LONDON – The world's leading foundries are facing difficult times due to the creation of too much manufacturing capacity, according to Paul Otellini, president and CEO of Intel Corp., the world's leading chip maker.
"I think the foundry business is in big trouble over the next few years," Otellini told analysts at a conference here Thursday (Feb. 17) and accessed by EE Times by a telephone replay. He put it down to "significant overcapacity" mainly due to attempts to grab market share by new player Globalfoundries Inc. (Sunnyvale, Calif.). On Friday in Hillsboro, Oregon, Otellini said Intel would build a wafer fab in Arizona and create 4,000 U.S. jobs in 2011.
However, talk of overcapacity does not chime with the view of some analysts, such as Malcolm Penn, founder of Future Horizons (Sevenoaks, England). Penn has heralded the coming of a fab-tight era, due to a lack of spending on manufacturing capacity by both IDMs and foundries over the last couple of years.
Some have said that the chip world is rapidly collapsing to a handful of leading-edge chip makers most of whom will have some foundry interest. Intel and Samsung still make the majority of chips for sale under their own brand while Taiwan Semiconductor Manufacturing Co. Ltd. (Hsinchu, Taiwan) and Globalfoundries are pure-play foundries.
Nonetheless Otellini does not seem eager to be joining the foundry crowd any time soon. "Foundries make money not on the leading edge but on the trailing edge, with long running products," he said.
"Leading-edge capacity excess, driven by Globalfoundries' expansion primarily, means pricing is going to come down. As leading-edge pricing comes down, so does the trailing-edge [pricing], hollowing out the business. That means real trouble for Globalfoundries and TSMC over the next few years," Otellini predicted. He conceded that, with its memory and logic operations alongside foundry, Samsung was a different case.
Otellini said chip building was all about the efficiency of the development and manufacturing operations and that Intel's was second to none.
"Our competitors have to take what TSMC and Globalfoundries offer them," indicating that Intel liked having control of its manufacturing choices. But Otellini also glided over the collorary of his prediction of price erosion at the foundries; that fabless chip companies will be getting good deals on both leading- and trailing-edge silicon making those companies, the likes of Qualcomm, Broadcom and Nvidia, better able to compete with Intel.
Nonetheless, Otellini's conclusion was that Intel and Samsung would increasingly pull away from the manufacturing pack, particularly Globalfoundries and TSMC.
I wonder how much the production interruptions in Japan could play into the "over-supply" of Fabs? It seems wise to have diverse Fab locations and while very expensive to build/run, owning your Fab does give you control over your destiny (for good or for bad).
if there was oversupply why fablite/fabless companies suffered during '10-Q3&Q4 coz' of delayed shipments/outcomes where big players had the bigger pie always on time? Where was this overcapacity and why did foundries let their small size customers down?
Another Otellini speech where he is trying to mirror focus on other direction. I truelly agree with u @Rick, where are intel's serious moves against ARM? soon after it will be too late
Intel is a large and profitable chip producer. I am sure they are keenly aware that Windows and x86 are running out of steam. And they're expanding capacity while at the same time pointing out an overcapacity in the industry? So what gives?
There is more than one metric that needs to be addressed to be successful in the evolving mobile market. Of course power consumption is of prime importance. Also of great importance is cost.
Intel could manufacture an Arm license but that incurs royalties. Intel in the past, with some notable failures, has attempted to architect to meet specific applications. The 432 comes to mind. However, Android type environments could benefit from a specific architecture. There is, in my opinion, the likelihood of increasing power/workload efficiency while reducing cost.
To do this Intel needs to step back and approach the silicon/software solution much more holistically. It would mean substantially increasing the software side of the equation.
mranderson, How will GF gain? Who's using GF's 40 and 28nm today?? Have not heard anyone claiming they are using GF technology today, have you? If you know, please copy the links here.
TSMC is the only foundry who're having these issues. Others even don't know what the issues are!
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.