While the hype machine on the latest and greatest technology always reigns supreme in the fiber optics market, Intel appeared to take Silicon Photonics (SP) to an unprecedented level for an extremely long period of time. Just the sheer amount of money spent by the supplier, specifically the Silicon Photonics Solutions Group’s (SPSG) promotional efforts, including the use of professional graphics artists for its presentations, was remarkable in not only helping to convince certain potential customers of the viability of these solutions, but it likely cannot be ruled out that the funds played a role in misleading high-level executives at the supplier in exaggerating progress. It remains to be seen what the long-term effects will be on Intel’s credibility of not coming through on the promised deliveries of these devices several months ago, especially as the vendor has somewhat pulled back to a more realistic stance in purchasing FPGA vendor, Altera, which provides chips that are often used in optical products.
We believe that the top executives at Intel were shocked and furious because apparently, the problems with the modules were unknown to them. Nevertheless, the buck has to stop with the CEO (in concert with other CXOs), who was directly responsible for the creation of the SPSG. The company was hyping SP for so long that there should have been at least a little doubt that crept in regarding the internal claims that everything was moving along just fine.
The news had to be a big blow to those customers as well when they were told kind of at the last minute that Intel was going to drop the ball on their deliveries. Actually, some of these companies built products around Intel’s modules including switch ports, etc., especially a hardship to those planning to have 100G capability.
We have also picked up some intelligence, which indicates that Intel’s new SP priority is to develop a singlemode fiber WDM solution by the middle of 2017. Of course, this new mandate will be even more challenging to accomplish than has been the case with multimode.
In general, Intel will continue to face the same deep-seated shortcomings of SP, some of which we have expressed in the past. For example, any claims about low power cannot be made with any credibility. Also, companies, such as Finisar and Luxtera, are working with hybrids that encompass a photonic die bonded to an ASIC (such as CMOS) that create cost issues.
There are certainly size problems with a two-die situation. Economies of scale will not necessarily be achieved if the number of chips per wafer is not so high.
One wonders whether Intel will decide to throw in the towel if it happens to disappoint customers for a second time, only further exacerbating a potential problem concerning confidence in the optical realm. The company has been doing R&D work on SP since 2000. Despite going on a promotion campaign that could easily be described as unreal, perhaps it will have had enough in continuing to take on those same rigid laws of physics.
[written by Mark Lutkowitz]
(To see “Intel Behind $200M Photonics Competition?”, please click here. To learn more about our take on the impact to Finisar of Avago Technologies’ M&A activity, please click here. To read more about Luxtera, please click here.)
it is funny to see that afte almost two decades of SiP hype, there is still not a solution nor application for direct optical link to the ASIC.
In Compass we solved this issue from a different angle, and already have in deployment a system based on direct large-scale parallel optical link between different ASIC’s in the system.
Comments are closed.