Your manufacturing process only runs as fast as the slowest process in it. My impression (though I am not a semi engineer) is that CMP is viewed as an annoyingly low-tech nuisance. You've just etched with all these high-tech lasers and photomasks, done all this wacky CVD, and now you have to run it through a machine which is essentially a bunch of very expensive Brillo pads -- plus it eats lots of slurry and dumps lots of federally-regulated waste out the back.
This means that if you have a clearly superior machine, you can charge more than the competition and get away with it (look at SFAM's margins). This is especially true in manufacturing equipment, where the incremental value of processing X more widgets per hour (for instance) translates into X * 150+ hours/week more revenue for the manufacturer. Let's also remember that most semis require multiple CMP steps, so multiply Y steps * X * 150, and any difference at all is magnified greatly.
The bad part comes if others' equipment is roughly equal to yours. Then you are in a commodity market, where price and service (which costs you $) are the main decision factors, and your margins go to hell.
To avoid this, Speedfam must continue to spend on R&D. Their investment will be a good one as long as they can maintain technical superiority, and therefore reasonable profit margins. If they can only maintain pace with AMAT/IPEC/Ebara, they (and we longs) will be hurt badly.
This, incidentally, is my educated guess as to why IPEC never makes a profit. They can keep pace well enough to get INTC's business (and some others), but are not far enough ahead to demand a premium over their own cost of goods.
Long,
Steve |
