Gee.. Cary so many things to respond. I do enjoy having discussion with you anyway...So here I go again..
RE: "...during the downturn...the next upturn..."
I don't believe the industry is in a downturn and I am not waiting for the next upturn. 2004 was a big improvement over 2003 and I expect 2005 to be about level with 2004. I expect gradual revenue improvements in 2006 and beyond.
- Well it is a matter of wording. Many including me call it downturn when the order fall by more than 10 to 15% range and the fall occurs for more than 2 consecutive Q's. Certainly this is where we are right now and I am just calling it downturn by this definition.
RE: "...these orders were for 130 nm node or larger..."
I think the conference call said something like 85% 300mm and 85% below 100 nanometers. During an "inventory glut", it does not make sense that there will be big orders for older production technology.
-Be careful here. Just because your tools were part of 100 nm processing, that does not mean you are actually etching at that scale. These days the chips are so complicated that it has more than 10 layers of conductor and passivation (dielectric)layers. I have seen some logic chip with more than 16 layers on a research level. At the very first transistor level, etch scale actually corresponds to the node size. This is the toughest etch of all and it is considered "critical".
Starting level up, the etch holes actually gets gradually larger. However, introduction of new processes such as dual damascene makes the etch of next 5 to 6 layers up challenging and hence they are becoming more "critical". Finally for last 3 to 4 layers, etch gets considerably easier as you are dealing with easier feature and capping layers of the chip. I believe LRCX is strong in the middle layer and have made some in-road to the transistor layer against Tel who is so strong here. So not all the etch process for 100 nm chip is created equal. For those critical etches, vendors have better pricing and are able to sell more tools as the process time is typically substantially longer. Chipmakers like to give another vendor with non-critical applications as they hate to rely on one vendor for everything.. But for non-critical applications, productivity is a key and pricing pressure is fierce. By check indicates AMAT had very poor gross margin with the Etch product and I am deducing from this fact that AMAT had more product sold for non-critical applications.
RE: "...next technology driver is ramp on 90 nm process..."
This ramp has been under way for a while. The next driver is 65 nm.
- No. I beg to differ. Full implemetation of 90 nm has not occurred yet worldwide. Selective companies who has aggressive spending plan last year such as Intel and Samsung has inplemeted 90 nm manufacturing more extensively than others. But most are still doing it at small scale as they have recently exited pilot production stage to improve the yield further. These guys will do the upgrade hopfeully soon.
65 nm mode at best will be in the pilot stage by end of this year.
RE: "LRCX owns qualified processes for 10 out of the 11 major 300 mm 90 nm capable customers."
With 3 kinds of etch and critical and non critical layers, what does "qualified processes for 10 out of the 11" mean and how many "qualified processes" do AMAT and TEL have?
-Here I am talking about the oxide etch with respect to transisitor level and 3 to 4 layers above 1st layer and dual damscene etch for logic chips.
RE: "...with etch I prefer LRCX over AMAT."
If and when you prefer AMAT, NVLS will be buying the distressed assets of LRCX.
- If NVLS is buying LRCX (I doubt it will happen), I am definitely prefering NVLS over AMAT
RE: "...which it used the profit to grow the other business."
LRCX tried to do the same in CVD and CMP. It failed. Therein lies the difference between LRCX and AMAT and therein lies the investment risk with LRCX. Of course, with your finger on the pulse, presumably, you have less risk.
-I am willing to take the risk for now. Of course, if AMAT starts to capture more substantial etch market share, I am definitely buying AMAT in favor of LRCX. But it has not happened for last 5 years and it does not appear it will do so in the near future...
JH |
