Finisar’s current business model based on the all-encompassing optical components approach in the datacom/telecom market with a large amount of vertical integration from chips to subsystems, along with designing every conceivable flavor demanded by the space of a particular data rate, has become unsustainable. It is bad enough in a relatively slow growing business, particularly with 100G, to need to start developing the next form factor, so quickly after the supplier releases the last one, but also there is competing in a highly fragmented arena, especially against a lot of Asian manufacturers, which tend to put low prices way ahead of profits. Then to also have some of the major customers in recent years become competitors (either directly or indirectly) – probably makes it sheer insanity to continue such an operation.
With Jerry Rawls taking over the reigns again at Finisar, and after about 25 years of being in a leadership position there, one may guess that he had more than enough of a situation that will only get worse. Assuming he is successful in getting margins back up to a more respectable level (in hopefully no more than two or three years), it will probably be advisable to get his company out of these miserable conditions through a combination of divestments and spinoffs – and not have to personally deal with these headaches anymore.
While Finisar could try to imitate Avago Technologies’ strategy of shifting just to higher margin semiconductors (fibeReality was the first news source to anticipate as well as to report this event), we believe that the former is in a different situation in that it is totally an optical supplier, and that being so far out on the learning curve with running an efficient, internal supply chain can be used to its advantage in departing the stage. Moreover, we foresee Finisar becoming by far the most dominant module vendor in the world for the following three reasons: 1) our continuing expectation that Finisar will purchase Lumentum; 2) our bet continues to be that Oclaro does not survive or just becomes a fab player; 3) as we have noted in the past, Sumitomo is retreating as a VCSEL supplier. So, then we would contend that Finisar should insist on a premium across the board (even to some limited extent on 10G) because of its unique ability to ensure a high level of supply, based on an even greater amount of control of major facets of the food chain.
Obviously, there would be a certain amount of risk in Finisar remaining firm on higher prices, and daring system vendors to take their chances with its competitors that have far less production capacity and in many cases, they have not established nearly as long-term of a track record in providing reliable optical components in such a high volume. Conversely, it would not be in the interest of many of these major customers for Finisar to cease to exist in the short term.
As a result of the busted model mentioned above, we tend to think it is unlikely that well-established Japanese vendors, Sumitomo and Fujitsu Optical Components, which have the resources to increase manufacturing capacity for singlemode applications, would also be apt to challenge Finisar much on price. The generation of insufficient margins would defeat the whole purpose for both of those vendors.
In addition to the effects of even better economies of scale, a major way in which Finisar can increase its margins is to cut R&D expenditures (the largest portion of its operating cost) to an absolute minimum as quickly as possible, and immediately halt any current work on gear that will not see much, if any, action over the next several years including anything related to 50G, 400G, and silicon photonics.) So, its development concentration should be confined to not much more than getting its scheduled new products out. In addition, we recommend that Finisar take a page out of Avago’s playbook and not give any volume discounts with any of its upcoming devices in which it has a true competitive advantage – specifically, with those buyers that have hurt Finisar by becoming more vertically integrated and/or have funded new componentry competitors.
Still, has not Google, Microsoft, and some OEMs potentially set up Kaiam against Finisar in the same way that Cisco Systems strongly supported Finisar in the early part of the last decade against JDSU? The big difference is that under our scenario, unlike JDSU, Finisar would be looking to exit the market within a fairly short period of time. Also, we are not convinced that the financing would be forthcoming to even come close to making Kaiam a vertically integrated entity.
[written by Mark Lutkowitz]