Intel delays the 10nm 'Cannon Lake' CPUs until 2019
Intel has once again delayed the release of its next-generation ‘Cannon Lake’ CPUs. Previously planned to launch in Q2 2018, the company announced in their first-quarter earnings report that they wouldn’t be mass producing chips based on the 10nm process until 2019. The 10nm node is essential because a smaller node than the current 14nm one will improve the power efficiency, decrease the heat output and help achieve faster processing speeds. Moreover, AMD has announced plans to start sampling 7nm process-based CPU this year, and if Intel is to maintain its edge over AMD, then it needs to begin rolling-out 10nm node-based chips soon.
It appears that it is becoming quite complicated for the chip manufacturer to transition to the 10nm process due to physical limitations that arise once the transistor sizes are reduced to sub-20nm scale. Initially, the ‘Cannon Lake’ processors based on the 10nm architecture were expected to arrive in 2016. However, since Intel couldn't bring 10nm-based chips into mass production, they have stuck with the same 14nm node from 2014 for their 7th and even 8th generation of CPUs. Nevertheless, CEO Brian Krzanich mentioned that the 14nm process used in the Coffee Lake CPUs performs 70% better than the one used in Broadwell chips. As for the complexities Intel is facing in transitioning to the 10nm process, Krzanich explained it is because they are trying to increase the transistor density by 2.7x this time. Comparatively, the usual increase in transistor density from node to node is under 2.0x, and that is why Intel needs more time to implement the 10nm node.
Interestingly enough, Intel competitors Samsung and TSMC have also been rolling out 10nm-based processors for quite a while now and are moving towards sub-10nm chips. In fact, Samsung is ready to develop chips based on the 8nm fabrication process, and we might see 7nm process-based CPUs from TSMC later this year.