An OFC Top Ten

OFC is always looked upon as a time to envision what the future holds for the optical components industry. We are a long way from the euphoria of the late 1990s, capped by the orgiastic OFC2000, when many tens of thousands celebrated the peak of the Nasdaq in Baltimore. Back then, we never imagined the contraction we have since undergone: Now, the prospect of just a few thousand of us inside the cavernous halls of the Georgia World Congress Center will serve as a constant reminder of our unfulfilled dreams.

That said, it seems some guidelines on what products to watch out for at the show are in order. On one hand, some products are mature, with innovations that change the rules and favor new entrants being unlikely. On the other hand, the science that goes into making products is complex, with people with different areas of expertise coming into the industry at all times, discovering new ways to make what we already have and inventing functions we don't yet have. There always is more than one point of view on what is possible in function, performance, size, cost, and reliability.

Here are ten product areas in which challenges to the status quo provide reasons to pay attention, in reverse order of significance.

10) Dispersion compensators
This is an intriguing area because it illustrates the choices that face equipment manufacturers deciding between solving a problem optically versus electrically. There have been a variety of companies that have developed different optical solutions for chromatic and polarization mode dispersion compensation. During the Bubble, it looked as if that there would be strong demand for these products, but now, with carriers not desperate to install the highest bit-rate pipes, there has been time for companies developing electronic versions of the same functions to catch up. To the degree that a system designer can use an electronic version that lends itself more to integration and ease of assembly, he will. Will there be reasonably sized niches for each, or is it winner take all?

9) Tunable filters
There are many applications for tunable filters, including use in Tunable Lasers, wavelength division multiplexers, reconfigurable optical add/drop multiplexers, and optical channel monitors. Each application has distinctive requirements, with a subset of these being transmission shape, tuning speed, wavelength range, insertion loss, size, cost, and reliability. Identifying the winners in tunable filters first requires knowing which applications matter most, so that one can identify the critical parameters and then determine the technologies that have the most merit. One could spend all of OFC sorting this out if not careful.

8) Tunable lasers
This has been the hot area for laser makers for several years, with promises each year that they would finally displaced fixed wavelength DWDM lasers. No doubt, they have been making progress, with design-in announcements growing and customer-backed multiple sourcing agreements coming out. At the same time, however, the challenge to tunable lasers has always been to meet the cost of fixed wavelength lasers, since tuning increases costs and, without a huge need for changing wavelengths live, also has performance consequences. Will this finally be the breakout year? (For more on tunable lasers see the Light Reading report: Tunable Lasers Revisited.)

7) Platforms that promise high degrees of integration
These platforms have been promoted for years, especially during the Bubble. We heard of a variety of platforms that make possible the integration of many functions by making use of materials such as silicon and indium phosphide, as well as manufacturing techniques such as MEMS and micro-robotics. These platforms are often undone by their requirement for large investments or by discrete implementations having superior price-performance ratios. Nonetheless, the allure of integration is still out there, and some progress has been made, so companies will continue to pursue this goal. Investigating the latest integration platform is always exciting because one has to be a detective to discern what the platform can do and what the vendor is keeping back.

6) Uncooled pump lasers
At prices of cooled pumps, amplifiers have been too expensive for widespread use in metro applications, so their volumes have been low. Pump suppliers have striven to lower costs by eliminating the cooler, which requires relaxing alignment tolerances and raising pump chip powers. Unfortunately, higher powers stress device reliability. In addition, metro pumps have to be more dynamic than long-haul pumps, so their specifications can be more demanding. Pump vendors have announced progress in uncooled pumps, so learning their approaches and their latest results will be of interest.

5) Lower per-port costs in DWDM filters
In 2000, thin-film filters were selling for close to $1,000 per 100GHz port. Arrayed waveguide grating filters were supposed to unseat thin films as channel counts grew, and scaling advantages of planar lightwave circuits were supposed to reign supreme. Channel counts, however, shrank, and thin-film production was transferred to Asia. Prices have fallen close to 90 percent, to just above $100. Equipment manufacturers will buy from whoever can meet their requirements, so the question becomes which vendors continue to chase this business and drive prices down below $100 a port. At the same time, other approaches have been knocking at the door for a few years, but the willingness of incumbents to keep their accounts, whether or not they make money, has kept these pretenders at bay. Will this be the year they crack the fortress walls?

4) Spurious claims about CWDM components
Metro DWDM is supposed to be hamstrung because the price per wavelength is just too high. Studies have come out showing that CWDM, which permits the use of uncooled lasers and wider band filters, is supposed to lower the cost per wavelength as much as 40 percent. This savings is supposed to be at the equipment level. The logic is a bit hard to follow, because network equipment has to provide quality of service, regardless of the cost of the optics. These service issues, according to some manufacturers, mean that CWDM can be no more than 20 percent lower than their DWDM counterparts. At the same time, new CWDM entrants are said to offer prices 50 percent below those of incumbent DWDM prices. What is the real price difference, and what are the reasons for this? (For more on CWDM see the Light Reading report: CWDM: Low-Cost Capacity.)

3) 1310nm VCSELs
The appeal of long-wavelength Vertical Cavity Surface Emitting Lasers (VCSELs) has attracted many companies to try to make this product, since one can then make a high-speed transmitter that can be used for access and not just intra-office applications. Scientists have been trying to defeat the mirror-gain region tradeoff for years over full operating conditions. Some say the problem is insoluble, but others say there is evidence of success. I am looking forward to seeing what the status is now.

2) Cost cutting in transceivers
The volumes here are the highest, with competition having been perennially the fiercest. Multibillion-dollar corporations and startups have forayed into this area, usually learning the hard way that there is no way to circumvent the commodity nature of this business. At every new distance/bandwidth combination, prices have to fall 30 to 50 percent per year for several years, with new product versions required annually and volume shipment needed within a couple of weeks, if not immediately. Miss a qualification window and you lose a year; miss two windows and you may as well shut down. I am looking forward to seeing who is now bold enough to make a new challenge.

1) The XFP transceiver MSA
This point picks up on the general transceiver point made just above. XFP deserves its own mention because it is supposed to be the unification of Sonet (Synchronous Optical NETwork) and SDH (Synchronous Digital Hierarchy) and Ethernet transceivers, at 10 Gbit/s, in pluggable small form factor. Equipment manufacturers have long complained that Sonet/SDH components should sell at Ethernet prices, since so much of the material is the same. Subtle differences do exist, such as power budgets and extinction ratios, creating justification for some. Still, the standards committee is moving towards creating as much commonality as possible. As well, XFP vendors are rumored to be planning to demonstrate innovative low-cost versions for increasingly greater distances. Which of these will be successful and how soon can they be available in commercial volumes will be exciting to follow.

And one last "Watch-Out-For" – in more of a commercial sense. Watch out for how the industry realignment progresses. Which companies will grow in clout, at whose expense?

JDS Uniphase Corp. (Nasdaq: JDSU; Toronto: JDU) still is the market leader, but the downturn came before it could integrate its different companies. The loss in talent during the last two years leaves the leader vulnerable. Intel Corp. (Nasdaq: INTC) has the key ingredients necessary for continuing its challenge: money, customer relations, and operational talent.

Chinese manufacturers have the edge in labor cost, numbers of optical engineers, and a domestic market with great growth potential. Bookham Technology plc (Nasdaq: BKHM; London: BHM) and TriQuint Semiconductor Inc. (Nasdaq: TQNT) are now potential challengers as well, along with several other large suppliers that have traditionally been in the upper ranks of the industry.

With sales still slow, jockeying can go on for a few more years, giving time for different companies to move to the fore and for segments such as broadband access to be the key to dominance.

— Jay Liebowitz is founder and president of Liebowitz Strategies, which helps companies with strategy, business development, and positioning for increasing revenue and achieving corporate financing milestones. Liebowitz, who can be reached at [email protected], has many years of experience providing advisory services, consulting, and market research services in optical and electronic components and subsystems, and has also managed businesses at Lasertron (now Corning Inc.) and Epitaxx (now JDS Uniphase Corp.).

For up-to-date information about the coming OFC Conference, please visit Light Reading's Unauthorized OFC Preview Site

gea 12/5/2012 | 12:30:45 AM
re: An OFC Top Ten It's been a long time since OFC was just a "components show".

Aren't there any systems trends you think will be worth following at OFC?
Physical_Layer 12/5/2012 | 12:30:27 AM
re: An OFC Top Ten Jay - you mentioned that you were a bit skeptical on the cost differences between DWDM and CWDM, and I can understand why, but do you even think 20% price difference is even close to being possible? The savings has to come only from optics, right? Personally, I'm thinking that the filters will have virtually no cost differences. In TFF, the actual filter is pretty cheap. The price is all in the labour/testing. In AWG (if you'd even bother going that way), same idea...the chip is cheap...the cost is in packaging/testing.

As for the lasers, that could be a bit different because of no real-estate and cost associated with temp-control electronics. But how much would that be per channel? And how much of the total cost of a system is in optics? It can't be that much considering how much optic prices have come down. Maybe 2-3 years ago the argument for CWDM was stronger, but with such lousy prices these days, is it really even possible for CWDM to have a significant price difference?

gea 12/5/2012 | 12:30:26 AM
re: An OFC Top Ten "The savings has to come only from optics, right?"

In some situations there could be a lot more savings. Remember, unlike DWDM, CWDM is a defacto (not just dejure!) standard. All companies making 8 wavelength CWDM will be making the same 8 wavelengths. This will do a lot towards commoditize pricing, which never really happened in DWDM.

More than this, however, is the possibility of not using a transponder whatsoever. If there's no EDFA (and there won't be with CWDM), and if the wavelengths are the same, AND if the client-layer (eg, Ethernet switch) has SFP or GBIC optics, you won't need a transponder. That's a HUGE savings.
lorent 12/5/2012 | 12:30:02 AM
re: An OFC Top Ten Sorry Gea,

You're not likely to see any standardization on the system level. Sure lasers and filters may be the same but there's no way that vendor 'A' is going to support vendor 'B' on the same ring. Across trib interfaces is fine but not on the line side.

I guess the main thing bothering me is how the economics of the business are going to support the industry going forward. Nortel used to do $10B a year in OC-192 long haul. I'm not sure if the entire 192 market is doing $10B a year. Overall revenues in the telecom space continue to shrink while competition has never been more fierce. Where does it all end? Cisco + two other big competitors (Nortel, Lucent, Siemens, Nokia, etc.)? Makes the most sense except that there's always CEOs involved that don't want to give up their power. I think it's going to most closely resemble the airline industry. Screwed up big companies. Little companies that come and go and few well run niche players.
Brian VanOrsdel 12/5/2012 | 12:29:50 AM
re: An OFC Top Ten Hi Jay -

Pleased to see that in your OFC "top ten" list you've emphasized cost-reductions in transceivers as high in significance.

The performance of the XFP package and optics has been demonstrated at the NFOEC show last year - my guess is that this year at OFC you'll see a lot more emphasis placed on XFP's compatability with various SERDES mfgs devices, and the ease in which XFP can be designed into existing 10Gb optics cards.

I believe that the goal within the XFP (and also SFP) transceiver vendor base is to be able to offer a full portfolio of devices covering all spans between 600m and 80Km...40Km technology (mostly 1310nm) is here now, with longer distance devices just around the corner. XFP-pluggable technology has already been demonstrated to be capable of replacing much more costly 300pin transponders in most client-side, and many line-side applications.

Thanks Jay, for highlighting what I agree is one of the key technologies which will drive down optical system costs.
mooseknuckle 12/5/2012 | 12:22:50 AM
re: An OFC Top Ten What is virtually no cost difference? It all comes down to yeild. Additionally all customers have their own spec it seems. Depending on the shape of the filter, a custom CWDM could be more than a generic DWDM, like a 100 gig. Throw in tight CD specs and that takes another chunk out of your yeild. Then you have filter packaging yeild.

The real savings is in the lasers.

Titanic Optics 12/5/2012 | 12:20:31 AM
re: An OFC Top Ten I attended OFC at the huge Atlanta convention center. There are still way too many companies that exist.

Will any of the startups simply decide to liquidate, and spare the stakeholders the misery of a long slow death? How many firms will be at OFC 2005? My prediction: 250

When the dust finally settles, and a recovery happens sometime in the next decade, what will the industry look like? Who will the major players be? My predicition: JDS will be the dominant player over the next few years, but will fade and be bought by the firm which will be the new top dog, which could be Intel, Triquint, or someone else. Intel may drop the business or they could be #1.

In the interim, with so many companies doing non-telecom, will OFC look like Photonics West, with a lot of small, niche players that have $1-20 million in sales per year, none of which are very exciting companies? My prediction: a lot of companies are choosing this route and will hang on, getting SBIR grants and doing work outside of telecom. Hey, its a living.

Your thoughts?
Curious George 12/5/2012 | 12:19:47 AM
re: An OFC Top Ten An interesting comparison is between 10 G serial and 4 x 2.5 G CWDM. Given packaging requirements, chip sets available, etc. the serial approach is less expensive. It would seem the opportunity for 4 x 2.5 G CWDM would be for applications that suffer from transmission impairments or those that require some form of wavelength provisioning for enhanced services or security reasons. With regard to transmission challenges the LX4 application provides an example where 10 G transmission over 300 m of MM fiber is required. Without some for of dispersion compensation, 10 G serial probably would not work.

Sign In