| TheRealMafoo said: My guess is whats doing it is slow CPU tech advancement. Most of the cost of a CPU, is R&D. Let's say it takes me $100 million to design something, and it cost $20 to make. let's asume I am expecting to sell that thing at 50,000 a year for 5 year, and then replace it with the next thing that cost me $100 million to design. I project then, that I will sell 250,000 of these things. The cost then, is $100 million across each ($400) + $20. It cost me $420 to make each one. But, if in 5 years I can sell that thing for another year, the cost drops to 20 bucks each, and I still get to sell it for something close to what I used to sell it for. This is the kind of thing that's happening. CPU's are no longer the slow part of computer system. People need a lot more ram or faster storage long before they run out of CPU. I am writing this on a 4 year old Mac Book Pro, and if the CPU in this thing was 10x faster, I would not notice it. |
Actually the fabrication capital costs are incredible. Its more likely that the reason is that they are producing a lot of chips on their heavily depreciated 45nm lines and the economic recovery means more chips sold for more computers. The cost of their 32nm process and fabs is probably greater than the R+D costs on the chips they fabricate with it.
For example: http://www.brightsideofnews.com/news/2009/10/29/tsmc-to-invest-six-billion-usd-in-two-gigafabs.aspx
TSMC is spending $5 Billion on one single fab alone.
 |
| 4. This figure illustrates the historical and projected cost for wafer fabs out to a 450 mm fab in 2012. |
So its a very expensive business to be in. This is why AMD divested itself of the fabs to spread the risk/cost over a wider base. They are transitioning to 450mm wafers in 2012, and that will be a killer. Expect potentially $5-10B fabs at that point in time.
