Effective, quick transitions key to survival and recovery - Special Section: Wafer Processing - semiconductor industry forecasts

Electronic News, July 22, 2002 by John Kibarian

Today, almost two years into the semiconductor industry downturn, it's natural that chipmakers are looking for evidence of growth wherever possible.

Certainly there are reasons for optimism in some sectors of the industry. However, a full recovery across the entire industry has not taken hold yet.

When the downturn started in early 2001, most chipmakers expected this down cycle would look like a V-shaped curve, with a steep downward lurch and then an equally fast recovery.

A few quarters later, however, the consensus moved toward a conception of the downturn as a U-shaped curve. That prediction proved closer to the truth, as the bottom of the cycle eventually ran through 2001 and into the early months of this year.

Looking ahead to the next six to 12 months, the question is not how much longer will things decline, but when will the industry finally see strong, broad growth?

While the recovery is still not visible, it shouldn't be a surprise--in previous downturns few industry observers were able to predict the timing of the upturn until after it was underway.

Today, some are questioning whether recent improvements in capacity utilization are the result of a real increase in end user demand or simply inventory correction. It will take several more quarters to answer this question, as only real demand will result in sustained and expanding growth.

From the perspective of the process-design integration sector partnering with leading edge design and manufacturing companies, most of the volume and growth lately has been in ICs for consumer applications. For example, in the communications IC business, chips designed for cell phones are recovering faster than the devices that support cell phones but are sold in systems for cell phone system operators.

What will the recovery look like? Historically, downturns have been a time when chipmakers invest in new technology. There will likely be some structural changes in the industry as it recovers. In this cycle there has been an acceleration in outsourcing of manufacturing, for example, and chipmakers increasingly working together to minimize R&D expenses and share risks.

Furthermore, in sub-0.18-micron process technologies, each new generation requires new materials, which pose a challenge.

Production costs using the latest technology nodes are high and the technology is more difficult, so risk increases as well.

Given the state of the industry, companies are moving aggressively to minimize risk and are only moving to newer technologies such as 0.13 micron for the devices that absolutely demand the new features offered. This is a rational reaction to the technology risks at this technology node.

In fact, simulations of the build cost, or production costs, for chipmakers on 0.13-micron products have found that they remain high.

Comparing this modeling to the comparative simulation models for 0.18-micron technology at the same point in its life cycle demonstrates that the 0.13-micron transition has been more difficult for the industry.

This creates a bit of a catch-22.

In general, chipmakers remain hesitant to move to new process technologies. However, using traditional technology ramp methods, factories need consistent production of the same chips month-over-month to work out production issues. Unless they are manufacturing high-end products such as large FPGAs with average selling prices in the $500-plus range or high-end microprocessors, few product lines are willing to commit to volume production on wafers that cost $3,700 each.

In addition, building high-reliability products at a reasonable cost at the 0.13-micron node remains a difficult undertaking and is one of the main obstacles slowing the recovery.

Since few chips have sufficient demand and cost more than $500--and 90nm technologies will cost even more--a core skill that IC companies must develop in the recovery is the ability to ramp new technologies with products that have huge volumes and lower price-points.

As it stands today, only the chipmakers that have a large target market and expect to see big volumes are willing to invest in next-generation technologies.

Unlike previous technology nodes, investments are taking place to develop consumer products, such as games and DVD players, rather than high-end workstations and industrial applications. To win in these markets, chipmakers will need to be effective at transitioning to newer technologies.

John Kibarian is president and CEO of yield management software supplier PDF Solutions.