ASML already forced the most advanced chipmakers—Intel, Samsung and TSMC—to put skin in the game by taking equity stakes in ASML and participating in a program to fund EUV and 450-mm wafer research. Now ASML has essentially done the same thing, putting its money where its mouth is to acquire a supplier critical to the development of EUV.
Get a 10% discount on ARM TechCon 2012 conference passes by using promo code EDIT. Click here to learn about the show and register.]
ASML of course already has billions of dollars at stake in the development of EUV. The technology haws proven its viability in terms of resolution, but EUV remains far too slow to enable economical volume semiconductor manufacturing. ASML says it expects EUV to achieve the throughput necessary to make volume manufacturing possible in 2014. But EUV is still no sure bet.
The lack of a light source with adequate power and stability remains the biggest hurdle for EUV. ASML and Cymer have been working closely together for more than a year to improve the source and increase throughput. The companies have improved the technology, but not nearly enough.
ASML said the throughput of its pre-production NXE:3100 is now seven wafers per hour. The company believes it can improve the throughput of EUV to 69 wafers per hour at 105 watts in time to enable volume manufacturing in 2014. Some in the industry remain skeptical.
It's not just the consolidation of a couple of equipment companies. Remember ASML invited Intel, Samsung, and TSMC to become part owners as well. So now we have everything from design through fabrication AND the equipment used to do it all intertwined together. At some point if the cost of making EUV work is so costly then maybe it's just not viable right now. Maybe the single minded pursuit of EUV litho is draining investment out of the semi industry that could better be spent elsewhere.
Just as the foundry business consolidated as the costs skyrocketed, now the equipment business will do the same. And just as everyone fears the growth of a single foundry domination to be bad for the industry, they will say the same about this segment. The costs just make any other option impractical, like it or not.
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.