LightTrends

 

Is Silicon Photonics Real?

View this newsletter in Chinese

Sure, silicon photonics exists as technology, but it has yet to be tested commercially by the market. Speculations about huge market potential of silicon photonics (Si-photonics ) have been fueled by Cisco’s acquisition of Lightwire a year ago (http://newsroom.cisco.com/press-release-content?type=webcontent&articleId=744182); Cisco paid $260 million for Lightwire—a Si-photonics start-up without a product but with a good IP portfolio. Intel fanned the flames more recently, announcing its development of Si-photonics–based optical interconnects to support next-generation server architectures at the Open Compute summit hosted by Facebook (http://www.opencompute.org/summit-2013/).

Cisco, Facebook, and Intel are perfect names to have on the list of supporters for any emerging technology. The term “silicon photonics” is also a perfect marketing slogan. It means much more than photonic integrated circuits manufactured on silicon substrates. It also implies that photonics may one day follow the path of silicon electronics towards integration of millions of elements and functions on a single chip. Needless to say, the global financial community would welcome any company with a potential of repeating Intel’s success in scaling a new semiconductor technology.

Intel has been conducting research on optical integrated circuits for years, if not decades. The company created a lot of publicity by starting a planar optical circuit business in late 1990s but shut it down quietly in 2004. Intel excited the optical interconnect world a few years ago with releases suggesting its Light Peak technology would bring high-speed optical links at consumer prices.  As we now know, Light Peak evolved into Thunderbolt with Apple and was implemented in copper.  Intel’s latest adventure into Si-photonics is a continuation of this story. It is surprising that these false starts are not mentioned more publicly.

A reality check on the current state of Si-photonics technology is long overdue. An industry panel titled “Silicon Photonics and Photonic Integrated Circuits” at the Photonics West conference in San Francisco last week offered a great opportunity to do just that (http://spie.org/app/program/index.cfm?fuseaction=secategorydetail&catid=300&event_id=896195&export_id=x13090&ID=x27509&redir=x27509.xml).

To the credit of conference organizers, this panel featured two speakers from companies (Infinera and OneChip) manufacturing Indium Phosphide (InP) integrated chips in addition to Si-photonics evangelists (Intel and Luxtera). The presentation was intended to offer an honest and balanced discussion of the state of these technologies, which would have been appropriate for the crowd of optical engineers and researchers attending the event but the end result was not the case.  The speakers got carried away with their marketing agendas, stretching the facts and often misleading the audience.

For example, Luxtera’s representative mentioned that the company shipped 2 million Si-photonics transceivers over the last few years and, responding to a LightCounting question at the event, stated that close to 1 million of them were shipped in 2012 for applications in InfiniBand optical interconnects used in high-performance computers. According to data collected by LightCounting from leading suppliers and consumers of InfiniBand interconnects (including Luxtera’s customers), less than 100,000 active optical cables were shipped and installed in 2012. The majority of these products were not based on Si-photonics.  Although a close read of a Luxtera press release indicates that the company counts channels at each end with an InfiniBand QDR active optical cable as incorporating “eight channels” and makes the story brighter, the statistics quoted still misled the audience.

Data presented by Infinera looked more reasonable, but its analysis of the market was heavily biased. Infinera’s representative stated that the company shipped close to 1,000 100G DWDM ports in Q3 2012—more than any of their competitors, illustrating the success of InP–integrated products. This speaker also mentioned that the 40G market is dead, suggesting that customers are only buying 100G products today. This is not true. Data collected by LightCounting shows that shipments of 40G DWDM ports almost doubled in 2012, reaching close to 60,000 ports, and demand for 40G products remains strong in early 2013. Shipments of 100G DWDM ports were about five times smaller in 2012 and are not expected to reach parity with 40G DWDM ports until 2015. Infinera is still facing a challenge of transforming an early lead in the 100G market into a sustainable leadership.

Are these speakers truly overexcited about the success of their companies or working on shaping the public opinion and potentially the market to favor their business strategies? It is not a secret that Infinera, being unable to deliver products on time, missed the 40G market. The company is clearly looking at the emerging 100G market for a rematch, which may be the last chance for the company to impact the industry. Infinera has been shipping networking equipment based on InP–integrated circuits for almost a decade, and this proves the technology can be viable.  Infinera was founded on photonic integration but became a systems company to build a business from it, and the company remains a niche player that has not reported a profit since 2008, when the company’s annual revenue peaked at $519 million, a small fraction of more than $15 billion global optical networking market.

Looking at the industry development over the past 10 years and filtering out the marketing hype, it is clear that photonics integration is widely used by optical component and module suppliers for incremental improvements in product performance or manufacturing processes. However, the idea that photonic-integrated circuits will change the optical industry just like Si-integrated circuits changed the electronics industry remains a distant and probably illusive target. Companies betting their business success exclusively on photonics integration have had very limited success so far. Unfortunately, this simple truth was not addressed by any of the speakers on the Photonics West panel.

Photonics West is a major industry event bringing together leading companies and researches from all over the world. Optical communications is not the major focus of this conference, as the majority of attendees are focused on biomedical or industrial applications of lasers and photonics. More than 4,000 scientific papers were presented at the Photonics West conference in 2013, illustrating work of many scientists searching for answers by collecting data and analyzing results, joined by the common goal of advancing our understanding of nature. It is a shame that the Si-photonics industry session at this event misled the attendees, many of whom are young engineers searching for areas of technology to apply their energy and talents. How many of talents like theirs were wasted in the telecom bubble? Is it time for the optical communications industry to be a bit more transparent? Analyzing failures of the past will certainly help the next generation of technology and business leaders. Why are we keeping them in the dark? LightCounting aims to contribute to this process by publishing unbiased market research, including reports on markets targeted by Si-photonics technology (http://www.lightcounting.com/activeoptical_2012.cfm).

On a positive note, Si-photonics products are very likely to be an integral part of the optical communications market. This technology is still developing, and many businesses are looking for the right products and applications to enter the market. It is too early to declare success. The future of Si-Photonics also depends on limitations of the more established technologies. For example, if VCSELs are not successful at 25Gbps data rates, Si-photonics may displace this proven technology in short-reach, high-volume optical interconnects.  But do not count on leading VCSEL manufacturers to give up this market without a fight.

About LightCounting
LightCounting, LLC, a leading optical communications market research company, offering semiannual market update, forecast, and state of the industry reports based on analysis of publicly available information and confidential data provided by more 20 leading module and component vendors. LightCounting is the optical communications market’s source for accurate, detailed, and relevant information necessary for doing business in today’s highly competitive market environment. Privately held, LightCounting is headquartered in Eugene, Oregon. For more information, go to http://www.LightCounting.com, or follow us on Twitter at http://www.twitter.com/lightcounting.