A potential good explanation is the fact that chips cost more and more to be made. In a quest for higher speed (more transistors/chip
Wrong ! For the last 30-40 years the cost of chipmaking tools climbs, but the cost per transistor declines dramatically. The more newer technology needed to fabricate smaller features always demands greater complexity and more expensive tools. But the newer generation has greater automation, and has larger wafer sizes. Over the last 40 years the cost per square inch has remained approximately constant.
tools.. have less and less useful life.
Who ever told you that ? The time between the various generations is staying at close to the historical norms.
The electronic chip is near the end of its useful life. Going further down with the critical dimension below 32 nm will be cost prohibitive
I have heard the same refrain every few years. However the latest mega fabs are indeed getting increasingly expensive, so that few companies can afford the cost of building one. This has led to most companies not taking on the risk of building a new fab, but depending on a shared foundry or manufacturing plant. This has not fazed Intel since it is the largest semiconductor company, has deep pockets, and an 80+ percent share of the largest chip market segment.
The only problem has been on the demand side. Computer hardware performance has outraced software needs. Thus the need to build "better platforms" and develop other metrics of performance than raw clock speed.
Silicon Photonics is still a question mark. A lot of engineering still needed, and several years before you see major products based on that.
The market is leery of chips because it cannot see where this years big growth will come from. It is assuming that the anemic post bubble growth rate will continue and assumes that telecom is dead for all time. But it can all change overnight. If the high speed demand from telecom does indeed materialize the chip demand will indeed explode.
News 
Market Data 
Discover 
