Fascinating! I guess that's why executives are paid a lot of money, the decisions they have to make can make or break a business.
PS. With all the best prior research and analysis, there is always an element of luck in it.
And then, talking with the Cymer folks, their EUV teech will be rolling out before 450mm is ready, even on the most optimistic predictions, which could increase yields on 300mm enough to make 450mm unnecessary in the short term.
What is also interestng is what happens to 300-mm wafer fabs when the leading-edge digital logic and memory ICs are made in a very few 450-mm megafabs.
History teaches that the 300-mm fabs will become the domain of More-than-Moore, trailing-edge digital, 57 varieties of analog, mixed-signal, image sensors, MEMS and so on.
And most 200-mm wafer factories will probably become economically unviable if they are trying to compete with larger 300-mm wafer fabs.
The result will be a complete changing of the order.
I am with you. I think every tool maker would like to get money up front for 450-mm development. But I am not sure that that many are negotiating from ASML's position of strength in lithography. Sadly, in more competitive equipment markets, bearing the cost of 450-mm development may be table stakes.
Not sure that ASML should set the standard for all tool vendors, though I'm sure any vendor will now like to ask for upfront $$ to fund R+D. Regarding 450mm photo development, it seems we're really talking EUV. I believe there are some toolsets like implant that face significant challenges to upsize from 300mm to 450mm, but I would think that this asml example will be the exception, not the rule.
Thanks MP. I agree with you, and I suspect that Bob Johnson of Gartner would agree also. There will be only a few companies that move to 450. As far as how they will fund it, I suspect that we saw an example of that yesterday, when Intel announced it would take a stake in ASML and provide additional funding for 450-mm and EUV development. This is the way I think it has to be, because the tool vendors themselves are not going to be able to afford all of the R&D on their own.
Blog Doing Math in FPGAs Tom Burke 23 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...