Intel has been in a lot of trouble lately, they are going through production woes and are suffering from huge delays in their shift to 10nm.
Intel first promised 10nm in 2015, but they failed to deliver. Even after 3 years, we are yet to see Intel’s shift to the 10nm process for their processors. According to a report from Semiaccurate, Intel might shift to 10nm in late 2019, even that might not be a true 10nm chip, but rather a 12nm in disguise. This delay can be attributed to both management and engineering failures at Intel. Semiaccurate also stated that the current condition of the 10nm development at Intel was awful and the performance yields were also minor.
Taking advantage of Intel’s misstep, AMD is going all in for 7nm. AMD’s CTO Mark Papermaster stated in an interview with CRN that 7nm was a big bet for AMD and they spent considerable resources making it happen. In fact we will be seeing AMD’s 7nm node in action with the upcoming Radeon Vega Instinct GPUs. The new Epyc server chips will also be on the 7nm node.
Further Delays in 10nm
Intel's latest roadmap leak shows Coffee Lake-R Refresh, Atom Elkhart Lake and a fresh round of 10nm delays for the nth time. No 10nm until… H2'20. OMG! pic.twitter.com/en5ePIjybv
— witeken (@witeken) September 18, 2018
According to a recent Intel roadmap leak, we might not see the 10nm processors before 2020. If the roadmap is accurate, we will only be seeing refreshes of current processors next year, which are also refreshes themselves!
This can be a huge blow for Intel. Intel’s 14nm node does have similar specs with respect to 10nm nodes from other manufacturers, like Global Foundries and TSMC. They are still ahead with respect to their competition, but if they implement 10nm by the year 2020, they will certainly lose the advantage.
Lithography figures are quite important as they signify the distance between transistors and how closely packed they are. Lower figures mean shorter distances between the transistors, which means electrons can transverse faster. Along with performance improvements, there’s also significant improvements in efficiency. With every shift to smaller nodes, the process gets significantly costlier and harder.
At this rate Intel might not be able to offer significant improvements in performance for their consecutive product lineups aka the Coffee Lake refresh. By the time Intel starts shipping out processors in the 10nm node, we might already see 5nm being used. If that happens, Intel will be the one playing catch up.