Manufacturing Software: Where Is the Market Heading?

Although this condition has steadily evolved over the past five years, it has become especially apparent in 2006. The irony of this situation is that the semiconductor industry already relies heavily on advanced manufacturing software for much of the productivity improvements that fuel its growth, and this dependence will only become stronger in the coming years. The largest single contribution in the past five years has no doubt been the transition to 300 mm wafers. However, when you break this topic into its component parts, you find that a great deal of the overall benefit was derived not from the increased number of devices per wafer, but from the automation technologies required to achieve acceptable yield and productivity at the 300 mm technology nodes. In fact, some of these technologies were applied in 200 mm fabs as they were being developed, and postponed the need for 300 mm as a result.

This phenomenon will certainly persist. The only real consensus about the timing of the next wafer-size transition and/or major changes in production lithography technology is that “they will take awhile.” Mature yields for high-volume devices are at acceptable levels, even for 300 mm fabs. For these reasons, the targeted rate of annual productivity gain must be achieved by contributions from other sources in the manufacturing process hierarchy. Increasingly, companies are also looking at other stages in the overall product lifecycle to shorten the “time to money,” with the ultimate goal of first-pass success in product and process design. These sources include improvements in device design, unit process design, and process integration, and explain the growing interest the EDA companies have placed in downstream manufacturing systems.

As diverse as these techniques are, they share a common characteristic: the prevalence of software. But how can an environment so rich with challenge and opportunity be such a difficult place to grow an independent software business? This question is further broken down into two parts: the demand side and the supply side.

The issues on the demand side are fairly simple. As software technology has become a potent source of competitive advantage, semiconductor manufacturers have kept a greater portion of their mission-critical software development in-house to better control the content, schedule and cost of this work, and to protect the resultant intellectual property. The industry’s cyclical nature exacerbates this tendency — during the downcycles, companies naturally protect their internal resources and continuity while cutting outside spending, which causes the gap to widen between their specific needs and the suppliers’ ability to meet them. The issues on the supply side are more complex. In any highly competitive industry, suppliers seek to grow the value they provide their customers while presenting a consistent image and maintaining sufficient economies of scope and scale to make money. As semiconductor manufacturers spend a smaller percentage of their total software investment on commercial systems, suppliers across the industry spectrum are looking for ways to capture synergistic portions of this business.

Two major acquisitions announced before this issue went to press illustrate these points. The first of these, the purchase of SiAutomation (principally a fault detection software supplier) by PDF Solutions (yield management systems) is an example of a company extending its scope into an adjacent market segment and creating synergy in both segments. The combination of these companies addresses two distinct gaps in the process design loop, as it connects a source of detailed equipment data and the analysis tools to characterize and understand actual process behavior with a detailed unit process model that can be used in product and process design tasks. This is a classic case of horizontal integration. The acquisition of Brooks Software by Applied Materials is more a case of vertical integration, as it broadens Applied’s scope of hardware and software products within the fab. It also gives Applied a more direct source of advanced software technology, for which they are a major consumer. However, from a marketing perspective, it is most significant in that a once-independent source of mission-critical software is now aligned with the industry’s largest process equipment supplier, which has ramifications throughout the food chain.

The semiconductor manufacturing software market is a moving target in a dynamic industry, so a snapshot at any point in time is going to be a little blurry. Nevertheless, some of the features of a new structure are taking shape, and I hope this perspective helps the readers analyze and understand this sometimes puzzling picture.