datasheets.com EBN.com EDN.com EETimes.com Embedded.com PlanetAnalog.com TechOnline.com  
Events
UBM Tech
UBM Tech

News & Analysis

Comment


FH1

7/13/2010 9:09 PM EDT

"Freeman also expressed concern about the rapid increase in capacity by ...

More...



pixies

7/13/2010 12:27 PM EDT

Last month I went to scavenge some used equipment at the defunct Qimonda fab in ...

More...

Analyst: Foundries gird for war at 28-nm

Dylan McGrath

7/13/2010 12:41 AM EDT

SAN FRANCISCO—It's a great time to be a fabless company with leading-edge designs, according to Dean Freeman, a research vice president at Gartner Inc. That's because, as Freeman sees it, the top foundry players are moving to ratchet up capacity at 40- and 28-nm, setting the stage for a market share battle and a corresponding drop in wafer pricing.

According to Freeman, the semiconductor industry has never before had more than one foundry that could offer process technology near the leading edge. But today, there are three—Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC), Globalfoundries Inc. and Samsung Electronics Co. Ltd.—Freeman said.

"I think you are going to see leading edge prices drop off very quickly as the competition moves into that area," Freeman said Monday (July 12) at the annual SEMI/Gartner Market Symposium here. "There is a huge amount of capacity coming on line at the 40- and 28-nanometer nodes."

 

By the fourth quarter of next year, TSMC, Globalfoundries and Samsung combined will have the capacity for nearly 280,000 300-mm wafer starts per month at 45-nm and below, up from about 70,000 in the fourth quarter of 2009, according to Gartner estimates. TSMC and Globalfoundries plan to spend a combined $7.9 billion on capital expenditures this year, according to Gartner. 

 

"Foundries are spending now to add capacity at 40-nm and spending to meet the 28-nm ramp," Freeman said. "You have a significant number of companies all introducing technology at the same time, which is going to lead to a significant market share battle."

 

Bill McClean, an analyst with market research firm IC Insights Inc., said in 2008 that foundries were pulling capital spending back to more realistic levels and focusing on increasing revenue per wafer. Reached Monday for comment on Freeman's prediction, McClean said the dynamic had changed with the emergence of Globalfoundries over the past 18 months.

 

But McClean expressed doubt that Globalfoundries has the muscle to start an all out price war with TSMC, noting that many of Globalfoundries' planned capacity additions won't come online for many months.

 

"I'm not sure how much Globalfoundries can exert in terms of pricing pressure," McClean said. 

 

Despite losing a small amount of market share last year, TSMC still held nearly 45 percent of the foundry market in 2009, according to Gartner. Freeman said both Globalfoundries and Samsung are pushing hard to cut into that lead.

 

Asked what is the best case scenario for Globalfoundries, Freeman said the fledgling foundry could steal significant market share from TSMC if and when parent company Advanced Micro Devices Inc. pulls the manufacturing of its ATI graphics processors from TSMC and gives the business to Globalfoundries.

Freeman also expressed concern about the rapid increase in capacity by foundries, calling the 300,000 wafer starts per month that the industry has added in recent times "way too much." He likened it to the situation in the 1990s, when DRAM companies and foundries added a great deal of capacity, creating a glut, in order to increase market share.

 

"What it looks like today is that they [foundries] are trying to buy market share again," Freeman said. "I think we many have some excess capacity out there. As you get into that excess capacity, that will be a huge advantage for the fabless companies."





KB3001

7/13/2010 7:03 AM EDT

Good news for fabless companies indeed. The market oscillates between fragmentation and consolidation depending on the balance between supply and demand, and this industry is no exception. One thing that I would want people from the fabless semiconductor industry to comment on is how do you prevent your direct competitors from accessing your know-how and R&D investment if they decide to use the same fab?

Sign in to Reply



resistion

7/13/2010 11:18 AM EDT

If everyone crowds 40 or 28 nm, then the problems will happen. 20/14 nm is going to be based on effectively doubling the critical layers, making it more complicated. I'd rely on foundries for FEOL, but for key prototypes might do my own BEOL, for example.

Sign in to Reply



pixies

7/13/2010 12:27 PM EDT

Last month I went to scavenge some used equipment at the defunct Qimonda fab in Richmond. It was an absolute shock for me to see a gleaming, stat-of-art manufacturing facility turned in to a ghost town. Such is the evil of overcapacity.

Sign in to Reply



FH1

7/13/2010 9:09 PM EDT

"Freeman also expressed concern about the rapid increase in capacity by foundries, calling the 300,000 wafer starts per month that the industry has added in recent times "way too much."

In your dreams; this is simply not true !!! Capacity is still falling overall, barely increasing (1% in Q1) at the leading edge.

Making a statement about 'plans to increase capacity' is just PR /marketing gobblegook not capacity. Even if firms did spend that much investment, it would take 2-3 years for it to be real wafer starts, even for Masters of the University TSMC.

The real real world? Leading edge Cap Ex is sold out; Cap Ex investment in new capacity in 2008 (2010's capacity) was negligible; likewise 2009 (2011's capacity) only this year has there being any real new capacity spend but this will not impact until 2011 at the earliest (if for an existing cleanroom). Welcome to the real world!

Sign in to Reply



Please sign in to post comment

Navigate to related information

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)