Johnson said a period of
semiconductor inventory level reduction, which lowered production
levels, appears to be over. But overall market weakness is continuing to
depress utilization levels, Johnson said. Increased demand and lower
yields at advanced nodes is creating equipment shortages for high end
logic production tools, but not enough to bring overall utilization to
desired levels, he said.
"In the memory segment, some suppliers
are even cutting production in an attempt to shore up weak market
fundamentals," Johnson said.
2013 fab equipment forecast differs substantially from a forecast
issued last month by fab tool vendor trade group SEMI. That forecast
calls for spending on front-end fab equipment—excluding test, assembly
and packaging gear—to increase 17 percent in 2013 to reach a new record
high of $42.7 billion.
SEMI's mid-year consensus forecast for
semiconductor equipment sales in 2012 is also more optimistic than
Gartner's revised forecast. The mid-year consensus forecast, issued in
July, predicts that equipment spending will decline about 3 percent in
Gartner predicts that tool spending by memory chip makers
will continue to be weak through 2012. The firm revised its forecast for
foundry capital spending downward for 2012 and 2013 due to early
improvements in 28-nm yield. But Gartner raised its capex estimate for
foundries in future years due to more aggressive development of 450-mm
wafers and extreme ultraviolet lithography.
I think there is another factor at play too. Equipment spending is heavily influenced by migration to newer technology nodes. Since fewer and fewer fabs are migrating to the more advanced nodes, due to complexity and cost, there are fewer technology migrations to drive equipment spending. Even if you argue that the fewer fabs will need to buy more equipment to cover the demand, it will at least induce more fluctuation in the spending as the process migration will be more sporatic with fewer players to even out the averages.
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.