I noted on Seeking Alpha on December 28, 2009 that although there is a lot of hype regarding semiconductor equipment growth in 2010, my analysis indicates that the market will grow only 20%, not the 50% figures being presented.
Clearly 2010 will be a year when semiconductor manufacturers purchase equipment that will move them to the next technology node, namely 40 nanometers (nm). Capacity purchases are out of the question, particularly when no new fabs are expected to be built, as in 2009.
A key factor in technology purchases is that semiconductor production ramps up quickly with yields above 80%, otherwise the $35 million spent on a single immersion lithography tool from companies such as ASML or Nikon will not meet cost of ownership mandates.
Back in late April, 2009 during a conference call, an analyst asked Rick Tsai, president and chief executive of TSMC (TSM), about yield problems with the company's 40-nm process. He said: ''There have been difficulties with the yields. 40-nm is a difficult technology to manufacturer. We understand the root of the problem.” Two months later, chairman Morris Chang assumed the role of CEO from Tsai.
Fast forward to January 2010. According to a new article in Taiwan-based DigiTimes, foundry chipmakers, including TSMC, have been struggling to increase their yields on 40nm to over 70%. Clients using TSMC's 40nm node include graphical processing unit (GPU) vendors AMD and Nvidia (NVDA), and FPGA chip supplier Altera (ALTR). In 9 months TSMC is still having a problem.
The DigiTimes article also noted that UMC is also working with customers to volume produce 40nm products and that FPGA chipmaker Xilinx (XLNX), which has placed orders for 40nm-generation products with its long-term foundry partner, has moved to diversify its production sources to minimize the risk of shipment delays.
TSMC and UMC are the leading foundries in the semiconductor industry and are leaders among manufacturers of leading-edge semiconductors. If yields are low ad 40nm, one must assume that yields are low throughout the industry. If equipment is not meeting production requirements, shortages in the semiconductors being manufactured will result.
Chip makers are naturally working to solve the yield problem, which is usually a combination of production and design issues. They have the option of purchasing additional equipment to overcome the bottleneck, which is what I suspect happened in the ramp in equipment purchases in Q4 2009, particularly by TSMC as part of its $4.6 billion capex in 2009. But with surplus 40nm equipment in place, I expect we’ll see a leveling off of capex for a few quarters going forward until semiconductor demand catches up with production capacity.