Article Info

PMC-Sierra Q207 Conference Call Notes

image Here’s a quick snapshot of the PMC-Sierra (PMCS) FQ207 conference call.

  • Revenue up slightly based on the strength of storage and FTTH chips
  • Headcount down from 1182 to 1057 with a corresponding reduction in opex.
  • “Market Conditions Improving”, with Video being cited as the big driver I still would like to see someone somewhere step up to the plate and quantify this effect. Without data this is another “internet traffic doubling every 9 months” type of proclamation

  • Expected continued market improvement through the end of the year.
  • Lots of storage design wins in SAS (Serial Attached SCSI) though it is still less than 10% of their storage business. Storage featured prominently in the “Design Win Parade” that characterizes investor calls for component companies (They never tell you what they lost).
  • China is biggest area of strength, particularly in Metro Optical and Wireless Infrastructure. “Chinese OEMs are driving telecom growth”.
  • FTTH revenue up in China and Korea. NTT committed to 3.2M ONU installs in Japan in the year ending March 08.
  • More R&D directed at GPON. Bigger focus going forward. They think world will be 60%/40% GEPON/GPON while rest of world thinks 40/60. I agree with PMC.
  • Some odd Q&A about Cable companies using PON. This is something I uncovered months back (see “PON Moves Towards the Enterprise“) but it is the first mention I’ve heard in a public forum from a public company. CEO Bailey backtracked a bit but it sounds like the cat is getting out of the bag. I’ve heard all kinds of stories about cable deploying fiber – I think the only conclusion one can reach at this point is they are going to do something beyond coax.
  • Outlook is $112-$115M next quarter, up from $108.9 this Q. No mention what the deferred component will be next Q. I am assuming $112-$115 includes deferred revenue, which they don’t report on the formal P&L. This Pro Forma accounting is out of hand and the P&L is tough to digest. The key metric: Net Cash is up about $25M in the last 6 months.

Market liked the call. What did you think?

Author owns no position in PMCS


Comments are disallowed for this post.

  1. They seemed to duck questions regarding numbers in Korea. First analyst was asked to re-queue the question, and when another analyst asked, they somewhat unwillingly answered. Previous guiding has been 800.000 lines in Seoul in 2007. He said that every start of a ramp can be a bit “lumpy” and that H1 was doing well and that they were expecting a straight line ramp, but as it turns out it might not happen that way… Does anyone know what might have changed in Korea re. PMC-Sierra and Dasan?

    Posted by olsen | July 20, 2007, 3:08 PM
  2. Two things:

    1) I’ve heard the same things you have about MSOs looking at PON for business services (particularly from Teknovus). Vik Saxena, CTO at Comcast, expressed an interest in PON for such applications during the OSA/Lightwave Executive Forum at OFC in March. However, I haven’t been able to get him to comment further; the company isn’t ready to discuss future optical strategies, he says. Overall, it looks like MSOs are going toward Carrier Ethernet for optically delivered business services. You can do that via PON, but some MSOs have already deployed gear from companies such as Atrica.

    2) I agree with you about GEPON getting the lion’s share of the world PON market. Looking at the overall FTTH market, however, I wouldn’t ignore Active Ethernet architectures such as the ones Cisco and World Wide Packets are pushing. These will be particularly popular in Europe if the EU succeeds in pushing open access down everyone’s throat. Supporting open access via PON is a bear. That’s why many muni networks in Europe (and UTOPIA here in the States) aren’t using PON.

    Posted by Stephen Hardy | July 20, 2007, 3:39 PM
  3. Stephen:

    Agreed on both counts. Pt to Pt fiber looks like the best bet for Enterprise connectivity. Hard to imagine the cable guys deploying a shared medium again. NTT went with PON, the local drops can be unbundled but at very high cost.


    Posted by Andrew Schmitt | July 20, 2007, 3:58 PM
  4. First a correction: Vik Saxena is not the Comcast CTO. He is the Senior Director of Network Architecture at the CTO Office at Comcast Cable. He is responsible for IP/optical transport and HFC/access network architectures. So PON definitely seem to fall in his domain.

    Overall, I would not count on Cable to make any meaningful impact on PMCS revenues. I don’t see Cable deploying PON to residential neighborhoods in the visible future. Wall Street will kill the Cable companies for even a mention of such a move. Cox, who for years told Wall Street that a 750MHz plant will be sufficient forever, chose to simply upgrade to 1GHz when they went private.

    Cable operators are, however, using individual fiber runs to provide business services whenever expected revenues justify installation costs (this includes cell tower backhaul). Currently, CWDM is popular due to its low cost multi-link capacity, with most links being GE. When you consider that in NA most of the cost is in laying fiber (and not in edge equipment), this is very understandable. PON cannot compete with the capacity of multiple individual GE links. Nevertheless, PON is always being considered, and at some point it may take over.

    Residential PON is considered whenever Greenfield deployments, such as new housing developments, come up. But so far, the difficulties in maintaining two distinctly different systems (HFC and PON) prevented cable operators from actually deploying this option.

    Posted by Adi Bonen | July 21, 2007, 9:57 PM
  5. I didn’t mean to imply Cable would make an impact on PMCS. But cable deploying fiber to reach the enterprise is a strategy change which would impact many.

    I have heard that cable guys are being forced to install fiber to the prem in new developments because the homebuilders are demanding it.

    Posted by Andrew Schmitt | July 22, 2007, 11:29 AM
  6. Cable has actually been using fiber to serve businesses (even PON) long before Verizon started its FIOS program. For residential deployments, there are still attempts to find a solution with simple RF E-to-O and O-to-E conversions that will enable making the PON deployment an extension of the HFC plant. The problem is in the upstream, where multi-frequency multiple-access is hard on the optics. Verizon actually started FIOS with its own version of an RF upstream solution. It was meant to enable utilization of cheaper traditional settop boxes. Later, Verizon abandoned that method in favor of MoCA enabled boxes.

    But you have heard right about new homebuilders demands. Some homebuilders still will let the cable company hook up coax as long as fiber is also installed for future use, but others don’t even install coax in the home. For now, what this means for Cable is a one-way DBS-like solution (no VOIP, Internet or interactive TV), and the Cable guys don’t like it a bit.

    Posted by Adi Bonen | July 23, 2007, 10:18 AM
  7. “I still would like to see someone somewhere step up to the plate and quantify this effect. Without data this is another “internet traffic doubling every 9 months” type of proclamation…”

    I don’t have the links handy, but here are some data points:
    – AT&T has reported that their IP backbone traffic is doubling every 18 months (60% CAGR)
    – Level(3) has shown charts that indicate their IP backbone traffic is doubling every 14 months (80% CAGR)
    – GlobalX has shown charts that indicate their IP backbone traffic is doubling every 13-14 months (85%-90% CAGR)
    – Infonetics Research projects a CAGR for IP backbone traffic of 60%

    The qualitative factors behind this include:
    – Number of broadband access users in the US is projected to double between 2005 (46M) and 2011 (92M)
    – Average capacity provided to each user is going to increase by a factor of between 10 and 30 (768 kb/s-1.5 Mb/s was typical a year or two ago – and still is probably the majority of DSL access; 10 Mb/s – 30 Mb/s DOCSIS or FTTX will be typical five years from now)
    – Average utilization of that capacity will increase (video has a higher duty cycle than web browsing).

    I don’t think “doubling every nine months” is realistic, but a 60%-80% CAGR seems in the ballpark.

    Posted by DG Lewis | July 24, 2007, 7:48 AM
  8. Just because backbone traffic is going up doesn’t mean it is internet video. I venture a big proportion of this is Enterprise.

    Duty cycle is an interesting term. We need the broadband equivalent of an Erlang….

    Posted by Andrew Schmitt | July 24, 2007, 8:01 AM
  9. Hey, an erlang’s an erlang. If your broadband connection is capable of handling 6 Mb/s downstream, 21.6 Gb in an hour is one erlang. 5% utilization, .05 erlangs.

    Look at it this way. YouTube (320×240, 30 fps MPEG4) videos run somewhere around 1.5 Mb/s. A good-sized web page is probably somewhere between 3 and 6 Mb. (As I write this, is 735 MB and the lead story on Barry Bonds is 368 MB, including all files.)

    So say I go to, download the top page (about 6 Mb – takes a couple of seconds), then I scan over it (maybe another 10 seconds), then I click on the Bonds story (another second or two to load), then I spend two or three minutes reading it. Total download about 9 Mb, total time about 200 seconds – total capacity consumed 45 kb/s on my 4.5 Mb/s cable modem, or .01 erlang.

    Instead, let’s say I go to YouTube and watch a three minute video. Now, I’m consuming 1.5 Mb/s of my 4.5 Mb/s nominal service for the entire three minutes – 0.33 erlang, 33 times the utilization of reading the article.

    That’s why internet video drives bandwidth.

    And that’s just crappy 320×240 three minute YouTube clips.

    Posted by DG Lewis | July 24, 2007, 5:14 PM
  10. A couple of comments after review the discussion above,

    1) I agree with data traffic doubling between every 1 to 1.5 years (with one year as very optimistic uplimit), due to rapid spread of fiber access, which driving core transport. There has been extensive studies in ieee world for this topic since last year and the industry rech consensus.

    2) First time after the bubble BW glut finally ends in core/carrier transport, take an example, tyco’s recent trans-pacific express networks.

    3) Cross-reference with several resources, video traffic will overwhelm everything else in the coming years. In 2010, 90% traffic will be projected video. I can provide even my presentations.

    4) Market conditions improving is industry wide, which many companies see benefits. Based on PMCS CC:
    “In the second quarter, we experienced improvement in our enterprise storage, fiber to the home, and telecom businesses on a sequential basis,” said Bob Bailey, chairman and chief executive officer of PMC-Sierra, in a statement. “We believe the overall business environment is improving as we go into the second half of 2007, and we are executing on our corporate restructuring to improve the Company’s operating performance going forward.”

    Posted by FC | July 28, 2007, 4:32 PM