Mahi Secrets Surface

Details about stealthy startup Mahi Networks Inc. are starting to surface. The company, which is building a multiservice provisioning platform (MSPP), is rumored to be close to signing a deal with a large RBOC. And it seems that one key to the startup's potential success could be in its partnership with semiconductor giant, PMC-Sierra Inc. (Nasdaq: PMCS).

Here's the scoop: Mahi, which anonymous sources say had a valuation of about $250 million after its $65 million second round of funding last June, is rumored to have a contract with BellSouth Corp. (NYSE: BLS) to develop its Mi7 optical transport switch router. If Mahi delivers the goods, the contract will be worth more than $300 million worth of business, sources say.

"This would make sense," says Dave Gross an analyst with Communications Industry Researchers Inc. He says that several people working at Mahi have strong ties with Bell South. For example, John Spencer, regional director of product marketing engineering, used to work at Bell South. Greg Peters the company’s president and CEO used to work at Advanced Fibre Communications Inc. (Nasdaq: AFCI), a company that sells telecom gear to RBOCs like BellSouth.

“One of the really good things about this company is that we have staffed it with folks that have a long history of meeting the needs of Bell operating companies,” says Chris Rust, a director on Mahi’s board and a principal partner at Sequoia Capital, one of Mahi’s investors.

Mahi denies that any contract has been signed. “It’s just not true,” says Ron Longo, vice president of sales and marketing for Mahi.

What makes Mahi so hot? The Mahi box promises to combine legacy transport, like Sonet and TDM; optical transport like WDM; IP routing; switching; and MPLS functionality into one platform. It’s specifically geared to metropolitan-area central offices where it can take the place of several, expensive boxes, putting Mahi head-to-head with other God-box companies like Maple Optical Systems.

Other startups have gone down this road before. Tenor Networks Inc. also tried to make a big MPLS God box for the metro core and has since found itself at a funding deadend.

What makes Mahi different from those other boxes is the technology inside. It has been able to leverage cutting-edge, off-the-shelf technology that should help it get to market at least six to 12 months sooner than if it had developed all its technology in-house.

Specifically, sources say the company is using the latest 320-Gbit/s capacity chipsets from PMC–Sierra for its switch fabric, instead of developing its own ASICs. Alcatel SA (NYSE: ALA; Paris: CGEP:PA) is also said to have a design contract for the same chipsets. The chipset, known as the TT1, came out of PMC’s 1999 acquisition of a startup called Abrizio. It’s believed that Mahi has been working with the Abrizio engineers for some time, since the two companies share a common VC: Chris Rust of Sequoia Capital. The TT1 chipset was announced back in April 2000, but has not shown up in commercial deployments yet, says Bob Wheeler, a research analyst from the Linley Group.

“Using off-the-shelf silicon can cut back on resources enormously, especially for startups like Mahi,” he says. “Not having to hire an ASIC design team removes a lot of expense and risk for the company.”

PMC is ahead of the pack in its development of 320 Gbit/s, with competitors like Vitesse Semiconductor Corp. (Nasdaq: VTSS) and IBM Corp. (NYSE: IBM) lagging behind with slower chips, says Wheeler. Its closest competitor is ZettaCom Inc., a startup that has also announced it is working on a 320-Gbit/s chipset.

If Mahi is able to deliver a product by the end of the year (which Greg Peters, its CEO, has stated previously as the company’s target) it could be the first system vendor to ship a product with an off-the-shelf switch fabric of this magnitude.

But the first generation of the TT1 may come with baggage attached. Analysts, including Wheeler, who cover the semiconductor industry say that it has an enormous chip count, which means more chips must be deployed in the system to support the large switch fabric. What this means for system companies is that chassis slots must be taken up with line cards to support the enormous fabric, which in turn means fewer ports for line cards with revenue-generating interfaces.

Wheeler says that PMC is expected to announce a more tightly integrated upgrade to the TT1 within the next few months, but these new features wouldn’t likely show up in a product like Mahi’s for at least a year. Despite these integration issues, Wheeler says that the PMC chipset still gives Mahi a leg up on its competitors.

“It puts them ahead in terms of time of market, and that’s important for a startup,” he says. “They [Mahi] can always go back and reduce the cost of the product when the next generation of silicon becomes available.” - Marguerite Reardon, Senior Editor, Light Reading

Page 1 / 2   >   >>
metroshark 12/4/2012 | 7:55:23 PM
re: Mahi Secrets Surface If Mahi is counting on PMC chips to build their box, I wish them good luck - and they will need a lot of it. PMC has OK solutions for Framers and Cross Connects, but as far as routing, MPLS, etc., I don't know what they can offer for a carrier class solution. Also, in terms of the cost of the system, they will be at a severe disadvantage.
optical 12/4/2012 | 7:55:23 PM
re: Mahi Secrets Surface OK, let's just sit back and see if they can deliver on the hype. I guess we haven't learned our lessons from the days when all these godboxes said they could "do it all" and in reality they can't even deliver a working product. I'm from Missouri, so show me first before I get all excited about a box that can do it all!
indianajones 12/4/2012 | 7:55:20 PM
re: Mahi Secrets Surface Not to mention power and size. The chipsets take up so much board size and power that it is not even funny
lightwatcher 12/4/2012 | 7:55:19 PM
re: Mahi Secrets Surface A PMC sales rep told me that Equipe and Lucent's Oz/MSC25K project both use the TT1....maybe that's why those switches are so damn big ?
Pseudopersonality 12/4/2012 | 7:55:19 PM
re: Mahi Secrets Surface What is the scalablitliy with this ASIC based architecture?

What are the differentiators with the "bleeding edge" techology?

ars 12/4/2012 | 7:55:18 PM
re: Mahi Secrets Surface Smart thing to use off-the-shelf technology. "Can do all" is a sure recipe for failure. Sometime vendor has a better chance of succeeding inspite of initial stuggle (What is a small piece for customer, is life and death for the vendor).

BTW, 2000 was the year of "off-the-shelf public relationship". Hopefully, this "off-the-shelf" is post-bubble.

sailor 12/4/2012 | 7:55:17 PM
re: Mahi Secrets Surface I had heard that Mahi lost several of their lead architecture/ASIC people. Maybe the choice of using PMC chips was the only way to product.
myresearch 12/4/2012 | 7:55:01 PM
re: Mahi Secrets Surface If you known something about PMC's switch fabric and the kind of boxes Mahi is trying to build, you know that they dont go together. I am not saying that PCM's or Mahi's products good or bad but they do not suit each other - you would figure this out very easily if you have some technical knowledge in this space:-)

Brattain 12/4/2012 | 7:54:56 PM
re: Mahi Secrets Surface ALL IN ONE GOD BOX?
so what, every vendor in this industry has
considered or attempted to put all services in one box. History has shown this has repeatedly failed for one reason or another.

sow what, every system vendor does this to save time.

Where is the uniqueness or innovation in
Mahi's approach???? I see no secret here.

itworksallthetime 12/4/2012 | 7:54:52 PM
re: Mahi Secrets Surface It is also THE ONLY, working 320 Gb/s switch fabric in its size/performance???????
Page 1 / 2   >   >>
Sign In