To compete in the fast-moving semiconductor industry, companies have to be ready to make billion-dollar decisions in a Silicon Valley second. It's a business ruled by a basic principle that says new technology will soon be sent to the dustbin of history by newer technology.
For chipmakers, that means that a major investment in a fabrication plant may pay off for only a few years before it has to be retooled or replaced to produce a new generation of chips. That has led many companies to move to a "fabless" business model, where the production of chips is outsourced to contract manufacturers, often an ocean away.
It's a decision, however, that carries significant risk management challenges, such as balancing savings on workers' compensation and property coverage costs against increased worries about business continuity and supply-chain management.
"All of sudden, some of the issues that you have--a huge property exposure, workers' comp issues--a lot of that goes away," says Thomas Whitenight, vice president of technology services for broking firm ABD Insurance and Financial Services, who has worked with clients moving to a fabless model. "You outsource all of that. It takes away a lot of the headaches."
Those advantages have to be weighed against new concerns such as the impact of a major typhoon or earthquake on Asian-based contract manufacturing plants, called foundries; disruptions to a supply chain that stretches for thousands of miles; and the loss of direct control over manufacturing facilities.
"As you deal with fabless semiconductors, the real exposure shifts to that contingent side, because they're using foundries for the most part to provide their chips. So it's not their own exposures they're worried about, it's the manufacturing at other locations," says Norm Jacobi, a managing director at Marsh Inc. based in San Jose, Calif.
In addition to soul-searching about their business models, semiconductor companies are facing new risks as their products expand into new areas, such as medical uses, aviation and home entertainment.
To deal with their new and evolving exposures, semiconductor risk managers say the best strategy is to take a proactive, holistic approach to risk management by evaluating all their risks, insurable or not, and developing ways to mitigate them.
FROM FAB-LITE TO FABLESS
Faced with the need to move to the latest cutting-edge, nanometer-scale technology, custom chipmaker LSI Logic Corp. said in September it would put its remaining fabrication plant in Oregon up for sale and move to a fabless business model. The veteran Silicon Valley company says the change offers not only the ability to adopt new technology more quickly, but also compelling savings on insurance costs.
Karen Miller, longtime risk manager for the Milpitas, Calif.-based company, says the move will bring savings in several areas as LSI sells the plant and trims 500 workers from its payroll.
"That impacts our workers' comp and reduces costs. We also have a self-insured program in that space, as well, so we would dismantle that," Miller says.
"From a property standpoint, it reduces our cost of property insurance because we no longer have that asset to insure," Miller says, "but depending on how that manufacturing is reallocated to our contract fabs, that really drives our contingent business-interruption exposure and costs."
The move has been the latest step from a "fab-lite" to a fabless model for LSI, which already sold a manufacturing plant in Japan and earlier put its Colorado plant up for sale.
Other companies using a fabless business model include communications chipmakers Qualcomm Corp. and Broadcom Corp. On the other side, chipmakers running their own fabs include Intel Corp., the largest maker of computer chips; its rival Advanced Micro Devices Inc.; and analog chipmaker National Semiconductor Corp.
For LSI, the decision hinged on how best to move to newer chipmaking technology at a scale of 65 nanometers and smaller--roughly a thousand times finer than a strand of human hair and half the scale of current 0.13-micron technology.
It's a common challenge in a business governed by Moore's Law, promulgated four decades ago by Intel co-founder Gordon Moore, which says that the number of transistors that can be put on a chip doubles about every two years. That has meant a relentless march to ever-teenier production scales.
At the same time, manufacturers are seeking to use larger silicon wafers, from which the chips are cut, to improve yields. The latest wafers, about 12 inches in diameter, yield more than twice as many chips as the previous 8-inch wafers. Moves to both the smaller-scale chipmaking technology and the larger silicon wafers require upgraded or new fabs.
The costs of building a new fab or retooling an old one are huge.
Intel, for instance, announced in July that it would spend $3 billion to build a new fab in Arizona, and AMD is investing $2.5 billion in a new fab in Germany. IBM had to raise the roof of its fab in East Fishkill, N.Y., by more than 4 feet to accommodate new equipment as part of its $2.5 billion retooling of the plant several years ago.
That kind of price tag can make a decision whether to build a new fab a make-or-break choice. So many chipmakers are turning to partnerships with contract manufacturers to make their latest chips. LSI, for instance, has partners that include Taiwan Semiconductor Manufacturing Co., the largest chip foundry.
Miller notes that the move to a fabless business model is a well-trod path in the semiconductor industry, and the risks have been well-studied.
"I anticipate that to be an easy transition in that our underwriters are very familiar with that scenario," Miller says. Overall, the move sharpens the company's risk management focus on business continuity, she adds. "It focuses (us) more on contingent business interruption, business continuity and supply-chain management," Miller says.
FABLESS, NOT RISKLESS
While the move to a fabless model reduces the costs associated with owning a manufacturing plant, some of the risks remain.
"You don't have some of those primary property issues necessarily, (but) you would have business interruption," says Brenda Shelly, executive vice president at Willis Group, based in San Francisco. "You still have the same risks with a fab being down, it's just not your fab."
One of the benefits of owning your fabs is the ability to better manage business continuity and loss prevention and mitigation, says Eugene Kiernan, risk manager for National Semiconductor Corp., which makes analog power-control and amplifier chips, as well as chips for electronic equipment, that translate digital signals into real-world phenomena such as light and sound. The Silicon Valley-based company has manufacturing plants in the United States, Scotland and Asia.
"Personally, I think if you go fabless, for risk management it creates more of a risk because you have less control. You have less information," Kiernan says.
"You have to work at identifying and maintaining and keeping abreast of the risk because it's being done somewhere else. Whereas if you own your own manufacturing, you have much better control over your costs and products," Kiernan says.
Going with a fabless model also heightens supply-chain concerns, Kiernan says.
"That takes a part of your supply chain and puts it in control of somebody else." Kiernan says. "You might not know what's going on on a day-to-day basis, or week-to-week basis, and then you have to understand how they're handling risks at their factory."
The business-continuity exposures can be critical in an industry where a missed launch date or the inability to supply product at the right time can easily spell the difference between success and failure.
"If that factory has a catastrophic event, where can they provide alternative manufacturing facilities? If they can't, then you're out of business because they're out of business," Kiernan says.
That risk is one that insurance may not fully address.
"If you lose a critical fab, it doesn't matter if you're going to have an insurance payment," says Marsh's Jacobi. "The key is are you going to be able to be in business tomorrow. So really relying on disaster planning, business continuity plans, is the key."
That means that when deciding whether to move to a fabless model, companies need to take a very hard look at their supply chains.
"From a risk management standpoint, as they move to that fabless model, we spend a lot of time looking at the supply chain as it moves through from idea to processing to packaging, and then the cargo moving around the world," says ABD's Whitenight, who is based in Redwood City, Calif.
Another area of concern for companies hiring contract manufacturers is protection of their intellectual property in a business where billions of dollars in sales may hinge on a new chip design.
"If they're partnering with the wrong people, piracy can become a big deal," Jacobi says.
Intellectual property losses, however, may not be best addressed through insurance.
"A lot of times when there's intellectual property and litigation, the remedy and the damages is licensing," says Shelly at Willis. "So that's not really going to be insurable. It's something that really speaks to reserving and self-funding or something that's managed best in the general counsel's office."
At LSI, Miller says the company has a unit devoted to intellectual property issues.
"A lot of what we do at the company is risk mitigation," Miller says. "We have people whose jobs, for example, are in the intellectual property area. They do the filings. They do the research on the patents. They do defense on patents and enforcements. We have a whole group that's set up to do that so we don't need the insurance."
NEW PRODUCTS, NEW RISKS
An emerging risk for semiconductor-makers is the expansion of computer chips into new product areas, such as in aviation, medical devices and home entertainment, which brings with it the potential for expanding liabilities.
"The big deal right now is products liability," Jacobi says. "As companies move into these adjacent spaces like aviation or medical or telecommunications or wireless, they start to take on the liabilities of their end users.
"Chip companies are designing chips that can go into aviation products, (and) it may be necessary to buy a specific aviation product liability policy," Jacobi adds.
"If they're going into medical products that are invasive, they may have to buy medical product liability," he says.
At LSI, Miller says the company has long been active in the aviation market.
"We've always had some presence in the aviation products space. So we purchase a separate products policy for that," Miller says.
"We really haven't made an entree into medical, but if we were to, we'd have to look at a specialized program for that as well," Miller says. "We're already in the consumer space, and that doesn't create a problem for us in regards to standard purchasing."
As technology converges into more areas of life, however, the insurance industry is watching closely.
"Insurance underwriters are asking many more diverse questions regarding both individual risks from both a products standpoint and premises liability," says National Semiconductor's Kiernan. "And renewals are requiring more and more detail than ever before to support the types of questions they're asking."
The key to managing insurance costs in this environment, Kiernan says, is to develop a deep understanding of the company's products and processes, and be able to explain that to underwriters so as to obtain the broadest possible coverage without critical exclusions.
"Our risk management strategy has been to better understand our own products and markets--now and in development--so we can not only respond to the underwriters' concerns, but we can more clearly address our own concerns," Kiernan says. "At the end of the day, the risk is with the company."
writes frequently on technology issues for
Risk & Insurance®.
March 1, 2006
Copyright 2006© LRP Publications