While fibeReality is unaware of the exact date of when Chuck Mattera, the CEO, of II-VI, negated the proposal to buy out Finisar’s IC group, it was around the time of Inphi’s rhetoric seemingly signaling a desire to exit. We understand that Ford Tamer, Inphi’s CEO, was pretty devastated by the refusal. Both Tamer and Michael Hurlston, Finisar’s ex-CEO definitely underestimated Mattera, at least on this deal. Perhaps if permission was received, it may have led to a relatively immediate acquisition of Inphi by Broadcom, for not only the InP capability that would have been gained from Finisar’s IC group, especially in anticipation of Broadcom moving upstream with transceivers (albeit, we continue to believe in a very targeted way), but also the other chip capabilities that the big IC firm either lacked, or would be able to gain additional market share. After getting burned on attempting to acquire Qualcomm, it would have been a good way for Broadcom to somewhat recover on the silicon side. With the Cisco Systems-Acacia Communications announcement, Inphi had apparently gained new life, as presumptively the new merchant DSP leader (with the assumption that no leapfrogging occurs by another player). While we would still lean towards Broadcom purchasing Inphi at some point in the future, the priority of the latter in the meantime is to gain as much leverage as possible to maximize its valuation. There will not be Cisco Systems competing with Broadcom, as ostensibly seemed to be the case with the buyouts of both Luxtera (as a defensive move) and Acacia. In addition, we do not buy into the theory that Inphi was interested in Finisar’s assets in order to even come close to becoming a full transceiver supplier.
Regarding this last point on Inphi, we provide the following seven reasons: 1) it had already been through the “hyperscale vortex” with COLORZ; 2) the safest way for an engineering executive to become critically important at the firm in the long term is on the ASIC side of the business (including chip validation and firmware), not with modules; 3) the tendency is to lean towards chips because of the concerns about keeping margins up, along with the concomitant perceptions of the analysts on the Street; 4) the Chinese suppliers, which can often undercut on price with transceivers, do not tend to excel as much with chip development, particularly involving coherent technology; 5) over six months ago, when there was somewhat of a turf battle between the silicon photonics (SiPh) and the coherent chip groups, we understand that the latter came out ahead; 6) the degree of impairment of Finisar’s IC group at the time of the proposed sale is unknown to us (we have heard that consolidation of two sites in California is being planned now, which may not necessarily be a bad sign in that it may involve a lot of low-end datacom devices, and coherent activities in Europe are pretty serious); and 7) assuming a buyout by Broadcom, the parent corporation would be in a much better position to decide the appropriate direction on transceivers.
With all of these factors in mind, it is still true that Inphi has put in a lot of work into the ZR module and is planning on a prototype before the end of 2019. There is also a big push to either demo this solution and/or a CFP2-DCO at OFC 2020. Naturally, there will still be questions on just how well these products work.
Another hurdle for Inphi is to make the ITU’s cFEC+ viable with the 400G chip. The game plan is to fight off Cisco/Acacia’s oFEC from the Open ROADM MSA. Although likely feeling forced to say it quietly now, Inphi’s pitch is that Acacia has been pulled out of the space, at least on the chip side, and that the former is establishing itself as the de facto standards bearer.
On SiPh, while it is hardly a priority in general at Inphi, in the same way as happened at Acacia, it is expected to remain just a component of a module, which will not result in unacceptable, negative effects on the total profitability of the full device. So, the analog team at the former will evidently use SiPh as part of ZR.
Once again, regardless of whether Broadcom actually purchases Inphi, the goal of the latter is to beef up its strategic advantage by increasing the functionalities around its chip business (including DSPs, drivers, and TIAs), such as from its last acquirement, eSilicon, which provides packaging and SerDes capabilities. The supplier had already combined the Inphi and ClariPhy analog teams. Thus, the overall “story” includes coherent/direct detect, datacom/telecom, and electronics/SiPh.
For our fast-growing, totally separate, quick update blog, which is exclusively on fibeReality’s LinkedIn page, please follow us here.
As always, fibeReality does not recommend any securities, and this writer does not invest in any companies being analyzed by us.
[written by Mark Lutkowitz]