It’s not just the technology Intel strives to better with every generation, but also its human taskforce.
“[The engineers] had better be better these days, way better than when I was an engineer,” said Permutter, adding that today’s engineers came at things “with a different mindset.” This, he said, meant the new generation was used to products with much broader market capabilities. “In some ways that helps a lot, in other ways it makes it harder,” he said.
Despite the seeming solidity of Intel’s roadmap, however, Perlmutter is well aware that the road ahead may not be entirely bump free.
“I always worry. I was worried when we moved to 1 micron. I was worried when we went below 100. I’m worried about going below 10. But my colleagues running the technology side tell me they are worried, too, and the moment they are worried, I’m no longer worried, because they do their job,” he said.
Having said that, ARM and its partners might do well to worry a bit more about Intel’s progress, with the firm’s 14-nm node already in “extremely advanced stages of design.”
Perlmutter said his firm is holding firm to its schedule of putting its first products on 14 nm in production later next year .
“We will launch many products on 14nm. Our plans are advancing and we believe that we will grow the gap and it will give us more advantage,” said Perlmutter.
“Experience is enabled by performance. The better performance you have the better you are. It’s all about performance,” he said, adding “high performance computing in a small form factor.”
Analysts say that Intel now has about a 2 year lead in process technology. This is a huge advantage. And as the lithography & equipment complexity & cost goes up with each new generation it's more difficult to keep up and even afford to build new fabs. Intels new 14nm fabs are about $7B each and the list of those who can afford it has gotten much smaller. Today, only Intel, Samsung & TSMC can afford them. This is one reason why Intel is getting into the foundry business too.
I'm not sure where you're looking but costs of Atom chips to Intel are 10x cheaper than their PC chips because they're so much smaller. 22nm chips are cheaper than 32nm and 14nm are cheaper than 22nm versions. Of course, the performances go up and the power down each new generation too. 14nm will be a killer chip for Intel that ARM won't be able to match
Human civilization continues to advance due to the species' unique ability to build on knowledge. Today's engineers are smarter (read Scientific American Sept 2012, p44). Dadi Perlmutter is as brillian an engineer and manager as you will ever meet, which explains his current responsibilities. Dadi "gets it" and has for more than 15 years. He also gets people, has the humilty and wisdom to attract and utilize the best among us.
- Kurt Robinson
One hope ARM has is that IBM's design for yield tools and SEM/TEM capabilities are helping GF get their Malta fab up and running faster. And ARM-GF have new long term partnership lately. IBM may have learned to be more agile and sneaky that in distant past when they let intel empower the cloning of their PC's by using the easy-to-rev-eng BIOS ROM chip. But its all about execution, so we will see over next 3 years. Maybe Warren Buffet realized this was a game he did not understand? Risk is very high next 3 years.
Before, when buying a PC, the CPU (Intel or AMD or whatever) was the key thing. Now, when buying a tablet or smartphone, most will not care if it's an ARM or Atom processor inside. They would only care about the final product design by Apple, Samsung, or whomever.