|
||||||||||||||||||||||||||||||||||||
|
|
||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||
|
|
Good and Evil in The Garden of Outsourcing: A Real-World Tale of Intrigue (and Woe)
by Alex Mendelsohn, ChipCenter Senior Technical Editor
Even as the shadow of a sluggish market grows longer, and layoffs dot the tech sector's economic landscape, American electronics OEMs continue to outsource, turning over phases of product development to third-party operations. Some folks refer to the process as strategic collaboration, but I find that a somewhat less-than-inspiring euphemism, at best.
Whatever you call it, we're immersed in an age of third-party integration, especially with respect to contract manufacturing. Obviously, the goal is to cut near-term costs. After all, if you can design a product, and hand it off to someone else to make it, you don't need an assembly line, nor do you need skilled, or even semi-skilled, assembly workers. Your cost structure becomes comprehensible and fixed as far as your accountants are concerned, and your engineering team can concentrate on what it knows best, namely design work, not manufacturing.
But, can product development companies manage contractors as effectively as in-house teams? Just how do you time the delivery of critical resources? How do you measure a supplier's performance? How do you handle ECOs (engineering change orders)? What about alternate sources?
The much-vaunted philosophy of concurrent design can get trashed in an outsourced environment. Moreover, who really owns your intellectual property (IP) when you hand it over lock, stock, and maybe barrel too, to a third party?
A Tale Of Woe
The sparks started to fly back in April, when ZF Micro Solutions was advised that National wouldn't be producing ZF Micro's ZFx86 FailSafe system-on-a-chip (SoC). After inking an accord to complete some integration engineering and fab the part, National Semiconductor suddenly turned tail and refused to build the ZF-contracted SoC for ZF Micro Solutions.
"That decision dropped out of the blue," says David Feldman, ZF Micro's chairman and CEO. Feldman told me that National decided independently, and out of contract, to produce and sell the jointly designed chip, even though ZF Micro had paid National to do additional NRE (non-recurring engineering)on ZF Micro's design.
Numbers Don't Lie
As you might imagine, ZF Micro was thrown into what Feldman characterizes as "urgent dialogue with National Semiconductor." In the discourse, ZF Micro Solutions contended that personnel at National were telling lies to National's upper management, ostensibly to cover their tail feathers. ZF Micro argued that it looked like National's lower echelons failed to perform and therefore the company couldn't meet the terms of the design and production contract.
But wait, there's more. ZF Micro Solutions contacted its customers to advise them that it was taking steps to resolve the unanticipated problem with its recalcitrant outsourcing partner. ZF Micro released a statement noting that National indicated it would accept a so-called "last-time buy" for the ZFx86 FailSafe chip (along with insistence on a cash advance).
ZF Micro decided it would help its own customers participate in the last-time buy program National was offering, but ZF's customers would, in turn, have to pay cash in advance for their orders so that ZF Micro Solutions could try and place orders with National Semiconductor. A 12-month forecast for the number of chips an OEM might need was requested.
At this point you may very well ask why downstream designers wouldn't have sought alternate sources? Wasn't there someplace else a needy OEM could get the devices from?
Apparently that wasn't a factor in the original deal. "National had a binding agreement with us for high-volume production of the ZFx86 FailSafe SoC," tells Feldman. "National is a big and well respected company. We didn't feel the need for an alternate source. We had placed orders based on representations by National that it didn't need a second-source for the ZFx86 SoC."
Feldman kept the faith, assuming National would be committed to be his foundry. Feldman says that Brian Halla, the CEO of National Semiconductor, promised him that the company would be "totally committed" to ZF Micro's volume production runs.
Consolation Concessions
In the heat of the quarrel, ZF Micro petitioned the court to permit a distributor, namely Pioneer-Standard, to release the product---whether under ZF Micro's or National's part number. At the same time, ZF Micro attempted to secure from National the foundry information ZF needed to develop an alternate source.
Things got a bit brighter in June when National advised ZF Micro that it had ramped up production for the ZFx86 on National's fabrication lines in Portland, Maine. National Semiconductor agreed to deliver chips.
Then, during the summer, ZF Micro Solutions received an e-mail (what a way to communicate!) from National Semiconductor. It informed ZF Micro Solutions that Pioneer-Standard and National Semiconductor had reached an agreement on how ZF Micro's customer requirements could be satisfied.
Unfortunately, ZF Micro had no control over the price National would charge Pioneer-Standard, nor would ZF Micro have any control over the price Pioneer-Standard would charge to already committed downstream OEMs. Issues of allocation arose as well.
In response, more than 100 OEMs formed a bloc called the ZFx86 Chip Consortium (ZCC). ZCC was desperately trying to ensure that member companies that had spent time and money designing-in the ZFx86 could get an alternate source. Some OEMs in the ZCC were literally staying alive by living off inventories.
"OEMs had invested millions to design-in our ZFx86," tells Feldman. "One company in the UK spent four million British Pounds---more than $6,250,000---on a ZFx86 design-in. At the same time National told us it would bill for one more fab run of 50,000 chips! We knew of one customer alone who needed 100,000."
As soon as ZF Micro's customers started complaining to National, the picture seemed to get brighter. National started selling the ICs to OEMs through Pioneer. But, to complicate matters, Feldman says Pioneer wound up feeding ZF Micro's confidential customer list to National.
Did ZF Micro have an IP protection plan, one that could have (or should have) been implemented with its dealings with National? Or, did ZF Micro go blindly ahead with its outsourcing agreement with National, hoping for the best, assuming that a big IC house wouldn't mess up?
The Web: Boon or Boondoggle?
To counter the possibility of illicit e-interception and communications failures in this day and age of e-commerce, the Electronic Data Interchange-Internet Integration (EDI-INT) and Applicability Statement 2 (AS2) protocols were crafted. Rolled out about five years ago by the Internet Engineering Task Force, EDI-INT and AS2 were designed to ensure secure communications over the Internet.
EDI-INT simply uses the Internet as a transport and delivery medium, defining the methods for sending EDI messages. EDI-INT's intent is to sanction and enable engineering and business transactions through SMTP e-mail, replete with encryption and security.
For Your Eyes Only?
Once the deals are struck and the dotted lines are inked, who is responsible if things go awry, as they apparently did in the case of National Semiconductor and ZF Micro Solutions? If you think about how critical time-to-market is in this era of truncated design-and-development cycles, no one needs a lawsuit, least of all the OEMs who are relying on delivery of components. Time-to-market is time-to-money.
Feldman's List Of Lessons Learned
"Ask for references. Ask to talk to companies that entered into similar agreements." Check to see if the company has been involved in legal disputes relating to the type of relationship you're considering.
"Make sure the contract provides unequivocal details as to IP ownership, not only who owns what going in, but who owns what when the project is finished," he adds. "Consider how you will treat jointly developed IP and what rights each party will have regarding such jointly developed IP.
"Very clearly list all IP that is yours going into the project," Feldman continues. "List all previously developed IP that you're basing the new project on, and specify whether it is to remain proprietary.
"Make sure your contract contains clauses that will guarantee both the delivery and return of all your IP, and have exit points built into the contract that guarantee delivery of all work in progress.
"Be sure to schedule frequent delivery of backups of IP and work-in-progress, so that as progress is made on the project you always have materials that are as close to up-to-date as possible.
"Once the relationship is underway, monitor progress carefully. Have milestones built-in where you can exercise the option to get out because of delays, or where you can impose penalties for non-performance if you feel there are unnecessary delays---whether intentional or unintentional."
Feldman concludes with some sage advice from Intel founder Andy Grove: "Only the paranoid survive."
|
|||||||||||||||||||||||||||||||||||
|
Copyright © 2003 ChipCenter-QuestLink About ChipCenter-Questlink |
||||||||||||||||||||||||||||||||||||