Online information source for semiconductor professionals

Foundry fun and games begin

10 September 2009 | By Mark Osborne | Editor's Blog

Popular articles

Oberai discusses Magma’s move into solar PV yield management space - 29 August 2008

‚??Velocity‚?? the new buzzword in Intel‚??s PQS annual awards - 12 April 2012

Applied Materials adds Jim Rogers to Board of Directors - 29 April 2008

TSMC honors suppliers at annual Supply Chain Management Forum - 03 December 2008

Sematech Litho Forum: Sematech mulling multi-beam mask writer effort - 12 May 2010

It’s always interesting to observe foundries waffle-on about the cost of fabrication, the growth in fab-lite/fabless business strategies and therefore the obvious need for foundries to continue to grow and prosper as more of the industry turns to foundries over time. What we haven’t seen them explain in any detail or with any interest is that there is overcapacity at the foundries and wafer ASPs have been in decline for several years. Also we have seen the slowing of node migrations as the cost of design and masks prevents all but a small top 20 elite of companies to push down Moore’s Law as intended.

Basically there has been a need for consolidation in the foundry space even before SMIC threw billions of dollars at becoming just another CMOS foundry. UMC has never been able to ramp either of its 300mm fabs to full production and they were ahead of TSMC in getting 300mm production up and running!

With the oil money coming into the semiconductor industry, things are starting to change and change fast. Globalfoundries’ acquisition of Chartered Semiconductor is the right move at the right time and kind of solves some of the major doubts I had over its ability to survive and prosper as a pure-play foundry.

It gains bulk CMOS processes and fabs at both 200mm and 300mm a year or more quicker than was planned with the second module fill-out in Dresden. More importantly it gets the customers and can now bulk-up against UMC, its initial competitor in the revenue rankings.

Just saying we are turning into a foundry has never proved to be a good business decision, however buying another foundry that already is, does.

I still think that Globalfoundries will need to bulk-up some more so another acquisition could be planned, perhaps in China and not necessarily SMIC, though it must be an attractive buy considering its share price and record of unprofitable operations and according to market share data has 5% share.

However, as Len Jelinek commented at iSuppli two other deals are on the cards this year. A merger between Hua Hong NEC and Grace Semiconductor and UMC’s purchase of China ‘partner’ Hejin. Maybe Globalfoundries will look at one of these as both have China operations?

The key for Globalfoundries is to close the gap on TSMC, which has such a dominant position and forces all other competitors to feed of the scraps. This is a worrying time for TSMC as the oil money, coupled to Globalfoundries advanced processing capabilities and now improved capacity, process portfolio and customer base is surely the biggest threat they have ever encountered, after beating UMC to a pulp in the last 10-years.

That to me suggests that TSMC may be the next to bulk-up. If so we have a major fight for market share looming and in this industry that takes lots of dollars and we know who has those!


Related articles

Foundry fun and games continue - 11 November 2009

The Foundry Company secures $1.2 billion package from NY State - 18 December 2008

GlobalFoundries elects to invest in production ready EUV lithography for Fab 8 in 2012 - 15 July 2010

A global low power design solution - 01 March 2005

Globalfoundries next two 300mm fabs ready for first phase tool install - 12 July 2011

Reader comments

Does the fact that GloFo is perhaps the most complex startup ever assembled in any industry; or the fact that the management team of AMD/GloFo has the all time record in cumulative operating losses and one of the two top places among the worst destroyers of shareholder equity in the industry; plays any role in your analysis? Is it the fact that the oilers have bought into the mirage of a Globally Subsidized Foundry operator affecting your analysis? Yes, consolidation may go on, Moore's will slow down and with that massive commodityzation of the product will further accelerate. Who wins in the end? History shows that the winner in this battle is the ‚??viable‚?Ě scrap buyer; the player that buys the companies just out or in bankruptcy, for pennies on the dollar and that operates the product line for a period close to its nominal lifecycle? Will the oil money be willing to continue to sink cash in an industry that seems to be headed for sustained long term survival efforts? So far GloFo continues to pile up money losing ops in the hope that at some point cost cutting will take place, doesn‚??t that requires attrition by someone at the very least? Whose attrition when GloFo and ATIC keep buying losers well before they crash? TSMC should wait, stay profitable and continue to bleed the petrodollar sources, and then buy them in bankruptcy court, when citizens revolts prevent further corporate welfare for a failed business proposition in a mature industry.
By Adolfo Gutierrez on 18 September 2009
Globalfoundries remains a startup, perhaps the best funded startup ever in the semi industry. However its consistent inability to perform, its continued reliance on subsidies, debt and equity dilution, makes it still a mirage until profits finally roll in. Will they ever? Is the semi industry becoming another heavily subsidized global industry, just another unfunded taxpayer luxury? or is it simply that the inefficiencies are so widespread that Globalfoundries will be regarded just as another mirage on the desert when it finally collapses? Perhaps, and after all, someone willing to lose money in copious quantities can be the last left standing, as long as the losses can be folded onto irrational oil surpluses. Is the Arab mirage enough to keep it alive?
By Adolfo Gutierrez on 15 September 2009

Post your comment

Please enter the word you see in the image below: