fibeReality has lately been discussing a lot about the collapsed ecosystem in the active optical equipment space, as well as the growing market for MultiMode Fiber (MMF) in the large data centers of the hyperscale players. In the passive fiber sector, there is a thriving ecosystem, with only a limited number of major suppliers globally, including Corning, OFS Fitel, Prysmian Group, and Sumitomo Electric Lightwave, but it is set up around the ubiquitous SingleMode Fiber (SMF) category. The vendors only have so many draw towers, etc., as well as operations to cable up SMF, which is experiencing very high demand, driven a lot by the hyperscalers themselves. In particular, Corning has its capacity in its worldwide facilities aligned to SMF, and in order to ensure it is adequately serving the needs of its customers, including the two that have been the biggest in the world, Verizon (counting in the massive 5G support effort) and AT&T, we anticipate that the supplier would be very reluctant to do any retooling for what would be a specialty solution. As happened in the past, when there was demand for different types of fiber, which were either mature in nature, required relatively limited quantities, or were just simpler to produce, and which also did not fit Corning’s business model, it has reached out to Fujikura Fiber to supply the products.
When it comes to MMF, despite being the largest supplier of OM3 to the US, even though at least some of it may be outsourced, Corning has made it clear that MMF is not a space in which it is anxious to be a player. For example, the company was pushing against standardization of OM5. (There has been some speculation in the industry that with Corning’s process, it is a little bit harder for it to make MMF, particularly at the high end, as compared with other competitors.) Also, understandably, the yields on MMF are not as attractive as on SMF.
In addition, as far as we can tell, demand for SMF is still significantly outstripping supply with the myriad applications: submarine, PON, and even terrestrial long-haul (for instance, we have heard that CenturyLink has been filling the “Level 3” conduits with additional fiber). Also, with the large number of dedicated 10G lines needed to enable 5G, Verizon is trying to figure out ways to get more fiber in its widespread, existing conduits. Therefore, unless a buyer has a secured supply, it is at the back of the line, and paying top dollar for SMF.
Historically, when it came to a specialty solution, that Corning did not want to include in its mainland plants, it used to first see if there was any available capacity in its little facilities in England, Germany and Australia. If there was none, it would cross-license the technology to Japanese vendors, especially Fujikura.
From around 1997 to 2001, Corning offered two main kinds of fibers, the conventional SMF-28 and the new LEAF. Although the processes were much more expensive to make the latter, the sales team did a good job of selling it, and so, the manufacturer built a new wing onto its largest draw facility to make nothing but LEAF – doubling the capacity of the plant with a total of 20 draw towers. Corning charged eight times more per kilometer than it did the SMF-28, while it did not cost that many more times to make the newer fiber – and it was sold out for 18 months.
In the meantime, Corning cross-licensed the SMF-28 to Fujikura. It was an attractive deal for the latter in that it was guaranteed the total share of Corning’s American business for that fiber type.
Another example is that in order to get PANDA fiber about 25 years ago, Corning had the patent. The company licensed it to Fujikura, when big demand later developed from Raman vendors, such as Qteras, and Corvis. So, Corning paid Fujikura to make the PM fiber as well.
Unlike Sumitomo (which has its VCSEL fab now in Yokohama, and is supposedly defining the device characteristics), Fujikura manufactures MMF (at its Sakura plant). In addition to OFS and Prysmian, the two other MMF vendors are YOSC in China, which is a big supplier of the fiber, and j-fiber in Germany, which is more on the low-end side.
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 .
[written by Mark Lutkowitz]