x
Comms chips

AMCC Retreats From France

Applied Micro Circuits Corp. (Nasdaq: AMCC) is closing its French subsidiary as the company backs away from the switch fabric market, at least for now, according to a Securities and Exchange Commission (SEC) filing yesterday morning.

During the next few quarters, AMCC expects to incur restructuring charges totaling $4.5 million to $11.5 million.

AMCC built its French operation from a pair of deals with IBM Corp. (NYSE: IBM). AMCC bought IBM's PowerPRS switch-fabric chipset for $51 million in 2003 and picked up an IBM PowerPC product line months later for $228 million. Dozens of employees were transferred to AMCC in the process. (See AMCC Switches On IBM's Fabric and AMCC Powers Up.)

Late last year, AMCC decided it would stop further development of the PRS. Moreover, most of the PowerPC-related employees were returned to IBM in February, the result of a labor decision in the French courts. According to the SEC filing, the combined events leave the French subsidiary too small to be worth maintaining, so AMCC has begun the process of shutting down the subsidiary.

Around the year 2000, several chip makers developed switch fabrics. But many router and switch vendors still prefer to develop their own switch fabrics and have been slow to adopt merchant chips, according to Sam Fuller, an AMCC vice president of marketing. Moreover, any further generations of the PRS would probably have to be reworked to accommodate today's emphasis on Ethernet: "The market is moving to 10-Gbit/s Ethernet as a backplane technology," Fuller says.

AMCC does have customers for the PRS and will continue supporting them for years, says Fuller. To that end, he adds, about one third of the 56 employees in France have been offered jobs in the United States.

But the PRS roadmap ends here. "We're not planning to develop a second-generation architecture based on the PRS," Fuller says.

AMCC has now taken three swings at the network processor market without knocking a home run. It had developed the nP5800 line inhouse, but shelved that in favor of the PRS. Separately, AMCC acquired switch-fabric startup YuniNetworks in 2000 -- development of that company's Cyclone chipset was halted some time ago, and Fuller believes the product doesn't have any customers left. (See AMCC's Triple Acquisitions.)

Fuller wouldn't rule out the possibility of AMCC trying yet another switch fabric. It's just that the company doesn't see the PRS meeting the needs of the next generation.

Startups Dune Networks , Enigma Semiconductor Inc. , and TeraChip Inc. continue to pitch switch fabrics to OEMs. Another company with a switch fabric and other related chips, Sandburst Corp. , was recently acquired by Broadcom Corp. (Nasdaq: BRCM). (See Siemens Invests in Dune, Enigma Unlocks Packet Switch Fabric, TeraChip Raises $7M, and Broadcom Builds on Sandburst.)

AMCC shares rose 19 cents (5%) to $3.99 yesterday.

— Craig Matsumoto, Senior Editor, Light Reading

Pete Baldwin 12/5/2012 | 3:58:07 AM
re: AMCC Retreats From France At what point do switch fabrics become a healthy market? they seem to be lagging network processors (as most people predicted years ago)... and for that matter, are we even at a point where the network processor market could be called "healthy"?
Mark Sebastyn 12/5/2012 | 3:58:07 AM
re: AMCC Retreats From France Switch Fabrics are a healthy market. It's just proprietary switch fabrics that aren't. And it is hard to see how that will change going forward given the standardized solutions available.
pjensen 12/5/2012 | 3:58:04 AM
re: AMCC Retreats From France With all the consolidation happening, I won't be surprised if AMCC would think about buying Ethernet based switch fabric/network processor company? Anyone knows what those target companies could be?
Pete Baldwin 12/5/2012 | 3:58:03 AM
re: AMCC Retreats From France With all the consolidation happening, I won't be surprised if AMCC would think about buying Ethernet based switch fabric/network processor company? Anyone knows what those target companies could be?

I like that avenue of thought, although i'm not sure AMCC would do the buying. Xelerated took the Ethernet route with its switch fabric some time ago.

Greenfield could be an interesting possibility.
sigint 12/5/2012 | 3:58:03 AM
re: AMCC Retreats From France schmitt:
Switch Fabrics are a healthy market. It's just proprietary switch fabrics that aren't. And it is hard to see how that will change going forward given the standardized solutions available.
___________________________________________________

Could you please elaborate? My own experience with Standards based switch fabrics (ethernet, ATM, PCIe) has been that they are often too restrictive for any serious USP or value add to be built in. Probably ok for low marging commodity stuff.
whyiswhy 12/5/2012 | 3:58:01 AM
re: AMCC Retreats From France Craig:

France has succeeded in converting itself into a Moslem third world country. Still with an attitude, of course.

-Why
Mark Sebastyn 12/5/2012 | 3:58:01 AM
re: AMCC Retreats From France You're right, they typically are too restrictive. The problems with proprietary ASSPs are:

1. You are permanantly wedded to one vendor
2. Still likely misses some of the key features you need
3. High cost

What then happens is the 2-3 high volume high-end players go with ASICs (aka Cisco, Juniper), the high-volume low end stuff goes to standard fabrics (aka Ethernet), and ASSP fabric companies are left to chase tier 2-3 players. It's hard to secure the big customer that will provide the major volume you need to support the R&D.

Telecom is now commoditized. If you can't do it off the shelf or with an open standard, or do the ASIC yourself, it just doesn't get done.

It's a very ugly business.
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE