Intel: We Will Not Skip 10nm For 7nm, Ice Lake On Track For End Of 2020 Launch
I honestly cannot remember the last time these conferences contained this many nuggets of information for the tech press. Intel has revealed at the recent UBS conference that it will not be skipping 10nm to go directly to 7nm - putting to rest rumors that had been alleging that Intel was planning to do just that. It also gives us their answer to the process ramp dilemma.
Intel: We will not skip 10nm, Cooper Lake on track for 1H 2020 and Ice Lake on track for 2H 2020 production
Before we begin, Intel sent out the following blurp to the tech press:
Intel remains on track for delivery of the Whitley platform starting with production of Cooper Lake in H1 2020 followed by Ice Lake production in H2 2020. We are also on track to follow Whitley with the delivery of Sapphire Rapids in 2021. - Intel Spokesperson
Intel's 10nm has been the bane of the company's existence for well over two years now. The process where the company lost its lead over the industry, 10nm has been in the pipelines for quite some time now - without actual delivery. Fortunately, however, it looks like the company will finally have high-performance server solutions on the shelf by end of 2020 with production slated to begin in 2H 2020. The company also confirmed that they are on track to deliver Sapphire Rapids (and along with it the Ponte Vecchio Xe GPU) in 2021.
So we think we have a strong position in 14-nanometer. And of course by the end of 2020, we'll come out with Ice Lake, our first 10-nanometer server solution that brings important performance per core benefits to the surface. So all-in-all, I think we have an aggressive road map that we feel pretty good about to deal with the competitive landscape.
-Venkata S. M. Renduchintala, Group President-Technology, Systems Architecture & Client Group and Chief Engineering Officer, Intel Corp
This is around the time that UBS' analyst asked Intel whether they were considering skipping 10nm and go directly to 7nm. This is a move that has been widely regarded as something that would be beneficial to the company allowing it to make up lost time and shift directly to an EUV based node that should be very easy to manufacture. It might also allow it to achieve parity with TSMC.
So it sounds like Ice Lake will be broadly available on the second half of next year just like you've said. But that will still nominally put you a year plus behind your peers sort of just from a process perspective. And I know that there's more to it than just process for sure. But I continually get the question well why not just skip 10-nanometer and go directly to 7-nanometer, if you feel so good on 7-nanometer, why not just our right skip 10-nanometer? -Analyst
Here is what Intel had to say, however:
So again it's a scenario where there is a lot of goodness in 10-nanometer that carries forward to 7-nanometer. And quite frankly, we're at a point in 10-nanometer where we have come through [ph] the order of (00:31:19) disentangling the core fundamental technology issues. And we're now in a point where we're ramping in high volume production. And by the way even though we went through that trial and tribulation that delayed 10-nanometer for a couple of years in coming out, we've actually delivered it now without compromised even to performance or the scaling capability that would be required of it.
So having gone through that we believe there's great merit in continuing forward in delivering on the values of 10-nanometer. By the way let me just make sure that 10- nanometer isn't a single flavour, there's going to be 10-nanometer, 10 plus nanometer versions of that. So you'll see continuous goodness coming through 10-nanometer that will be delivered concurrently with the ramp on 7- nanometer.
So, again, going back to my former point we see nodes overlapping and living for a lot longer. We believe that we've got a very, very solid technology platform in 10-nanometer. We want to make the most use of it. And I think you'll see that in Ice Lake and in many other client products that are coming out. Ice Lake client product that came out earlier this year will be following that up with our 10 double plus product in the early part of next year. So we're really buzzed about the roadmap we have on 10-nanometer and see no reason to foreclose on that as we continue to ramp 7-nanometer.
So there you have it. The company has no plans of ditching its 10nm process. In fact, Intel will be doubling down on it and introducing + and ++ generations on it as well. This is something they have pulled off with great success on the 14nm node and should give them more time to prepare for 7nm. TSMC on the other hand, and ironically, seems to be aggressively pushing forward into 5nm (equivalent to Intel's 7nm) and should have products on the sheld by 2021.