* Giant mergers are up (Intel+Altera, KLA+Lam, etc.), concentrating the industry more than ever.
* And ultimately: A 5-year-old PC still works just fine.
When I say this is the end of Moore's Law, I'm not trying to be dogmatic. Of course there will still be a semiconductor industry and of course there will still be amazing technological progress. But it seems the rate of that progress is slowing, and now the industry is adjusting.
And ultimately: A 5-year-old PC still works just fine.
I suspect that is really the dominant factor here. It's not that the progress within the chip industry has stopped; there is far more number-crunching power in the CPU and GPU in my latest PC than in the one from five years ago. And if you're doing things like playing demanding games or modelling skyscrapers in a CAD package, that progress is probably very useful. It's just that for what most people use PCs for, it doesn't matter, because they were already good enough so unless their old one broke they don't need a new one anyway.
Not only that, but when it comes to replacement, smaller and more convenient devices like smartphones and tablets will do everything a lot of people need without needing a PC at all these days. If you mostly used a PC for things like staying in touch with your friends or retrieving information from a web site, rather than anything creative beyond a quick bit of typing or anything that needs more powerful equipment, you might not even have a laptop any more.
Given that Intel has never had much penetration in the mobile device market compared to ARM designs, it doesn't seem that surprising that the demand for their products is waning as the mass market moves in that general direction.
Moore's law does explicitly talk about semiconductors - with the recent strides in light-based computing (however small) we might be able to continue the overall computational power trend in the future.
Interesting point. If we are witnessing the end, would we recognize it at the time? Moore's Law has been something we've become accustomed to, so it almost seems hard to believe that it will end even if we are told that it will.
Makes me wonder what research money will be spent on instead of just faster chips.
Worst case scenario: nothing. The R&D goes away. Look at aerospace for a nightmare scenario.
In the early 20th century we got fixed wing flight.
In the teens and 20s we got motorized fighters and the first passenger planes.
In the 40s we got jets.
In the 50s we broke the sound barrier and orbited Sputnik.
In the 60s we landed on the Moon.
In the 70s we... stopped going to the Moon.
In the 80s nothing much happened except declassification of a few things (stealth) that were developed in the 60s and 70s.
In the 2000s we grounded the Concorde. Passenger flight got slower and more expensive.
In 2016 we fly on passenger planes no faster than what we used in the 70s and 80s, and we're stuck in low Earth orbit.
1969 was the peak of the aerospace industry. With the exception of SpaceX (which is really just picking up where NASA left off), we are less advanced today than we were in the 1960s.
There's many places we could go beyond conventional Moore's Law: multi-dimensional chips, optical, quantum, exotic materials with very low power consumption, etc. But if what we have is "good enough" and there is little demand for anything faster, the R&D dollars won't be spent. If anything the shift toward mobile computing and wimpy thin client endpoint devices might actually lead to a pull-back and loss of capability similar to the one we saw in aerospace after the 70s.
The consolidation we are seeing is not a good sign. This is what happens when an industry decides it's now a cash cow and it's time to go out to pasture.
We also could have a base on the Moon and Mars right now and be working on our first interstellar probe to Alpha Centauri. Physics didn't stop us. Economics and politics did.
That's not entirely wrong but you're ignoring the massive price difference for air-travel between the 60s and "slower" 2016.
The price drop in air-travel can be attributed to:
1. Technology. Cheaper (per seat), more efficient planes.
2. Consolidation of airlines (and airplane manufacturers).
3. The disruption of full-service airlines by low-cost carriers.
In semi-conductors we have been getting along with 1 and a bit of 2. I would say that the disruption of Intel by ARM is an example of 3 because intel is not incentivized to compete at those low price points.
3. The disruption of full-service airlines by low-cost carriers.
In the US at least, this was due to political deregulation. The process was started by Nixon, but finished in law by Jimmy Carter (!), note also the leading lights of the Democratic party in the signing picture, e.g. Teddy Kennedy 2nd from right: https://en.wikipedia.org/wiki/Airline_Deregulation_Act
Maybe it is. But if the general public is happy with what they already have, manufacturers tend to iterate instead of innovate. Getting from europe to australia two times faster, for ten times the normal ticket price makes me settle for the slower option.
Even a 200% increase in CPU performance won't bring the general public back to desktop PC's they don't really need, because they are satisfied with what they have. We need something truly disruptive and get people interested in something else.
Right now there are a lot of experiments (AR, VR, wearables, transparent screens, lightfield, new batteries, etc) which will, probably when combined into a truly attractive package, change everything. Again.
* Tick-tock is dead.
* 10 nm is severely delayed.
* EUV is severely delayed.
* Significant layoffs in R&D
* The ITRS roadmap is vaguer than it's ever been.
* Giant mergers are up (Intel+Altera, KLA+Lam, etc.), concentrating the industry more than ever.
* And ultimately: A 5-year-old PC still works just fine.
When I say this is the end of Moore's Law, I'm not trying to be dogmatic. Of course there will still be a semiconductor industry and of course there will still be amazing technological progress. But it seems the rate of that progress is slowing, and now the industry is adjusting.