Intel's presentation at the International Solid State Circuits Conference was held recently (via ExtremeTech) and they unveiled some additional information on process plans. They also assured investors that Moores Law was still alive, albeit taking its last strides now. They also talked about how the last problems with 14nm had been rounded up and fixed, and also the reason for the delay while mentioning that the pilot program for 10nm was still on schedule.
An unofficial Intel roadmap @unknown.
Intel at ISSCC: We understand 14nm now, 10nm will arrive in half the time of its predecessor
Intel's 14nm process entering mainstream market was delayed by about 6-9 months depending on your perspective, however, even with this relatively big delay, Intel managed to keep its node lead over other third party fabrication facilities like TSMC and Samsung. As we have already revealed before Intel's 14nm node is very clearly the best one out there - with no competition. These advantages include better gate length, pitches and SRAM scaling among other things.
Intel made it a point to mention that all these advantages and improvements will remain in affect for atleast 10nm. The scaling will remain the same while the density rises, and cost per transistor plummets. Intel has entered the sweet spot at the top of the line and now the only major problem remains beating physics every time. Not surprisingly however, Intel has given no concrete roadmap for 10nm mass production. And all the technical jargon and marketing aside, the slide itself is a jumble of technologies that are being researched for 10nm and beyond. At 7nm and below physics breaks down pretty quickly and quantum physics effects quickly start becoming a serious problem.
To assuage fears, Intel claims that the rollout time to 10nm will be upto 50% faster than the delayed 14nm. The slides also show assorted materials and technologies that are currently being researched to be able to function as workable computer transistors. These include highly practical and existing tech such as EUV (Extreme Ultra Violet Lithography) to III-V wafers and even tech that has yet to leave the lab properly like nanotubes, nanowires and graphene. Naturally Intel has yet to choose its race horse in the fight after 7nm but this indicates that all of the technologies are being researched concurrently.
Its worth pointing out that ASML holding has already delivered test products and while the technology is still being prototyped, volume production remains far into the horizon. According to Intel plans, the tech will most probably be utilized for tech that is 7nm and beyond. 10nm is going to arrive on the same ol' conventional fabrication plants. Ofcourse the upcoming architectures from Intel wont be without surprises either, Intel is actively developing 2.5 and 3D based stacked dRAM and future processors will definitely feature something of the sort.
