The latest update of Gartner Dataquest's semiconductor capital equipment spending forecast paints a grim picture of the market's prospects this year, with capital expenditures expected to slide 19.8% compared to 2007's results (down from the 13.2% downtick forecast issued in December).
The wafer fab equipment (WFE) sector's revenues will drop 17.4% this year, according to the research firm. With the DRAM folks finally getting a grip on their overspending, and the logic and foundry guys proceeding cautiously, only the NAND flash segment is expected to increase its capex in 2008. And don't forget the macroeconomic situation: it's official, Gartner concedes, we're in a recession.
I spoke with the company's Bob Johnson earlier this week, who believes "this correction was long overdue. The underlying thing that I'm looking at here is that we had a couple of years that were bubble years, largely due to the memory market doing its periodic insanity, and now we're paying the price. The reductions in DRAM spending were much more severe than we projected last December, and that's probably welcome news for the long term."
He likens the current situation to the infamous memory-driven implosion of the mid-1990s. "One thing you have to look at, is that in 2006 and 2007, there was a definite spending bubble there, and it was driven by DRAM. If you look at that on a graph, it bears an uncanny resemblance to 1995 and '96, when the DRAM market crashed."
Bob also alluded to some changes going on in the foundry space, where "people are moving designs into the smaller feature sizes slower than they have in the past...because of the cost of the design. You find companies like Broadcom doing everything, moving it down, but the other guys, their competitors, are going very selectively and even keeping some stuff up at 90- and 130-nm and looking at packaging solutions, just because the cost of a design is so huge at those smaller feature sizes. That's one of the reasons foundry spending is slowing."
We briefly discussed the equipment forecast in more granularity, noting that, "to no one's surprise," as Bob put it, that the lithography sector will be the least affected by the challenging conditions. I asked the Gartner VP to send me the tables with the latest tool segment data, and most of the numbers can't airbrush the even gloomier picture.
Even litho doesn't escape the relentlessly negative market environment: the scanner folks are expected to drop nearly $791 million this year to about $6.7 billion, or a 10.6% hit, making it the least damaged segment, small consolation to the likes of ASML, whose latest earnings make this trend painfully obvious. Their cousins in the photoresist processing/track business will plummet 24% in 2008, the worst percentage decrease of any of the tool groups. Rapid thermal processing and oxidation/diffusion (-19.8%), deposition (-21%), and the largest segment, etch/clean/planarization (-18.6%), all will suffer downturns in excess of the total WFE average of -17.4%. Ion implant/doping (-16.5%), manufacturing automation/control (-17%), and process control (-16.4%) won't fare much better.
Although WFE as a whole is expected to rebound in 2009, rising 9.8% overall, only the litho segment will exceed the number it enjoyed in the bubble year of 2007. As a result of the downward revision, all the forecast totals out to 2012 have been lowered by a billion or two. Even 2007 wasn't as flush as at first blush, with the revised wafer-fab tool total off $270 million from the number cited in December. That's a bit deceiving, since several segments--RTP, etch/clean/CMP, automation and control--actually came in with higher sums than shown on the previous forecast.
If you want to look at something really sucky, then cast a glance at the compound annual growth rates, or CAGRs (which sounds almost like "kegger," when Bob pronounces it). The total WFE CAGR for 2007-2012 is now projected at a whopping 1.0%, according to the Gartner's latest figures. If you throw in the packaging/assembly/test data (which I begrudgingly do), the number rises to an even more impressive 1.3%.
The individual equipment segment CAGRs for the period range from -1.4% for RTP to a reasonable 5.2% for the lithographers. Track, implant, etch/clean/CMP, and automation/control all show slightly negative or parity rates as well, and only process control gets into the black, barely. When compared to the same numbers on the December forecast, almost every segment tallied a 2% or better CAGR for the 2006-2012 period (albeit not exactly the same time span).
If you drill down deeper into the tool segment info, a handful of bright spots can be found, but they are few and far between: the litho alternate exposure/write tools market should experience a 70% CAGR between 2007 and 2012, ultra-high-dose doping equipment should grow at a 20.5% clip, and electrochemical deposition gear should increase at a modest but healthy rate of 7.2%, while single-wafer wet processors (+7.1%), spin/spray processors (+5.9%), and other clean processes (+9.8%) will also head in a positive direction. On the process control front, optical CD and macrodefect inspection subsegments are expected to climb about 10% each.
The only billion-dollar-plus market opportunities in 2012 among these groups? Those surface preparation categories, which total about $2.67 billion when agglomerated. But before you throw all your wafers into those tools, also keep in mind that two of the worst individual CAGRs lurk among the removal equipment categories--metal etch at -21% and batch spray processors at -9.8%.
The sobering data noted here underscore what Bob thinks about the chipmaking community's prospects in the years ahead. "The semiconductor industry is entering into a long-term period of single-digit growth, and I don't see double-digit growth rates returning."
Given the givens, some people may be quite happy just to hit single digits on a consistent basis.
|
Comment by GUEST on 2008-04-18 11:33:50
great piece, thank you; there's currently a groundswell rally building in all the chip stocks, who have had the bejesus kicked out of them over the last 6 months, reading this makes me think that long term, it would be prudent to sell the upticks and the only name you really want to have left on the pad is ASML | Comment by GUEST on 2008-04-22 17:40:22
great piece, thank you; there's currently a groundswell rally building in all the chip stocks, who have had the bejesus kicked out of them over the last 6 months, reading this makes me think that long term, it would be prudent to sell the upticks and the only name you really want to have left on the pad is ASML |
|
Blogs 
