-oriented talent with the marketing/sales management running the Maryland-based firm.)

While Infinera may see the shift in focus by Ciena as vindication, the latter appears to still be buying its fair share of optical components (including for building line cards in-house) from external vendors, and we do not expect it to come even close to hampering its flexibility in its development of products by placing such a massive amount of its functionality on an integrated circuit. Actually, assuming that Infinera would have encouraged “Transmode” (after the closing of the buyout) with devoting an R&D effort with one of its customary six-month programs out of Sweden (rather than place so much focus on integration of the metro product lines), the PIC-based player could have around now possibly introduced a competitive pizza box for the Data Center Interconnect (DCI) space.

So, we believe that Ciena is in agreement with Cisco Systems (for example, the former is evidently still buying ROADM chipsets externally), when the latter stated last year, “We are not religious about ASICs, custom versus merchant….When we invest in an ASIC, it’s a $50 million plus investment, so we don’t take that investment lightly.” In addition, given that ballpark cost, the $32 million price for the TeraXion technology does not seem to be out of line.

As we mentioned above, the one aspect that is potentially troubling is the extent to which the upper management at Ciena could be willing to sacrifice its advantage with WL3 in order to get its quarterly numbers to look more attractive. fibeReality is still alarmed generally by the cavalierness of the vendor to engage in open line systems with DCI customers. Ciena has certainly been willing to use its valuable WL3 solution as barter with at least one large vendor to gain proficiency involving an ancillary capability that was needed in the past.

[written by Mark Lutkowitz]