“In the last couple of years, with Intel being forthright about getting into in the mobile space and quoting process node advantage as a way to leadership in the marketplace, that has woken up the foundries to the challenge,” says ARM vp Noel Hurley.
Now, TSMC, Samsung, and GloFo are all saying they’ll be out with mobile SOCs on 14/16nm this year. That is the same timing as Intel which says it will be producing on 14nm this year.
“The foundry group have stepped up to the challenge and, on the IP side, we have stepped up,” says Hurley.
But, although Intel plans to have 14nm ready for production this year, Intel will not be using its 14nm for integrated mobile SOCs. For the next eighteen months, maybe two years, Intel will be using the foundry companies for its mobile baseband processors.
“We’ll bring that (baseband) in on our 14nm process either late 2015 or early 2016,” says Intel CEO Brian Krzanich. By that time Intel should have got its 10nm process up and running, so it looks as if Intel has decided to keep integrated mobile SOCs off its leading edge process for the next two nodes (14nm and 10nm) at least.
This means Intel gets no process advantage when it comes to integrated mobile SOCs and is a long way behind Qualcomm in mobile design – with no intention of integrating its baseband and apps processors – as Qualcomm has already done – until late 2015 or early 2016.
While Intel has decided not to use process as a competitive differentiator for its baseband processors, ARM is gradually encroaching on Intel’s position in server processors.
“We now have 16 licensees in the server space,” says Hurley.
Asked when we can expect to see more server chips from these licensees , Hurley replies: “In the next quarter.”