Market research is a tricky beast at the best of times. It provides an interesting snapshot but has to be taken with a healthy dose of skepticism: are the numbers coming from the supply side or the demand side? Are the participants giving accurate numbers? And how are the numbers analyzed? Years ago in "Physics Today," I saw a cartoon that showed a curve in the first frame, a data fit in the 2nd frame, and an extrapolation of the 3rd. The first frame showed an uneven curve; the 2nd frame showed that the curve fit the back, head, and trunk of an elephant; and the 3rd frame showed that the extrapolation fit three elephants linked trunk to tail. Certainly, the disparity you mentioned between the report and the WSTS numbers is notable. The TrendForce report appears to hinge on increased DRAM demand by PC manufacturers wanting to ensure an uninterrupted supply. As an analyst, what's your opinion of that phenomenon?
It's interesting to see TrendForce predicting 30% growth while WSTS just announced (and the SIA sanctioned) a forecast with memory growth of only 5.9%. Why all this disagreement in a market that has been around for 40+ years?
Memory is the big swing factor in semiconductors. It is the market that should be exopected to grow or shink the most in any given year.
As for your parting sentence: One of the most certain signs that a semiconductor cycle is about to undergo another dramatic transition (from boom to bust or vice versa) is when people start to say that the semiconductor cycle is over and done with.
Funny you should ask about process nodes – I'm in the middle of putting together a slide show highlighting the state-of-the-art for flash, DRAM, and embedded. Stay tuned – it should be up sometime next week.--K
Blog Make a Frequency Plan Tom Burke 17 comments When designing a printed circuit board, you should develop a frequency plan, something that can be easily overlooked. A frequency plan should be one of your first steps ...