Mergers & acquisitions

Intel Snaps Up Templex

Intel Corp. (Nasdaq: INTC) has quietly acquired Templex Technology Inc., a small optical components company in San Jose, Calif., for an undisclosed sum. The deal closed late last month, Atul Tambe, Templex's vice president of engineering, told Light Reading on Monday.

At press time, Intel had not returned calls for comment. There were no press releases or SEC filings published by Intel regarding the event, but it may be that the deal was small enough to go through under the radar.

Templex is not a large company -- it had only 25 employees at its peak and now employs about 10 -- but the acquisition is interesting because of the technology involved. Though Templex was selling Fiber Bragg Gratings (FBGs) for Wavelength Division Multiplexing (WDM) networks, the company has been developing a technology for adding more capacity and management capability to fiber optic networks that some view as a replacement for WDM.

Templex was founded in 1995 to develop Temporally Accessed Spectral Multiplexing (TASM), a technology developed by University of Oregon physicist Thomas W. Mossberg and former University of Washington professor W. Randall Babbitt. (Mossberg was the company's first chief technology officer, but he returned to teach at Oregon in October 1999.)

TASM's claim to fame was that it allowed for intelligence to be encoded and processed in the optical domain, a trick that could add speed and capacity to optical networks. The most immediate application of TASM technology was in the implementation of optical code-division multiple access (OCDMA), an innovative way of adapting CDMA cell phone technology to add more capacity and management features to fiber optic networks (see Optical CDMA Product Stirs Debate).

OCDMA has yet to gain commercial acceptance, though a couple of companies have (unsuccessfully) positioned it as a technology alternative to dense wavelength-division multiplexing (DWDM) in metropolitan networks (see CodeStream Goes Under).

Of course, it's too early to say what Intel will do with the OCDMA technology. At the time of its acquisition, Templex was only selling fiber Bragg gratings and hadn't shipped an OCDMA product (see Startup Unveils Fiber "Switch"). It was, however, tinkering with ways to provide a migration path from WDM to OCDMA.

"We're not integrated into Intel just yet, so it's hard to say what's going to happen with all the stuff we have," Tambe says. "I think the time for OCDMA will come; I'm just not sure that we're there today."

Intel's purchase of Templex is not surprising, given that Intel had invested some $5 million of the nearly $8 million the company had raised since its inception. In 1997, Templex, then in Eugene, Oregon, completed a small funding round with Bison Ventures (now defunct) as the lead investor.

By the end of 1999, Templex had attracted investments from Intel; the Oregon Resource and Technology Development Fund; AVenture Partners LLC, a venture firm run by Intel veterans; Bison Ventures; and eFund, a venture firm formed by former Bison partner and Templex board member, Joe Tanous.

In 2000, Templex left Eugene for San Jose and now occupies an office less than five miles from Intel's corporate headquarters in Santa Clara.

Intel has snapped up several components companies this year, including its February acquisition of nSerial Corp.; its April acquisitions of VxTel and Cognet; and its May acquisition of LightLogic (see Intel Scoops Up Chips)

In related news, Intel's director of components research, Gerald Marcyk, is hosting a technology briefing conference call on December 18 at 10:00 a.m. PST.

Over the past few years, Intel has been paying more and more attention to developing products for optical networks and the service providers who build them. Last week it made much ado about its plans for providing servers that handle computing-intensive telecom applications, such as IP messaging (see Intel Targets Telecom Server Market).

— Phil Harvey, Senior Editor, Light Reading
Page 1 / 2   >   >>
flanker 12/4/2012 | 7:25:33 PM
re: Intel Snaps Up Templex OCDMA means no wavelength translation.

Wonder if Intel has the balls to buy a private optical switch and transport vendor and see if they can patch together an OCDMA network solution.

MadMark 12/4/2012 | 7:25:32 PM
re: Intel Snaps Up Templex Everybody working on OCDMA as a possible "replacement" for DWDM is wasting their time. There is simply, absolutely, fundamentally NO WAY that OCDMA systems can achieve capacities even remotely approaching those of DWDM.

In wireless CDMA systems, the biggest problem is interference from other users. Spreading and despreading techniques are used to encode and decode wanted channels, and make all the unwanted channels look like so much white noise. For a long time, it was hard to do, but modern digital signal processing means that it is a solved problem. Researchers always knew that it was a matter of getting the signal processing right.

But in the optical domain there's a problem that no amount of signal processing will ever solve. At radio frequencies, quantum noise (shot noise and spontaneous emission noise) is negligible. Aside from the interference from other channels, the noise sources are all ordinary thermal processes, and they're independent of the number or power of the channels.

However, at optical frequencies quantum noise processes are not negligible. In fact they can be the dominant noise sources. One hundred channels hitting a receiver generates one hundred times the shot noise of a single channel. It doesn't matter how good a job your signal processing does of suppressing the interfering signals, or whether you do it in the optical or electrical domain, you cannot suppress all that noise! And no, you can't filter out these other channels -- that would be DWDM! :-)

Some OCDMA approaches use coherent broadband sources (i.e. short pulses). Others use incoherent broadband sources (e.g. LED's, or ASE noise sources). Some use time spreading, other use coherence effects. All are ultimately limited in capacity by either shot noise, or "excess" (spontaneous emission) noise. And that capacity is WAY below even a modest DWDM system. In the optical domain, there is no substitute for filtering!

Research papers on OCDMA usually demonstrate an encoder, with "matched" and "unmatched" decoders, so you can see that a pulse is reconstructed by the matched decoder, but not the unmatched one. Some demonstrate the transmission of a small number of channels (often only two), to show that the matched decoder extracts the wanted channel only. But no paper I have ever seen demonstrates this using a realistic number of channels. And you'll never see a theoretical analysis that produces an optimistic assessment of the capacity of OCDMA that correctly includes those pesky quantum noise sources.
openmind 12/4/2012 | 7:25:26 PM
re: Intel Snaps Up Templex My Apologies for sloppy spellings in my previous post.
openmind 12/4/2012 | 7:25:26 PM
re: Intel Snaps Up Templex MadMark,

Generally a good analysis but still "inside-the-box" thinking. There is no reason to replace DWDM jsut enhance it with codes if price of the solution can be brought down. Even small number of codes on one wavelength on top of DWDM can give you more number of channels then otherwise possible. The enhancement of DWDM could be in many forms not just raw capacity improvements.

History is full of "Wet Blankets" theat were very successful so who knows.

- Openmind
flanker 12/4/2012 | 7:25:25 PM
re: Intel Snaps Up Templex Everybody working on OCDMA as a possible "replacement" for DWDM is wasting their time. There is simply, absolutely, fundamentally NO WAY that OCDMA systems can achieve capacities even remotely approaching those of DWDM.

If you read between the lines, he's really saying "I'd like to buy the world a Coke".
HarveyMudd 12/4/2012 | 7:25:08 PM
re: Intel Snaps Up Templex Intel so far has acquired over 10 optical networtking companies. But the acquisition of these companies have not added to the bottom line of Intel.

Intel is essntially chip ompany with no significant group of scientists to make the optical equipment work. IMprovements are needed ac the acquision of the companies was at a raw level. Given the history and tradition of Intel, it can not make significant impact both in terms revenue and marketability of optical products.

There is no end to claimimg the analog of technologies in the wireless dommain to the optical arena such as ( OTDM, etc.)This is a big joke.

Intel thinks that it can turn optical products into commodity products. To the utter dismay of Intel, this will not happen in spite of the very fast declining optical networking industry.
rjmcmahon 12/4/2012 | 7:25:07 PM
re: Intel Snaps Up Templex Intel so far has acquired over 10 optical networtking companies.

Where were they all located? Maybe their making a play for government support and a new telecom act?

Intel's primary business does seem gated by the expense of bandwidth. Many others, like Intel, cannot grow when bw is priced so high.

Convincing consumers to pay for *content* seems like the other missing piece of the puzzle before free trade will be realized over these networks.
buzzkill 12/4/2012 | 7:25:05 PM
re: Intel Snaps Up Templex How come Intel didn't make a bigger announcement out of this??
grendel 12/4/2012 | 7:25:05 PM
re: Intel Snaps Up Templex
These guys have been around for about 4 years, making big claims but showing very little. After all this time what do they have to show for it? $8 million total VC money, 25 total employees. $5 million came from Intel. At a guess this means that Intel owned over 50% of the company.

I'm guessing that Intel paid off the previous investors for the remaining shares, probably for less than they originally put in, and left the common shareholders with nothing. So, they toss an extra $1.5 million or so after the $5M that they wasted and buy the equipment to make FBGs in house. FOr this investment, they avoid the headline "Intel's OCDMA Startup Tanks", and the embarassment of having invested in this joke in the first place. Notice the lack of an Intel press release. They ain't braggin about this one.

The Oh So Talented 25 employees who failed to create anything of value after 4 years are probably just thrilled to have jobs with Intel, since the alternative would be job hunting in tough times having failed in their most recent endeavor.

I see nothing in this to indicate that ANYONE, Least of all Intel thinks OCDMA is going anywhere.

Steve Saunders 12/4/2012 | 7:25:04 PM
re: Intel Snaps Up Templex grendel,

bitter, much?

Page 1 / 2   >   >>
Sign In