x
Optical/IP

MEMX Starts Anew on 3D MEMS

Three-dimensional MEMS (micro-electro-mechanical system), an approach to building all-optical switches, is not dead. At least that’s what startup MEMX Inc. says.

MEMX is a just getting started. The company, which is being spun out of Sandia National Laboratories, yesterday officially announced the appointment of its president and CEO and the close of its second round of funding, the amount of which hasn't been disclosed (see Mike Hodges Joins MEMX).

MEMX’s emergence comes at an interesting time in the development of all-optical switch modules. Simple 2D MEMS, which switch lightpaths from one mirror to another across a two-dimensional matrix, are currently being used in most all-optical switches, but these fabrics don't scale easily beyond 32 ports. To increase scaleability, companies have been developing 3D MEMS, which can be used to build a three-dimensional array of mirrors that can reflect light in multiple directions.

But several issues have plagued 3D MEMS development. For one, the packaging of the devices is still done by hand, requiring fiber alignments in the sub-micron range. Also, the control of 3D MEMS is analog, which makes it much more complicated than the digital control of 2D MEMS. And because the silicon used to build 3D MEMS is also immature, there are reliability questions.

Just last week Onix Microsytems Inc., which had recently raised $77.5 million, cut back its staff by 25 percent and put its 3D MEMS product development on the back burner (see Onix Follows in OMM's Footsteps). A month previously, OMM Inc. also shelved its 3D MEMS development and cut 100 more jobs (see OMM-inous News). Both companies are instead focusing on development of their 2D MEMS products.

“We need to stretch our cash,” says Meng-Hsiung Kiang, co-founder and director of business development and product management for Onix. “So we have to be careful how we spend our money. The market for large-scale switch fabrics seems to have been pushed out.”

Another MEMS startup Integrated Micromachines Inc. (IMMI) is also rumored to be sidelining its 3D MEMS endeavors. But company officials deny these rumors.

"We're vigorously pursuing 3D MEMS and we're talking to multiple customers," says Steve Walker, IMMI's director of subsystem integration. "Everything we've talked about for the last six months is full steam ahead."

Why would MEMX pursue 3D MEMS when other startups are retreating? The company is still keeping specifics of its technology quiet, but two factors bode well for the company. For one, the new CEO is Mike Hodges, a CEO-for-hire with a track record of jump-starting startups. His most successful venture so far has been Tellium Inc. (Nasdaq: TELM), where he served as CEO from January 1999 until February 2000. While there, he hired the company’s current CEO, Harry Carr.

From Tellium, Hodges went to Bandwidth9 Inc. for four months and soon moved onto Onetta, an optical amplifier startup that has raised $68 million, where he hired former Redback Networks Inc. (Nasdaq: RBAK) CEO Dennis Barsema as permanent CEO for the company. Since leaving Onetta, he has served as CEO for Silvan Networks -- a Layer 7 IP routing company -- until the middle of the summer, when he finally took on the role at MEMX, another small company which still has only about 15 employees split between two development sites in Albuquerque and Sunnyvale, Calif.

But this time, Hodges says he plans to keep his title, instead of moving on in a few months. Others who have worked with him in the past say they have no doubt he will be successful as a long-term CEO.

“He’s a very strategic thinker,” says Krishna Bala, co-founder and chief technical officer of Tellium. “He was able to absorb Tellium’s financials and our business in a very short amount of time. If he decided to stay with a company, I’m sure he would do a great job. He’s a sharp guy.”

The second positive for the company is its tier-one venture backing. While details of the latest funding round haven’t been released, the company has announced that big venture firms Sequoia Capital and Austin Ventures are involved. Additionally, Sandia, MEMX’s parent company, had already pumped about $100 million into 3D MEMS development before the company was spun off last year, says Hodges.

Hodges has his own ideas about why competitors are getting out of the 3D MEMS market. He says that 3D MEMS technology is complicated and that his competitors don’t have viable products, so they are using the carrier crunch as an excuse to scarper off.

“I guess I would say the market is there if the technology works,” he says. “They didn’t have a technology that works. We do.”

But in light of recent research from analysts, the 3D MEMS market doesn’t seem all that appealing. Over the next four years, the market for large, scaling, all-optical switch fabrics is only expected to be a small fraction of the overall MEMS market, they say. Marlene Bourne, with Cahners In-Stat Group, says photonics switches greater than 32x32 ports (specifically 64x64, 256x256, and 1,000+), are forecast to generate revenues of only about $130 million in 2005. Compare this to the total MEMS market, which is expected to rise from $67 million in 2001 to $2.3 billion in 2005, according to a report she published last week.

What’s more, she adds, the bulk of revenues in the 32x32 port and above category are for fabrics that would support 64x64 ports -- and rarely for the massive fabrics that scale to 256x256 ports and above.

Tim Anderson, an equities analyst with Salomon Smith Barney, agrees that the need for large-scale switches is not yet on the radar screens of most carriers. "The market for all-optical switches still exists," he says. "But it’s the timing and size that are still moving targets."

Despite the exit of OMM and Onix from the 3D MEMS scene, there are still plenty of players competing in this space, including big names like Corning Inc. (NYSE: GLW), Lucent Technologies Inc. (NYSE: LU), and Nortel Networks Corp. (NYSE/Toronto: NT).

Tellium's Bala is especially cautious of startups in the current economic climate: "I’m not completely convinced that they will all be around at the end of the day. But if other companies are focusing on 2D MEMS solutions and pushing 3D development further out, it could give a startup like MEMX a chance to really get the technology right.”

- Marguerite Reardon, Senior Editor, Light Reading
http://www.lightreading.com

HarveyMudd 12/4/2012 | 7:52:12 PM
re: MEMX Starts Anew on 3D MEMS Mr. Bala of Tellium is absolutely correct regarding the future of optical start-up companies. About 90% of the optical companies would disapear from the radar screen. Dr. Rohit Sharma of ONI Systems made similar predictions.

Many VCs have funded optical networking companies and continue to prop up their respective funded companies in spite of the fact that there is no chance of survival of these companies.

I strongly believe that any significant break throgh in the optical arena would come from Nortel, Lucent and Alcatel etc.
oxc 12/4/2012 | 7:52:08 PM
re: MEMX Starts Anew on 3D MEMS A startup frog jumps into pot of boiling water, realizes its hot and jumps out. Your legacy frog already in water can't feel heat and dies slow death.

Your ideal VC-less world is socialist failure to appreciate the spirit that drives the marvel of innovation.



lambdaswitching 12/4/2012 | 7:52:07 PM
re: MEMX Starts Anew on 3D MEMS Crazy or crazy like a fox?

LR said:

GǪMEMXGÇÖs emergence comes at an interesting time in the development of all-optical switch modules. Simple 2D MEMS, which switch lightpaths from one mirror to another across a two-dimensional matrix, are currently being used in most all-optical switches, but these fabrics don't scale easily beyond 32 ports.

But in another place in the article quoting Ms Marlene Bourne of Cahners et all.

GǪWhatGÇÖs more, she adds, the bulk of revenues in the 32x32 port and above category are for fabrics that would support 64x64 ports -- and rarely for the massive fabrics that scale to 256x256 ports and above.

So how long does it take to hatch a big 3D MEMS switch? Six months, 12 months, maybe 18 months. Seems like a reasonable business model to go and build a GÇ£better mouse trapGÇ¥ using VC money over a period of time when demand is not there, at least not quite yet, and be ready when the demand picks up. When you look at the companies that have, for now, dropped out of the quest for bigger switches, you have to ask why?

Simple answer, there is no demand for now for Big Ass OOO switches, and the companies that to date have been working in this arena have to answer to investor pressures. VCGÇÖs who funded companies two years ago, want to see positive returns and right bloody now, no patience for companies burning money for a demand that is at best 18 months to 24 months away. Dance with the devil and pay his tune when the note comes due.

So back to MEMX, crazy or crazy like a fox? Seems to me that MEMX is a GÇ£second waveGÇ¥ company and can learn from the mistakes made by others, watch the burn rate, do the engineering right and be ready to kick a little ass when the tide turns.

In my book, crazy like a FOX !

@lambdaswitching
manoflalambda 12/4/2012 | 7:52:02 PM
re: MEMX Starts Anew on 3D MEMS So how long does it take to hatch a big 3D MEMS switch? Six months, 12 months, maybe 18 months

At least 18 to get the product into customer labs, then beyond that to refine it, certify it and start getting some orders. Lucent's LambdaRouter is just at that stage and none of the others seem to be (OOO's that is, where are Calient and Xros these days?).

?What?s more, she adds, the bulk of revenues in the 32x32 port and above category are for fabrics that would support 64x64 ports ... to 256x256 ports and above.

Any signs of any 64x64 products on the horizon? Agere? Polatis?

Salute,
Manoflalambda
lambdaswitching 12/4/2012 | 7:52:00 PM
re: MEMX Starts Anew on 3D MEMS Manoflalambda said:


"At least 18 to get the product into customer labs, then beyond that to refine it, certify it and start getting some orders. Lucent's LambdaRouter is just at that stage and none of the others seem to be (OOO's that is, where are Calient and Xros these days?)."

So is that a vote for MEMX being Crazy like a Fox ?

So it seems Lucent is out with LambdaRouter getting certified, trying to close sales all those normal product introduction sort of things. It looks like they will have the field all to themselves for a while, is this a good or bad thing, hard to tell. As a target for others to shoot at, anyone like to hazard a guess as to how much a big (256x256) LambdaRouter is going to cost ?

"Any signs of any 64x64 products on the horizon? Agere? Polatis?"

No burning press releases here on LR about product introductions. I would think that 32 x 32GÇÖs based on square matrix configurations using 16 x 16 parts is just packaging engineering, 64 x 64 using multistage designs would be a bit more challenging. Hard part in any design will be getting through certification programs and those sort of hoops.

[email protected]
hawkman 12/4/2012 | 7:51:52 PM
re: MEMX Starts Anew on 3D MEMS It is a difficult endeavor, but all the startups that say there is not a need for a 3d MEMS OXC are covering for a lack of money or more likely, failed engineering. The big systems makers know they need an all optical OXC, and they are willing to pay for it. MEMS is the most viable answer right now, and although the 1000x1000 hyped days of XROS may be over, the smaller OOO switches (32x32 up to 256x256) are in demand. MEMX is "Crazy like a fox" but they are way behind companies like Intellisense, IMM and Calient. It is going to be great to watch who is still standing at the end of the year.
lambdaswitching 12/4/2012 | 7:51:51 PM
re: MEMX Starts Anew on 3D MEMS Hawkman said
GǣIt is a difficult endeavor,GǪ., the smaller OOO switches (32x32 up to 256x256) are in demand.
MEMX is "Crazy like a fox" but they are way behind companies like Intellisense, IMM and Calient. It is going to be great to watch who is still standing at the end of the year.GÇ¥

I donGÇÖt disagree with your contention that the field does have a lots of players right now, but the crux of my argument is that MEMX appears to be the start of a second GÇ£waveGÇ¥ of companies entering this space. Time in a sense is on MEMXGÇÖs side, time to build the product to meet the need that is not quite here yet. Investors in MEMX will give the company some amount of time to piddle around in the lab, some amount of time to build a market presence and some amount of time for the spending taps to turn back on, my guess probably on the order of 24 months or so seems to be the maximum horizon most investors can stand right now. Competitor companies who were funded in late 1999/early 2000 are being kept on a much shorter leash it seems to me, lots of GÇ£Fish or cut baitGÇ¥ messages being sent out.

IMHO the rules of the game have changed a bit over the last few months, from one of patient (some would say too patient) VCGÇÖs funding just about anything with GÇ£OpticGÇ¥ in the business plan to one of VCGÇÖs being put on the rack by investors saying GÇ£Show me the moneyGÇ¥ and right now, and it does flow down-hill.

It would be an interesting exercise to go through the history of the PC business and see who were the first wave companies, who were the second, third etc. companies and whoGÇÖs standing today. If my poor memory serves me right, DEC was one of the GÇ£First WaveGÇ¥ companies and yesterday the remnants of the company that brought the PDP-11 to market, was bought by HP which is now a printer company after having spun off itGÇÖs technology core.

History does have a way of repeating itself.

@lambdaswitching
iamnoone 12/4/2012 | 7:51:49 PM
re: MEMX Starts Anew on 3D MEMS "behind companies like Intellisense, IMM and Calient. It is going to be great to watch who is still standing at the end of the year.GÇ¥

It's amazing that in all this discussion the companies that are actually shipping 3D MEMS products to customers is missing -- Lucent/Agere. I guess if you've got it, you sell it, and not spend time talking about it.
[email protected] 12/4/2012 | 7:51:44 PM
re: MEMX Starts Anew on 3D MEMS So, hereGÇÖs my perspective GǪ port count is important, but so are a lot of other issues. reliability, loss, cross talk, scalability, switch speed, footprint, High on the list is insertion loss (or power lost going thru the switch). Many of the big switches have a 6 dB - 9 dB loss. 9 dB means 7/8 of the light is lost going thru the switch. This then requires an amp => $. This kind of defeats the purpose of OOO. Also high on the list are problems like the nearest neighbor problem where switching one mirror causes the mirror next door to wiggle a bit which causes the power on that connection to modulate.

The point being that on the big switches there a some tough problems which have yet to conclusively solved by anyone. Secondly, OEO switches work real well at the larger sizes. They can regenerate the signal (ie, reshape it and send it out with full power) and change the wavelength, both of which can be done by any OOO switch today. While raising money for Tellium in 1999 at a time when we were solely focused on OEO there was a lot of hype re OOO switches. I maintained that OOO had its place and that it was 5 years away. That makes it 2004. I think this is great time to start an OOO company.

mxh,
ceo memx

*OOO = all Optical switch
*OEO = Optical GÇô Electrical GÇô Optical switch
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE