Intel Renews Hyperscale Optic Machination
October, 2018
According to fibeReality’s newest intelligence, we have never been more persuaded that Intel is selling optical components to hyperscale data center operators below cost, while the vendor has very little confidence it can achieve profitability, as a result of sufficient volume . Ever since the supplier had started offering quantity discounts to the space with PSM4 devices (and then later with CWDM4 hardware), we have believed that it has played a major role in bombing pricing, which in due course, led to the devastation of the optics ecosystem. We have also received information that even when Intel was running its Optical Platform Division, before divesting those assets last decade, that it had a clear inclination to treat these sales no better than it would loss leaders. For example, we understand there was irreparable damage to the 40G SFP market, when Intel aggressively undercut the price. A few years later when Emcore finished buying these component assets, the acquirer found out the division was losing the same amount of money as its total revenues (around $100 million). Most recently, after a lengthy hiatus, Intel has returned to heavily promoting its supposed advantages with Silicon Photonics (SiPh) in terms of it being an important enabling technology. In the recent past, we have thought that when it came to transceivers, which may have been shipped in any kind of volume, Intel did not use the technically complex SiPh platform, similar to Juniper’s Aurrion solution, but just another traditional, hybrid-integration product approach using conventional componentry. In addition, there are obvious similarities as to when Intel started to promote its optical capabilities now and several years ago. Furthermore, Intel hiring more people to support its current efforts in the sector needs to be kept in perspective.
To further back up our assertions on how Intel treated its optical module business in the previous decade, there was little uncertainty in the industry that it was operating in the red before it was fully sold off. While these results partially had to do with poor execution, the cutthroat pricing tactics were undeniable.
Regarding the parallel situations when it came to increased emphasis on optics very close to the present as well as the past, we assert that they have been attempts at distracting analysts on the Street. In the prior years, after the bubble, Intel needed to find a means of dealing with declining margins with its traditional offerings, and so it provided a story of how optical gear (as part of its overall strategy in the communications realm) would be very instrumental to its future success. Likewise, today, Intel is battling a narrative, fairly or unfairly, that it is falling behind in the data center space, and so, all of the hype over SiPh is reintroduced, along with a promise of adjacent market opportunities, such as for 5G.
Although Intel will certainly play a visible role in supporting the next generation of wireless, IOT, etc., with its chips in general, given history, we cannot take the firm seriously when it comes to optics, including SiPh. We have written a lot on the shortcomings of SiPh, and we continue to expect its implementation to be quite limited, regardless of the manufacturer.
Yet, remarkably, Intel, perhaps in combination with Facebook (we believe the reports to be true that the supplier has received orders to ship CWDM4 gear to the social media company shortly), is in some cases, evidently able to get away with financial investment firms, that the supplier is ostensibly leveraging its server CPUs to get this transport business from the customer. Anyone on the Street not automatically realizing that it has to be the other way around has given insufficient thought to the crisis in the optical ecosystem.
Concerning the hiring of a couple of executives and more personnel, it appears to us that it is not really about additions, but replacing people who have left. We can only imagine that one of the reasons for departures would be a lack of satisfaction, perhaps even on a moral basis. They may have decided not to work for an optics operation in which the emphasis is apparently not on a thriving concern, but on helping some of its largest customers of other gear drive down prices to the point of making it impossible for the largest American optical component companies, which would ordinarily be responsible for future R&D efforts, to sustain themselves solely in this space without outside capitalization.
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]
https://www.fibereality.com/blog/intel-renews...achination
October, 2018
According to fibeReality’s newest intelligence, we have never been more persuaded that Intel is selling optical components to hyperscale data center operators below cost, while the vendor has very little confidence it can achieve profitability, as a result of sufficient volume . Ever since the supplier had started offering quantity discounts to the space with PSM4 devices (and then later with CWDM4 hardware), we have believed that it has played a major role in bombing pricing, which in due course, led to the devastation of the optics ecosystem. We have also received information that even when Intel was running its Optical Platform Division, before divesting those assets last decade, that it had a clear inclination to treat these sales no better than it would loss leaders. For example, we understand there was irreparable damage to the 40G SFP market, when Intel aggressively undercut the price. A few years later when Emcore finished buying these component assets, the acquirer found out the division was losing the same amount of money as its total revenues (around $100 million). Most recently, after a lengthy hiatus, Intel has returned to heavily promoting its supposed advantages with Silicon Photonics (SiPh) in terms of it being an important enabling technology. In the recent past, we have thought that when it came to transceivers, which may have been shipped in any kind of volume, Intel did not use the technically complex SiPh platform, similar to Juniper’s Aurrion solution, but just another traditional, hybrid-integration product approach using conventional componentry. In addition, there are obvious similarities as to when Intel started to promote its optical capabilities now and several years ago. Furthermore, Intel hiring more people to support its current efforts in the sector needs to be kept in perspective.
To further back up our assertions on how Intel treated its optical module business in the previous decade, there was little uncertainty in the industry that it was operating in the red before it was fully sold off. While these results partially had to do with poor execution, the cutthroat pricing tactics were undeniable.
Regarding the parallel situations when it came to increased emphasis on optics very close to the present as well as the past, we assert that they have been attempts at distracting analysts on the Street. In the prior years, after the bubble, Intel needed to find a means of dealing with declining margins with its traditional offerings, and so it provided a story of how optical gear (as part of its overall strategy in the communications realm) would be very instrumental to its future success. Likewise, today, Intel is battling a narrative, fairly or unfairly, that it is falling behind in the data center space, and so, all of the hype over SiPh is reintroduced, along with a promise of adjacent market opportunities, such as for 5G.
Although Intel will certainly play a visible role in supporting the next generation of wireless, IOT, etc., with its chips in general, given history, we cannot take the firm seriously when it comes to optics, including SiPh. We have written a lot on the shortcomings of SiPh, and we continue to expect its implementation to be quite limited, regardless of the manufacturer.
Yet, remarkably, Intel, perhaps in combination with Facebook (we believe the reports to be true that the supplier has received orders to ship CWDM4 gear to the social media company shortly), is in some cases, evidently able to get away with financial investment firms, that the supplier is ostensibly leveraging its server CPUs to get this transport business from the customer. Anyone on the Street not automatically realizing that it has to be the other way around has given insufficient thought to the crisis in the optical ecosystem.
Concerning the hiring of a couple of executives and more personnel, it appears to us that it is not really about additions, but replacing people who have left. We can only imagine that one of the reasons for departures would be a lack of satisfaction, perhaps even on a moral basis. They may have decided not to work for an optics operation in which the emphasis is apparently not on a thriving concern, but on helping some of its largest customers of other gear drive down prices to the point of making it impossible for the largest American optical component companies, which would ordinarily be responsible for future R&D efforts, to sustain themselves solely in this space without outside capitalization.
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]
https://www.fibereality.com/blog/intel-renews...achination
1
