x
Optical components

Optical Vendors Face PIC Challenges

Photonic integrated circuits (PICs) may stand out as a solution to the cost associated with building out high-capacity optical networks, according to a new report by Heavy Reading. The report, Photonic Integration & the Future of Optical Networking, also points out that today's components and systems manufacturers face severe challenges in developing PIC technology.

The need for higher capacity optical products is clear, as carriers and network operators are being forced to meet rapidly expanding capacity requirements. "The applications that ride on optical networks continue to evolve and grow, but the ability to spend additional money for that capacity has dried up," says Sterling Perrin, Heavy Reading analyst and author of the report.

To ramp up their networks, service providers are asking optical vendors for 40 Gbit/s transponders at 2.5 times the current cost of 10 Gbit/s transponders. However, optical vendors have been unable to meet these demands, with 40 Gbit/s transponders typically at five to six times the 10 Gbit/s price.

Optical vendors clearly have to do better, but it's an open question how they will lower prices while increasing capacity. "The optical market continues to grow, but components costs have not stabilized," says Perrin.

That is causing optical components and systems manufacturers to look into alternative models, such as outsourcing, to lower the costs of optical products. But Perrin says that over the long term, driving down the cost of optical equipment "is not going to work simply by outsourcing. You need real technological advances."

"For the industry to achieve the future it envisions, a new revolution in optical communications will be needed. Photonic integration – with the PIC at the center of it – may be the biggest contributor to this revolution over the next ten years," he writes in the report.

Infinera Corp. (Nasdaq: INFN) CEO Jagdeep Singh argues that the current "discrete approach" of pushing more bits per wavelength is not economically viable and doesn't meet carrier needs.

"Fundamentally, if you look at the industry, the work that's going on is going on in terms of getting more bits per wavelength. But carriers want greater capacity, cost per bit, flexibility, and reliability, and getting more bits per wavelength doesn't solve that," Singh says.

"The metric that correlates best to carrier needs is bits per chip. The more capacity per chip, the more separate components are being eliminated," Singh says.

While researchers have seen the promise of integrated photonics for 30 years, most optical components and systems vendors are just now beginning to look into the technology. However, due to the telecom downturn of the early 2000s, there's a gap of innovation in developing PIC technology and no clear solution for correcting that.

"There has been a lack of investment in photonic integration over the last five years, leaving major gaps in innovation. The telecom market contraction that hit North American and Europe in 2001 continues to affect the PIC sector, as components vendors were at the bottom rung of the telecom supply chain and were hit harder than both their customers (equipment suppliers) and their customers' customers (telecom operators)," Perrin writes.

The ensuing carnage forced large component suppliers such as JDSU (Nasdaq: JDSU; Toronto: JDU) and Bookham Inc. (Nasdaq: BKHM; London: BHM) to contract and left many optical startups, including those focused on developing PIC technology, without funding.

The one exception to that trend is Infinera, a PIC-based success story in the optical industry. "The lead that Infinera has is pretty amazing," Perrin says. "A significant lead in the optical business is typically about six months, and this company is realistically about four years ahead of systems suppliers and component suppliers."

It will be difficult for another startup to duplicate that success, due to upfront costs required to enter the PIC market. Before going public, Infinera raised more than $300 million in venture funding, which is a sum that VCs seem unlikely to commit to telecom equipment in today's market.

It will also be difficult for larger systems vendors to move to PIC-based DWDM systems to compete with Infinera. That's because many have already divested their components. In his report, Perrin writes: "Infinera built a vertically integrated company at a time when big telecom operators – such as Alcatel-Lucent (NYSE: ALU) and Nortel Networks Ltd. – were shedding their components divisions to become specialized."

With little vertical integration left and few new startups to develop PIC technology, systems vendors will be leaning more heavily on components suppliers for photonic integration, a trend Perrin writes is already underway. "Heavy Reading has learned that such conversations are already taking place between systems suppliers and their components suppliers," he writes.

While it's unclear how optical components and systems vendors will proceed in the short term, the good news is that they recognize the need for photonic integration and, with Infinera as a proof point, are looking to develop their own PIC-based products.

"Infinera paved the way for photonic integration to come in earnest. The acceleration you're seeing is coming now that companies have seen Infinera succeed with the PIC model," Perrin says.

— Ryan Lawler, Reporter, Light Reading

Page 1 / 3   >   >>
NoCopper 12/5/2012 | 3:45:10 PM
re: Optical Vendors Face PIC Challenges For metro reach transponders the ratio is about 1:4 up to 1:5 for plain transponder cost. But you have to consider that typically 4 x 2.5G transponders require more slots in a shelf, more power and obviously 4 times the optical spectrum then 1 x 10G. So the overall ratio for 10G is better then 1:4 but certainly not as good as 1:2,5. Besides that you can run 10GbE on one 10G lambda and you can mux 9xGE in a 10G lambda, but only 8xGE in 4x2,5G lambdas.

In long haul I guess most lambdas are at 10G anyhow because of the reasons mentioned above. No carrier wants to waste the investment in the optical line (filters, roadms and amplifiers) with 2.5G lambdas. Any vendor still offering 2.5G long haul tranponders in a recent optical long haul system?
Stevery 12/5/2012 | 3:45:10 PM
re: Optical Vendors Face PIC Challenges service providers are asking optical vendors for 40 Gbit/s transponders at 2.5 times the current cost of 10 Gbit/s transponders. However, optical vendors have been unable to meet these demands, with 40 Gbit/s transponders typically at five to six times the 10 Gbit/s price.

What's the present ratio of 10 Gb/s transponders to 2.5 Gb/s?
howsweetitis 12/5/2012 | 3:45:09 PM
re: Optical Vendors Face PIC Challenges JDSU,ALU NT etc just don't have the brain trust to do what has taken 7 years to develop at INFN. The 4 year lead will not only increase but others simply can't put together a technical team that was and still is the perfect storm at INFN. INFN has just begun to touch the tip of the iceberg. There level of traction in this space is hugh. Tier 1 carriers will be jumping on this or will simply not be able to handle the speeds and capacities needed by there backbones. It not an "if" it's a "when".
rationator 12/5/2012 | 3:45:08 PM
re: Optical Vendors Face PIC Challenges PIC is great technology. INFN seems to have a great product where customers seem to be rewarding them with revenues. However, every technology has tradeoffs. Just because INFN has found a great market in DWDM for PIC does not necessarily mean that PIC is the solution for every problem.

Fact 1: INFN is a systems company first.
- Systems companies are addressing a larger market and therefore have access to larger rounds of funding from VCs
- Systems companies are profitable as a space. Please check which optical component manufacturers are profitable

Fact 2: INFN uses PIC technology for DWDM lambdas
- Look at INFN boxes on their product pages, they are using modules for client side
- Interoperability at the DWDM wavelength physical interface is not a firm requirement for systems
- Interoperability at the client side interface is a must have
- Interoperability of a multi-lambda PIC technology is much more difficult to specify and build
- INFN uses some form of electronic dispersion compensation to clean up the performance of PIC to PIC and insure interoperability. This would be difficult to extend in a multi-vendor interop environment

Fact 3: INFN DWDM technology is currently limited to 200GHz channel spacing.
- Large existing operators primarily have networks based upon either 50GHz or 100GHz DWDM channel spacing
- Most competing transport solutions offer 100GHz or 50GHz based systems (Systems makers in the boom times were researching <10GHz spaced channels)
- Moving toward broader channel spacing reduces the number of wavelengths per fiber and therefore reduces the revenue per fiber
- Other system vendors are focused on offering ROADM technology to provide flexibility
- ROADM technology is more valuable in a higher channel count network which is why non-INFN DWDM systems makers are offering this technology

"If the only tool you have is a hammer, you tend to see every problem as a nail.
Abraham Maslow"
doodler 12/5/2012 | 3:45:08 PM
re: Optical Vendors Face PIC Challenges Why was Santur overlooked on the list of PIC suppliers? Haven't they proved their capability in that every one of the tens of thousands of tunable laser array chips they have shipped is a PIC?
cw.774 12/5/2012 | 3:45:08 PM
re: Optical Vendors Face PIC Challenges Yield is low. Partnerships will be needed for some time (and all the disorganization associated with it).

Nobody wants to pay PhDs to run a production line that still tosses out PICS left and right.

Economy of scale doesn't solve this problem of production yield for the next 10+ years (not 5). By then, other analog devices will challenge the "low cost" OEO PIC on InP substrate.
Mostly Harmless 12/5/2012 | 3:45:07 PM
re: Optical Vendors Face PIC Challenges hello rationator,

i think you have not really understood realities of dwdm market. i quote you comment and put my reply:

>>- Look at INFN boxes on their product pages, they are using modules for client side

yes, is true. but client side optic will be single lambda, gray optic. not a pic application.


>>- Interoperability at the DWDM wavelength physical interface is not a firm requirement for systems

line side interop is not a requirement at all, period. line side optic will have fec, and all fec are proprietary. nobody use itu fec. nobody interoperate on line side optic.

>>- Interoperability at the client side interface is a must have

agree, and what is your point? why is this different for pic and non-pic?

- Interoperability of a multi-lambda PIC technology is much more difficult to specify and build

not true for two reason:

1. nobody else make pic except infinera. so anyone who use pic in system must buy pic from infinera (but they dont sell i know). so interoperability would be assured because both line interface are truely made by infinera not matter whos badge on box.

2. but this oem will not happen i think. so if nobody interoperate on line side no matter tthey use pic or not, why you say is <more difficult=""> to use pic for interoperability?

>>- INFN uses some form of electronic dispersion compensation to clean up the performance of PIC to PIC and insure interoperability. This would be difficult to extend in a multi-vendor interop environment

ok, several people use edc. nobody interoperates on line side. so why would infinera edc make it <more difficult=""> than anubody else?

sorry to say but i think you look for excuse to critic pic approach.
</more></more>
Stevery 12/5/2012 | 3:45:06 PM
re: Optical Vendors Face PIC Challenges So the overall ratio for 10G is better then 1:4 but certainly not as good as 1:2,5.

So the folks asking for 40G to be 2.5x the price of 10G are asking for a better ratio than the current 10G:2.5G price ratio?

Or put another way: Service providers are asking optical vendors for a better price ratio for cutting edge technology than for commoditized technology?

In what universe is that realistic?

In truth, it sounds like another throwback to the unrealistic thinking of the bubble.
Sterling Perrin 12/5/2012 | 3:45:06 PM
re: Optical Vendors Face PIC Challenges Stevery,
The 10G market took off at 2.5x the price of 2.5G - meaning 4x the bandwidth at 2.5x the price. The operators we've interviewed are asking for similar pricing to justify the move from 10G to 40G. So, they are not asking for a better ratio, they are asking for the same.

Vendors are devising clever arguments for why 5x or 6x the price is justified for 40G, but I don't believe operators will listen.
boozon 12/5/2012 | 3:45:06 PM
re: Optical Vendors Face PIC Challenges "The metric that correlates best to carrier needs is bits per chip". Good marketing mr. Singh!
The reality is that customers only care about bits per dollar...
So, apart from quoting Mr. Singh's gospel, why doesn't the article provide the relevant numbers to support the PICs superiority? Yes, those that one can believe in without necessarily having the faith in PICs...
bummer!!! it's all about dollars, remember?
Just buy the report...
Page 1 / 3   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE