LONDON – Competition within the ARM ecosystem aimed at mobile devices is driving a faster pace of process development and will likely prompt foundry production on the 22-nm/20-nm process node in 2013, according to Eric Meurice, CEO of lithography vendor ASML Holding NV (Veldhoven, The Netherlands).
Meurice, speaking to analysts on a conference call to discuss ASML's financial results for the first quarter of 2012, said the industry is competing in the mobile devices space and that is driving the development of a new production node each year.
Meurice said that ASML expects to have six to nine months of strong sales in support of ramping production of the 28-nm production node but that by year end there will be the start of demand for equipment for 22-nm production.
The leading chip company is already in production on 22-nm with its own FinFET-based process but leading foundry Taiwan Semiconductor Manufacturing Co. Ltd. is struggling to meet demand for its 28-nm production, according to recent reports.
"We have another six to nine months of production and shipment of machines before we have delivered to the market a mature, not high, level of [28-nm] wafers per month. After that there is another engine for 2013, which is to be proven, but we are getting now significant interest. The customers want to go from 28-nm to 22-nm already."
Meurice told analysts that competition in the ARM environment around mobile applications is driving the desire for improved products and architectures. Rather than migrating to a new node every two years or 18 months Meurice sees the faster ramping to a new node every year.
In a presentation prepared to accompany the companies financial results ASML said: "Demand for 28-nm in logic continues to be large and structural and will be accompanied by the start of pre-production for 20-nm late in 2012. For memory we expect technology demand for sub 20-nm NAND, sub 30-nm DRAM later this year."
Meurice said that successful production at 22-nm or 20-nm in 2013 was not automatic. "These are very complicated nodes technology-wise and you could have difficulties; yield, ramp, etc. But this would probably not impact the litho business." In fact, Meurice concluded, such difficulties could provide a business upside for ASML.
A business upside for ASML to be sure! Wouldn't Mr. Meurice rather sell 2 monopolistically profitable immersion tools, or unprofitable, barely post-R&D EUV/X-ray tools, running at single digit wafers per hour?
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.