SAN JOSE, Calif.—Morris Chang, chairman and CEO of Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC), acknowledged Tuesday (April 17) that the foundry giant has experienced issues at the 28-nm node, but said the problems were related to capacity, not yield.
Chang told an audience at TSMC's annual technology symposium that TSMC's 28-nm yields have from the beginning been in line with the firm's projections.
"Yes, we had some difficulties with 28-nm, but those difficulties had to do with not having enough capacity, not yields," Chang said.
Chang acknowledged that TSMC has been dogged by persistent rumors about its 28-nm yields. In January, Mike Bryant, a technology analyst with Future Horizons Ltd. said TSMC was in trouble with its 28-nm technology, saying the processes were not yielding well. Last month, unsubstantiated rumors circulated that TSMC had shut down its 28-nm production in February. At the time, a TSMC spokeswoman declined to comment, citing company policy against commenting on rumors, but added that the firm's 28-nm process was "normal."
Chang said TSMC's investments in capacity expansion for 28-nm production have helped the firm alleviate the 28-nm capacity crunch. He noted that TSMC planned to invest more than $7 billion in capital expenditures in 2012, up from about $7 billion in 2011, $5 billion in 2010 and $2 billion in 2009.
"With capacity coming online, I do believe the worst is behind us," Chang said.
Taiwanese IT publication Digitimes said in a report Tuesdaythat two stalwart TSMC customers, Qualcomm Inc. and Nvidia Corp., approached other foundries about producing 28-nm chips for them because of tight 28-nm capacity at TSMC. The report cited unnamed sources at semiconductor tool makers.
I would agree with Mr. Nenni 100% on that. I checked with several sources and found that it was highly unlikely that TSMC shut down 28-nm production. I reported this in the EE Times Forum, also in March:
The point of mentioning this rumor in the story was to support the statement that TSMC has been dogged by rumors about it's 28-nm process.
Mr. Nenni of SemiWiki has been saying since March that the 28nm rumors about TSMC shutting down production were false, http://www.semiwiki.com/forum/content/1091-tsmc-absolutely-did-not-halt-28nm-production.html
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.