The impact of the latest semiconductor industry downturn goes further downstream than the red ink on the balance sheets. According to SEMI’s World Fab Forecast updated recently, 19 fabs closed in 2008 and as many as 31 facilities are expected to have closed by the end of 2009.
I can’t recall that many being closed in 2001/2, during the worst downturn in the industry’s history. This more so when you add in a further 18 closures SEMI expects to see in 2010, making a grand total of 49 fabs!
Not surprisingly, SEMI also expects installed fab capacity to decline by 4% in 2009 mainly due to the number closed fabs and the fact that new fab expansions were put on hold and capital spending reined in to unprecedented low levels.
Of course there has to be a rebound as fab utilization rates have rapidly increased in the second half of ’09. However, I doubt we will see many new fabs come on stream in 2010, simply because there are actually very few sitting idle or where construction started in the last 18 months and could therefore be fitted-out when required.
There has developed a big hole in the timing of new constructions and tool installs that I don’t think can be corrected until the second-half of 2011. That isn’t a definite either as we would expect to see a number of new fab announcements before the end of this year for that to start filling the hole and its later part of November now.
Capacity expansions next year will primarily be from technology node migrations, especially from DRAM and NAND flash producers. As far as we can tell at the moment, Micron has a dormant 300mm fab in Singapore and Toshiba has significant space remaining at its newest NAND Flash fab to fill but few others in the memory space can say the same.
Equipment spending could increase 65% to US$23B in 2010, according to SEMI, which seems a lot but barely takes us back to 2008 levels and still below those in 2007. This also suggests that spending is more technology focused next year than real fab capacity buys.
The foundries are able to expand capacity at existing facilities (UMC, Chartered, SMIC, GlobalFoundries), while TSMC has been extending fabs for several years so has the capacity to fill.
However, looking around I can’t see many new 300mm fabs on the horizon, without new plans being announced. According to SEMI, fab construction spending is expected to reach US$2.7 billion in 2010, a 70% increase over this year but this is only 23 construction projects of mixed type. Importantly, it’s the same number of projects that were cited in the last report in August, 2009.
At the moment it would seem that it’s a case of out with the old fabs but not in with the new!