Comms chips

Will Intel Trash Telecom?

With Intel Corp. (Nasdaq: INTC) preparing a sweeping reorganization, the company might have to make some tough decisions when it comes to its communications efforts.

Intel will go through 90 days of review before launching the reorg, which CEO Paul Otellini announced on April 27. So, it's early to start declaring what Intel should keep or chuck. But its telecom-related products seem a particularly apt sector for soul-searching, since so many observers have wondered over the years whether Intel really belongs there.

"Intel has to decide if they want to be a serious player in the communications market," says Linley Gwennap, principal analyst with The Linley Group .

Otellini promised the reorg will leave no division unprobed. (Ouch.) "What we plan to do is a very deep analysis, go through operation by operation," he said in a presentation to analysts last week. The goal is to cut spending by $1 billion, or roughly 8 percent. Headcount is expected to drop 2 or 3 percent, but through attrition, not through massive cuts -- at least, that's what Intel has said so far.

His decision was spurred by increasingly tough competition from processor rival Advanced Micro Devices Inc. (NYSE: AMD) and lackluster first-quarter earnings reported last month: profits of $1.3 billion, or 23 cents per share, on revenues of $8.9 billion, compared with profits of $2.2 billion, or 34 cents per share, on revenues of $9.4 billion a year ago. (See Intel Reports Q1.)

Some are drawing parallels to Intel's 1980s transformation, when it ditched commodity memories and turned into a microprocessor company -- a move that worked out, one could say.

Intel declined to comment for this story, pointing out that the 90-day review is in its first week. And Otellini did tell analysts that the company had no preconceptions, saying, "We don't go into this knowing the answer."

Veteran Intel watchers aren't convinced the company will give up on communications, but some changes might be in order. "If they want to be in the comms space, they have to look at how to be a major player in this market, to the extent of a Broadcom Corp. (Nasdaq: BRCM) or Marvell Technology Group Ltd. (Nasdaq: MRVL) or Texas Instruments Inc. (NYSE: TXN)," says Linley. That may mean making a big acquisition somewhere, thereby picking up a cluster of executives who understand the networking-chips space better than Intel's home-grown teams do.

Intel's communications acquisitions have mostly been small, many coming around the year 2000. They were driven by then-CEO Craig Barrett, who pegged networking as Intel's next growth driver, prodding Intel to get into businesses like network processors, Sonet/SDH chips, and optical transceivers. (See Intel To Acquire Optical Chipmaker, Intel Swaggers Into OC192 Market, Intel's 10-Gig Shopping Spree, and Intel Capital Still Looking for Deals.)

Of course, the dotcom bubble popped, and some of those acquisitions have quietly gone away. But what really tripped up the strategy was trying to "force those companies into the Intel Way," Gwennap says. "If they were really serious, they'd have to bring in some people who know about the comms business and let them do their thing."

Network processors, arguably the most prominent of Intel's networking products, still show some promise -- "I'm not sure that's one of the areas they would want to cut," says Needham & Co. analyst Charlie Glavin, noting that Cisco Systems Inc. (Nasdaq: CSCO) has shown some short-term interest in Intel's IXP family of chips.

The problem is, Intel had to back off from high-end network processors, which haven't formed into much of a market. (See Intel Isn't Through With NPUs.) Intel is finding success in DSLAMs and access boxes, competing with the likes of Wintegra Inc. But with network processors only a $175 million pond to begin with, according to The Linley Group, Intel has to consider whether a fraction of that market is worth fighting for.

"For a company the size of Intel, you can't just keep segmenting your market down and down until you find a niche you can succeed in," Gwennap says.

Intel also has its optical transceiver and tunable laser products. But optical components have been a tough market, and at OFC/NFOEC in March, executives from companies including Intel questioned whether stubbornly low optical margins could continue to finance R&D in the sector. (See Smiles Abound at OFC/NFOEC.)

On a panel preceding OFC/NFOEC, Intel vice president Gary Wiseman pointed out several emerging markets where Intel sees prospects, including transceivers for 10-Gbit/s Ethernet and 4-Gbit/s Fibre Channel, and the growing use of tunable lasers. He also noted Intel is working on optical interconnects related to the computing industry.

That last point -- up-and-coming research -- could pose more tough questions for Intel. Because of its resources, the company has created research projects all over the place, in "spaghetti on the wall" fashion, Glavin says. He's particularly fond of Intel's silicon-laser experiments, which involve pumping a silicon chip with photons -- taking advantage of the famed Raman effect [ed. note: so it's more like noodles on the wall] -- to produce a laser. (See Intel Claims Laser Breakthrough.) That kind of R&D should have a place, even if it's tough to see how it could lead to revenues, he says.

"Everything is kind of interesting, and if Intel in their restructuring says certain things are primary research, and should be treated like a VC portfolio, that's fine," Glavin says. "My big concern is that Intel, in the process of restructuring near-term, kills a lot of primary research that could be very strong."

— Craig Matsumoto, Senior Editor, Light Reading

Page 1 / 2   >   >>
Pete Baldwin 12/5/2012 | 3:55:08 AM
re: Will Intel Trash Telecom? The angle here might seem arbitrary, but ... for the past year or two, competitors have talked about Intel losing interest in network processors and/or optical. Intel denies that at every turn, but you do have to wonder. If they ever wanted to get out of a particular business, now's the chance.

Like the story says, analysts aren't ready to say Intel should jettison either one. If you ran Intel, what would you do?
paolo.franzoi 12/5/2012 | 3:55:08 AM
re: Will Intel Trash Telecom?
Intel has abandoned the embedded systems market and communications many times. You have probably forgotten the litany of dead Intel things:

- 8085
- 8051
- All the peripheral chips (8255, 8275, etc.)
- i432
- 80376
- 80186/80188
- Good lord even the LXT310 (Where are you Dr. Pepper?!?)

Intel has had this behavior ever since IBM in Boca Raton built the IBM PC off the 8088. Focus on the PC market waxes and wanes. Every time it waxes or Intel decides to "focus, embedded systems developers and communications developers get screwed again.

I remember fondly the day my Intel sales team showed up and said, "Guys, we are dropping out of the 8085 business. You can not buy anymore from us as of today. I hope you can find other vendors to fulfill your needs. Can we sell you some 8088s?"

Mark Sebastyn 12/5/2012 | 3:55:07 AM
re: Will Intel Trash Telecom? I wrote about this last week.


The divisions return on invested capital is disgraceful. Regardless of whether Intel keeps the division or not, new leadership and metrics needs to be established with a better eye for returns. If Intel can't stomach this themselves, then get KKR or Silver Lake to do the same thing they did with Agilent's storage business.

I think the best solution is to spin it off, and then start pulling in all of the zombie companies that never pulled out of the post-bubble doldrums, consolidate product lines, eliminate duplicative SG&A and R&D in the business, and build a solid telecom silicon supplier.

As I said in my post, I think this is the best opportunity yet for consolidation in the telecom semiconductor business.
gadfly 12/5/2012 | 3:55:07 AM
re: Will Intel Trash Telecom? It is only a matter of time until you read that headline.
History shows many others attempting to master the optical/photonics components business then dropping the ball and walking away.
This spans from IBM who did this decades ago to the many recent "post optical bubble" dropouts from the active optical components business such as Alcatel, Nortel, Corning, Marconi and the hundreds of startups.
But the special irony of the Intel case is that over the long term optical communications will be an important performance edge in systems applications where Intel must remain a key player. So maybe they will drop a few more billion into this dry hole before recognizing that commodity producers will be the most efficient sources for those components.
whyiswhy 12/5/2012 | 3:55:05 AM
re: Will Intel Trash Telecom? Optical at Intel:

"If it don't fit, you got to quit"

OPtical clearly does not fit the Intel business model, which is as many have pointed out: x86.

Meaning this quarter its 386, next its 486, etc. with zero backwards compatibility at the hardware level.

Buy more chips or die.

That does not work in optical. The systems are not complex enough, at least not at the module level.

That's takes a rack play like Cisco.

More on Cisco later.

^Eagle^ 12/5/2012 | 3:55:04 AM
re: Will Intel Trash Telecom? I read your note about Intel posted in your link. Nicely written.

I agree with you regards "As I said in my post, I think this is the best opportunity yet for consolidation in the telecom semiconductor business."

What do you think about the OPTO part of the business? Their transponder, pluggables, etc. (I am leaving out the silicon optics r&d that is being done in Intel "skunk" works labs... I think that r&d is quite interesting and should be continued.. for other reasons than telecom..*).

What do you think about what I like to think of as their "classic" opto telecom modules business?

I think they have spent roughly a Billion dollars to get into this business via acquisitions. Ok.. maybe 800million plus.

That is not counting the ongoing burn rate that is one of the highest out there relative to sales in otpo modules.

Excellent products and some great pockets of engineering, but wrapping Lightlogic, New Focus, Sparkolor, and others with "Intel" has made their cost structure way high it seems to me and made it difficult for them to compete with more flexible and dymanic companies. Their one strength in the market has been that they are a safe bet compared to the "risky" start up plays. After all, buying a module from Intel is safe right? They have lots of money and are in the game for the long haul. I think a lot of the business they have is for that reason.

Customer feel safe especially given how many start ups have folded over the last few years.

If Intel makes customers feel at risk regards ongoing supply of opto modules from Intel and long term commitment to sector, their strongest card in gaining market closing business for Opto modules (I am leaving out silicon for processors, framers, et al) is taken away. This accelerates the bleeding. Intel has to either make some very positive and concrete statements that they are IN THE GAME for the long term, or they have to move quickly to get out as the uncertaintly takes away their best sales tool.. Intel opto is strongly backed and a safe supplier.

As someone (I think seven) posted, Intel has dropped customers for communications products and telecom before. This has got to be making their large customer feel a bit shaky right now. And the poor sales & marketing staff trying to ramp their sales for optical transponders.... and the product managers that are working on next generation roadmaps.. I think these recent announcements could cause customers to slow purchases and make other decisions. Hypothetical: if there was a 10% or 15% reversal in sales growth due to uncertainty, would that not accelerate losses in this group? Especially as they have been investing in people and manufacturing to support the growing market and growing demand. Investment always precedes any growth in revenue or margins and shows up on balance sheets as loss. but if they lose market share due to uncertainty?

How much margin do they have to make on the transponders and other modules to pay back the investment made and make a profit? How many transponders do they have to sell to justify the investment? Since their results are hidden in this large division it is hard to say.

Re: spinning out the business. I would suspect that you are correct there. Their big customer, Cisco would want to keep that supply chain alive if possible I would guess. They could find other suppliers, but it would take some time and cost.

If you have any thoughts on the opto sub sector of their communications business, I would appreciate hearing them.

I think Intel has to be very careful and somewhat decisive when they make what ever decision they are going to make, otherwise the very process of evaluation and the side affect that all momentum gets stalled while the customer base waits could accelerate things out of their control.

jmunn 12/5/2012 | 3:55:04 AM
re: Will Intel Trash Telecom? Let's not forget the 960 processor line.

"You can still buy them but forget about
that roadmap we showed you."

If they seriously burn embedded again it will be difficult to recommend using Intel parts in future embedded systems.

Unfortunately for Intel almost all new designs are "Embedded Systems" from STBs to game machines to cellphones/pocket computers.

The other thing I fear is Intel central control falling into the GM like mindset that "People really don't want anything besides faster PCs so we don't have to worry about these little gadgets they are hooking to their TVs."

For Intel to actually grow they have to move beyond just the mature PC market.

I just feel sorry for the people whos lives will be caught up in the "Oops, we weren't paying attention to the business and it turns out we don't need you anymore" realignment.

st0 12/5/2012 | 3:54:58 AM
re: Will Intel Trash Telecom? (1) Before intel get into the telecom, there was intel mil-grade processor. Similar story. Still remember the last buy call came in with shock wave for new aircraft design that not even pass the fly test. Does anyone know the price of change key components after Critical Design Review? Easy for the sales guy come in to tell you that newer and better product you can use, but not form, fit and function (by the way, the temperature range just a bit of narrow)... You can still find many old mil-grade Intel chip stored in long term storage as support part some places in the field, designed to support of 15 years...made some "after market ware house realy rich) that lead to another point...
(2) Just like Mil-grade, Telecom required support of many years (15-25). Lack of will at Intel to commit any platform for such a long time make Intel disqualify as a game player in the Telecom. (Told them years ago, I wouldn't use Intel after the Mil event... many people have long memories... However, many of us fade out in the field..new chaps just have to learn their own lesson... ). Unless the committment issue is resolved, Intel's business model does not fit the long term requirements of the Mil or Telecom products. (leave it to the expert, please).
(3) I am sure Intel lost some of the expertise that design the high rel product group (I do not meant comply to the specs). The mind set of high rel product development is rather different compare to the commercial grade. Got 1st hand discussion with some engineer regarding triple redenduncy use 3 chips... compare with on-chip redenducy use 3 path... the newbies just couldn't see the difference. (lack of concept in alpha particle damage in the design group is shocking... The design guys just too far from the reality..too much Spice simulation cause brain damage... Moore's law compliance got into the flashy design).
(4) Don't get me wrong, Intel is excellent company to get into any playing field their heart desire. (beside a bit too much heat..that is another story... PC do not have building code that have to comply, but the network gear requires GR to pass the cooling requirements... too much heat limited its application) Just the committment issue (sounds like boyfriend/girlfriend issue... hmmm). Short term, no problem, Long term, ???
(5) just my 2 cents.. Don't flame me using your duo-core please.. Really want Intel back on track.. what ever the path they choose.
sr_ns 12/5/2012 | 3:54:52 AM
re: Will Intel Trash Telecom? To: st

Could you please write/explain your point 3 in better English?

I'm sorry, I don't really understand what you are trying to say..

redface 12/5/2012 | 3:54:51 AM
re: Will Intel Trash Telecom? It may not make a lot of sense for Intel to trash telecom, since there is currently a lot of renewed interest in FTTH PON and telecom in general. What Intel should do is to ditch their moribund "silicon photonics" and start over with something else.
Page 1 / 2   >   >>
Sign In