Canon struggles in litho as tool prices soar

Litho prices off the chart
In other words, Canon kept its rivals honest. Without Canon as a factor, the market is somewhat less competitive. ''As the competition diminishes, the prices (for lithography tools) have increased,'' Hutcheson said. Some believe Canon had little or nothing to do with soaring lithography costs. There is a direct correlation between tool costs, as chip makers marched towards the advanced nodes.

In any case, analysts have sounded the alarm bells for good reason: Tool costs are out of control. When the IC industry broke the one-micron barrier in the mid-1980s, a stepper sold for about $1 million. In comparison, the average selling price (ASP) of an immersion system from ASML is $45 million now, compared to $42 million in 2008, according to The Information Network.

Hutcheson said lithography tool prices are doubling every two nodes. So, for the 22-nm node, a top-of-the-line, 193-nm immersion scanner could go for as high as $80 million or more per machine, he said.

From another perspective, Rinnen said that lithography prices are increasing ''50-to-70 percent for every node.'' Because tool complexity is so high, vendors ''demand to be paid fair value,'' he added.

What that implies are several ominous data points. First, fewer chip makers can afford to build leading-edge fabs. In chip production, lithography is the most expensive tool in the fab. And so, in today's climate, a dwindling number of chip makers can afford to procure state-of-the-art scanners. And without Canon as a factor at the leading-edge, there is little or no chance that lithography prices will slow down.

Intel, Samsung, TSMC and Toshiba can afford future leading-edge tools. Beyond that list, it's really unclear which vendor can afford to buy $80 million scanners. To be sure, it's a short list.

Besides soaring tool costs, there has been a sea of change in lithography over the years. Perhaps the most obvious change is the reduction of the vendor base. Back in the 1980s, for example, the market included the following lithography suppliers: ASET, ASML, Canon, GCA, Eaton, Nikon, Perkin-Elmer, SVG and Ultratech. It also included two X-ray hopefuls in Hampshire and Micronix.

Then, at that time, there was a massive shakeout. Industry pioneer GCA of the United States stumbled and is no longer around. U.S.-based ASET was a minor factor and disappeared. U.S.-based Eaton had little understanding of the business and failed.

Several years ago, U.S.-based SVG more or less picked up Perkin-Elmer's business. SVG, once the main lithography vendor to Intel Corp., was sold to ASML. U.S.-based Ultratech, which championed 1x lithography, eventually exited the business.

And the X-ray players disappeared as well. Sadly, by the start of this decade, the U.S. had no presence in optical lithography. The business was dominated by ASML and two Japanese vendors in Canon and Nikon.

Simply put, there was a clear failure and lack of leadership among U.S. chip makers--and perhaps the government--to keep optical lithography in the United States. Today, the landscape is somewhat brighter. The U.S. boasts a collection of nano-imprint lithography vendors (Nanonex Inc., Molecular Imprints Inc.) and an electron-beam supplier (Vistec).

The real problem, however, is that an inordinate amount of R&D funding has been funneled into EUV--without any payback to speak of. The other emerging technologies--nano-imprint and maskless--are grossly underfunded. Companies have rolled out nano-imprint and maskless tools, but not for mass IC production--yet.