Defective R&D Funding Models

For years, the semiconductor equipment industry has been dealing with an R&D funding gap.

Here’s the basic problem: Chipmakers demand certain tools for their next-generation processes, but they are not always willing to foot a large percentage of the R&D bill. And so, the equipment vendors develop the tools and assume a large part of R&D funding–and the risks.

Fair or unfair, this system has worked for many years. This was because there were a multitude of leading-edge chipmakers that bought tools. So, in many cases, the fab tool vendors themselves could manage to obtain a decent return-on-investment.     

Not surprisingly, this business model is falling apart. As before, the tool makers assume a large chunk of the R&D bill. But unlike in the past, there are fewer leading-edge tool buyers at each node. Needless to say, it is more difficult to get any return in a difficult business climate.

Chipmakers realize the old model is crumbling and are beginning to address the problem. In one example, Intel, TSMC and Samsung took a radical step by recently investing millions of dollars in ASML Holding. This is supposed to accelerate ASML’s efforts in extreme ultraviolet (EUV) lithography and 450mm tool development.

These types of arrangements are the exception to the rule, however. “In lithography, there is a huge gap in enabling EUV. But look what ASML is spending per quarter. It’s about 240 million euros per quarter in R&D or something like that. So, they are spending $1 billion a year in litho R&D,” said Michael Lercel, senior director of nanodefectivity and metrology at Sematech.  “I don’t see a billion dollars being spent in metrology and equipment R&D. It’s not supported by the market.”

This is one of the main problems in the fab tool industry today. For example, the process control industry in total may not need $1 billion in R&D per year, but the business will require substantially more funding to enable new breakthroughs in the arena. But the problem is that the process control market is not “big enough to support all of the investments needed to support a disruptive technology,” Lercel said.

So, what’s the solution to develop and fund next-generation process control tools? The same question can be asked about next-generation tools for 450mm and other expensive projects on the books.

In 450mm, for example, the industry has formed a plethora of consortiums to help share the risks and the R&D bill. In some of those consortiums, various governments have provided some funding.   

Still, the current R&D funding model, and the consortiums, don’t quite go far enough. Here’s some possible and future solutions to the problem:

*The current consortiums are too broad and serve too many tool markets. One possible idea is to create tool-specific consortiums based on needs. For example, if the industry really needs a multi-beam e-beam inspection tool, why not create a consortium just around that technology alone? Have Applied Materials, Multibeam, GlobalFoundries, Intel, Samsung and TSMC form and fund this product-specific consortium.

*This approach, of course, has been tried with mixed success. In another approach, perhaps the industry should funnel the funding to one company. Multibeam, for example, is keen on developing multi-beam e-beam inspection. So is Agilent. Why not just fund one or two vendors in the arena and make it transparent?   

These are just some possible solutions. But if something isn’t done soon, the R&D funding gap could widen. And if push comes to shove, the equipment industry may end up playing hardball. For example, one analyst constructed a possible scenario in future 450mm tool development. “The model needs to be completely changed,” said Malcolm Penn, chairman and chief executive of Future Horizons, in a recent interview. “I can see a situation where equipment vendors will want (a chipmaker) to guarantee them a fab line, or $100 million in orders, or something like that. There has got to be down payments, cancellation penalties and real financial constraints and commitments all the way down the supply chain.”