Suggested Texas VCSEL Strategy for II-VI

Avatar photoPublished on
Blog 1542312357 244

fibeReality has recently expressed its point of view that the II-VI-Finisar merger only really seems to makes sense in the context of Apple’s understandable concern over the 3D Sensing (3DS) manufacturing facility in Sherman, TX. Perhaps in hindsight, Apple’s “shot across the bow” was really meant to push the two firms to get a deal done. We have also written about the design ties between Finisar and Apple, which have been imperative in keeping the former in the running for VCSEL chip business from the latter. However, even assuming that II-VI fully makes any necessary adjustments on the process order at the plant, there is still a high level of risk of being a player in this space, especially increasing price competition, as we have discussed in the past. Not to mention that both II-VI and Finisar still have yet to have a proven track record that they can successfully produce even close to the quantities of six-inch devices demonstrated by Lumentum/WIN Semiconductors, including dealing with the inordinate amount of stress on VCSEL wafers at that size. There is no doubt that in combining the large resources of these two companies, the odds of overcoming any technical obstacles should increase immensely. We strongly believe that it has also become clear that the rewards in potentially becoming the dominant supplier of such gear to Apple, particularly in the very long term, would have been difficult for II-VI to pass up. It is also important for the supplier to have a game plan for the Sherman fab, which will mitigate both the inevitable fluctuations in demand for these components, specifically for Apple’s iPhone products, as well as keep operational costs as low as possible.

One aspect that we think is very important for II-VI to keep in mind is that it should not count on a great deal of sales to Android smartphone producers. According to our latest outlook, there does not seem to be a big rush by them toward facial recognition in general, and even in cases in which they may decide to move in this direction, they either may not use VCSELs, or tend to favor ams as their supplier. Naturally, there is also Lumentum, which may be selected instead by Apple’s competitors.

Therefore, in the foreseeable future, even with a presumption that II-VI takes over the bulk of the Apple account, a factory simply cannot be maintained. In other words, it would be unworkable to make, say, 50,000 to 200,000 six-inch die, only once a year. It is a much different situation with WIN Semi in that it is already producing six-inch GaAs chips for other purposes on a regular basis, and has some idle capacity available for occasional smartphone needs.

So, for an indefinite period of time, it is imperative to have other lines in the Sherman plant, and bringing over the smaller-sized VCSEL wafer manufacturing there would certainly be extremely advisable from an efficiency point of view. For every iPhone type of VCSEL use, there are lot of other apps, where the margins are higher, but the volume of material is less.

Perhaps with the newly combined corporation, it would also make sense for II-VI to more broadly take on Broadcom in competing for VCSEL bare die business in the traditional data center transceiver market, which is definitely much better served by the tinier wafers. Penetrating that space was an action we had suggested to Finisar in the past.

Needless to say, another big intricacy is that the VCSEL world is changing. Traditionally, for transceiver applications, they were very tricky to make, and it used to be considered a must to have Vertical Integration (VI), at least for transceiver applications.  Now, as much as Apple wants the Sherman plant to be a winner, it will not stop the customer from playing on both sides of the fence in leaking the technology to as many GaAs IC foundries as possible.

In order to combat the heavy increase in the democratization of the key VCSEL processes, II-VI will have to fully prove that it can gain a meaningful cost advantage over other competitors with six-inch VCSELs through that old-fashioned means of VI. The vendor should also be in a position to differentiate itself from these RFIC firms by promoting its in-house expertise in comprehensive testing and integration.

Moreover, although Apple does not usually buy hardware assets, II-VI’s 3DS business unit could potentially be sufficiently unique and important (particularly the high-power variety) that the customer might acquire the unit down the road. Gaining even greater control over this technology may be hard for Apple to resist.

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]