Article Info

Cisco’s Optical Illusion

It’s common knowledge in the close circles of the optical module business that Cisco (CSCO) has built an extremely profitable business on the backs of optical module companies. What is not appreciated is its magnitude and the corrosive impact it has on the profitability of the optical module business.

The Landscape

The over investment of the optical boom created a large glut in optical module manufacturing capacity. This glut remains and is fueled by too many second tier suppliers that remain sufficiently capitalized to continue operations despite the fact their core business is unprofitable. Industry consolidation has yet to take place and until it does module suppliers will face a hazardous pricing environment. The problem is compounded by the fact that though there are too many suppliers, only one customer accounts for the majority of unit demand. That customer is Cisco.

The Catalyst

Cisco dominates the market for Ethernet Switching equipment with a 60% market share. As a result, they dominate the supply chain for components used in those applications. We estimate that Ethernet optical modules account for 90% of module unit volume, and that Cisco purchased 6.1 Million of the 8.7 Million optical modules (70%) used in Gigabit Ethernet applications in 2005. As a result, Cisco is in a strong position to control supplier pricing, guide R&D investment, dictate contract terms, and control the destiny of small supplier firms.

A monopsony is a market situation in which the product or service of several sellers is sought by only one buyer. Cisco has established a monopsony in the market for certain optical modules that represent the vast majority of unit volume in the module market.

The Result

Pricing pressure on optical module supplier margins has been extraordinary. No other component industry has been squeezed as hard by Cisco’s much respected Global Supply Chain Management (GSM) organization.

Finisar and Avago sell short reach modules in large volumes to Cisco for $25, netting out around $5 in gross profit on each module sold. Cisco then resells this module (which has been pre-packaged and labeled for Cisco by Finisar/Avago) for $150 to $300, depending on the customer. Cisco has added no value to the product yet extracts a gross profit 25-50x greater.

Cisco has shrewdly built a high margin business reselling these optical modules to customers. The resale of standard form factor GBIC, SFP, and XENPAK modules was responsible for $0.26 (25%) of Cisco’s FY06 earnings, and accounts for the majority of Cisco’s operating income growth since FY04.

Virtually all of the value created by optical module manufacturers is monetized by Cisco. A reversal of this trend would be incredibly positive for module makers and quite debilitating to Cisco’s earnings growth.

The Future

Suppliers of lower-end commodity modules (the market Cisco controls) will find it difficult to realize higher margins on incremental revenue until the Cisco monopsony is broken. We believe this monopsony can only be broken through further consolidation and vertical integration of high volume module suppliers.

While the resale of optical modules does create real profit and cash flow today, it is not representative of Cisco’s generally perceived core competencies and its ability to generate long term returns on invested R&D. Significant unseen risks exist that could negatively impact Cisco’s ability to sustain 90% profit margins in this business unit. The next Michael Dell may sell optical modules.

The Report

Investors and the market are currently unaware of the earnings risks posed by a potential unraveling of the Cisco optical monopsony.

We examine these risks and their financial impact in this report. Detailed pricing, volume, and margin estimates for Cisco’s optical modules are included. This data is coupled with Cisco’s historic operating results to illustrate how this resale business impacts Cisco financials. The report also examines future technical and market risks to the profitability of this business and the resulting financial impact.

Contact us for more information.

“Perhaps the sentiments contained in the following pages are not yet sufficiently fashionable to procure them general favor; a long habit of not thinking a thing wrong, gives it superficial appearance of being right, and raises at first a formidable outcry in defense of custom. But the tumult soon subsides. Time makes more converts than reason.”

—Thomas Paine
Common Sense, 1776

UPDATE – Lightreading is running a poll to gather feedback. Vote here.


Comments are disallowed for this post.

  1. Great article in a recent Harper’s Magazine on monopsony, in their case referring to WalMart’s power over its suppliers. The article makes a compelling claim that WalMart’s status as a monopsonist is the result of an evisceration of the anti-trust laws that govern the U.S. marketplace (remember, this is Harper’s).

    The author finds this trend worrisome not because monopolists or oligopolists in the U.S. will charge consumers higher prices, but that they will dictate pricing from their suppliers, and that “it deprives the firms that actually manufacture products from obtaining adequate return on their investment. In other words, the ultimate danger of monopsony is that, over time, it tends to destroy the machines and skills on which we all rely.”

    That seems to apply more to Bookham and Avanex than to Fininsar and Agilient, but Andrew points out a very specific example here of monopsony at work: the customer (Cisco’s enterprise and service provider customers) do not suffer so much from high prices from Cisco’s near monopoly on Ethernet switching as do Cisco’s suppliers. This has opened the door for Chinese competitors to come in, and therefore has a negative consequence on domestic suppliers and our economy in some slight way.

    This pratice, writ large across the whole U.S. economy, definitely warrants some critical thinking. The author lays the blame at the feet of Reagan-era legislators, who pushed free markets so hard that the market in many cases now is anything but free (Wal-Mart being the best example for suppliers of consumer goods). When I read this article a few weeks ago, I couldn’t help but think of telecom, and this report seems to confirm my suspicions with some real detail.

    Thanks, Andrew

    Scott Clavenna

    Posted by Scott Clavenna | September 6, 2006, 1:09 PM
  2. The problem with “a Michael Dell” selling optical modules is that they most likely will not function when inserted into Cisco manufactured gear. On some platforms it is possible to circumvent the issue with software commands while on others GBICs and SFPs not having a Cisco code will be inoperational.

    Posted by Kaj | September 6, 2006, 10:09 AM
  3. You’re right. Unless you hack the copy protection. Which any undergraduate with a BSEE can do. And numerous Chinese companies already are.

    Posted by Andrew Schmitt | September 6, 2006, 10:19 AM
  4. The author seems completely oblivious to actual working of
    networks and coming up with an article, which is equivalent
    of political attention grabbing by promising cheap drugs
    from imported sources. Let’s break the monopoly, great!

    In my network, when I use so called equivalent components from from the same vendors which Cisco supplies and the
    components are not Cisco certified, I have always lended
    into trouble, without exception.

    Just to give an exact picture, we bought SFPs from the same
    vendor which Cisco supplies but the catch is that all SFPs
    are not the same. They have too many variations and they are
    not completely passive components but do communicate with
    networking gear. There are at least 10 different critical
    parameters which must be right of 99.99% (just 2 9s) uptime.
    In our components we found bunch of problems and the manufacture finally did acknowledge that compatibility tests
    with network modules are REQUIRED!

    Cisco did let our university these components allowing us
    to bypass checks but the end result was that we were better
    off getting Cisco approved optics.

    Try yourself before really falling for soooooooo much
    simplicity of a naive author who never seem to have worked
    on a network.


    Posted by TooSimplistic | September 10, 2006, 9:05 PM
  5. Mr TooSimplistic:

    First and foremost, I was responsible for maintaining and running the network for the UCSB College of Letters and Science in the early 90’s.

    Your explanation makes no sense. If Cisco tested each and every module they receive from a supplier, it would- but they don’t. So if in fact you bought modules from the same supplier to Cisco you would have an equal chance of failure as a Cisco branded module. Your ’10 different parameters’ are tested more extensively by a quality module supplier than Cisco itself.

    Finally, if you are going to point accusatory fingers, stop hiding behind an alias, and perhaps you could start by at least posting your message from the University you work at and not a generic IP address. It would lend more authenticity to your claims.

    Posted by Andrew Schmitt | September 11, 2006, 6:22 AM
  6. Andrew,

    Too Simplistic slammed you and you in turn slammed him/her. However, he/she is somewhat correct and I have seen similar experiences. There is a lot of work in qualifying SFP modules. They are not all alike even though they should be. There can be serious variations in quality, heat output, power consumption and wavelength sometimes even in the same module by the same manufacturer (they change parts and process without telling the vendor and it causes problems so we find out about the “small” changes). My understanding from the HW experts that I work with is that the SFP is a very loose standard.

    Companies, such as cisco and others, have to account for qualification testing and then standing behind these products (they can not test every one shipped) and they are mass produced and defect rate is actually fairly high. They also have to account for the cost of supporting/troubleshooting and replacing defective modules for their customers.

    That said, there is a lot of markup in the module but also very large discounting on modules and whole systems (of which the SFP module is expected to help make up for the loss on other components).

    I would like to know where you got your numbers because frankly, I do not believe they are very realistic (i.e. they are too simple). Even if cisco prices the $25 SFP at $250, which they do. They then often discount the system it is put in at 50-60% or more. To claim that they make 25% on SFPs based on how many SFPs they sell at list price is not accurate. If you said they make 5-10% I might believe it because they sell a lot of SFPs.

    Customers often come complaining about the price and when they try to buy the components themselves to “save money” as toosimplistic stated, they run into many problems with the components they get and they don’t get the volume discount that cisco does either. In the end the vendor does provide a lot of value that the optical company can not on their own. I would agree that cisco probably exerts their leverage better than anyone else and to some extent it may be unhealthy for their suppliers in the long run.

    Posted by Todd | June 8, 2007, 11:30 PM
  7. Trackbacks / Pingbacks

  8. Finisar & Luxtera - Gilder Telecosm 2006 at Nyquist Capital | October 6, 2006, 9:02 AM
  9. Cisco’s Scorched Earth Strategy at Nyquist Capital | November 14, 2006, 2:06 PM
  10. Why JDSU Bought Picolight at Nyquist Capital | February 28, 2007, 10:09 PM
  11. 10GbE and SFP+ - This Time It’s Different | Nyquist Capital | December 21, 2007, 8:07 PM