[ Ссылка ]
At the DesignCon conference today, I've attended the Text 100 sponsored panel on the "Future of the Electronics Industry" with representatives from AMD, ARM, Cadence, HP and The MathWorks and moderated by Gartner's Research vice-president, Bryan Lewis.
In his introductory remarks, I found Lewis picture of the state of the semiconductor industry a bit bleak:
- The semiconductor industry is slowing down. The growth rate is now in the 5-10% a year versus 20-25% in the 80s;
- Costs are rising really fast. On the design side, hhe verification and the engineering time as well as the embedded software are the main culprits. On the manufacturing side, it now takes about $3.5 billion to build a new fab facility. Finally, the development process R&D has also gone through the roof. When $1.5 billion was enough to pay for the 65-nm node, you now need twice more, yes $3 billion!, to get to 32-nm;
The implications are two fold:
- It's bad for innovation which tends to slow down as there are fewer designs created as companies are trying to amortise their R&D investments;
-Semiconductor companies have no choice but to partner, except Intel I guess, and create large consortium to reduce both costs and risks.
But Lewis final conclusion is really scary. Because the cost structure to move to the 32-nm process is so prohibitive, only a few will be able to afford it. Like Intel and perhaps AMD/IBM. The others will be left with a "lagging" technology and have to go after lagging markets to survive. 32-nm is going to be the definitive test that will separate winners from loosers.
[ Ссылка ]
