Intel has been a bit economical with the truth on 10nm but SemiAccurate can give you the real story. Officially 10nm is shipping now, and technically speaking it is, but when will it be more than a PR stunt?
On April 26, Intel held their Q1/2018 earnings call. It was a wealth of information about how things are going badly off the rails at Intel, as long as you knew how to interpret the doublespeak. Take this quote from Brian Krzanich for instance.
“We continue to make progress on our 10-nanometer process. We are shipping in low volume and yields are improving, though the rate of improvement is slower than we anticipated. As a result, volume production is moving from the second half of 2018 into 2019. We understand the yield issues and have defined improvements for them, but they will take time to implement and qualify. We have leadership products on the roadmap that continue to take advantage of 14-nanometer, with Whiskey Lake for clients and Cascade Lake for the data center coming later this year.”
Sounds great doesn’t it? They understand the problem, have a fix, and anticipate the ramp to take place next year. Lets parse this a bit, starting with the admission that 10nm was set for 2H/2018, something we told you was unquestionably not going to happen last December. We knew the schedule then, it seems curious that Intel management did not know. Then again SemiAccurate has been pointing out curious statements in Intel’s 10nm messaging since we exclusively revealed that they had Cannon Lake silicon back two and a half years ago.
Then comes the biggest shock, the ‘logic’ about the ramp. Officially speaking Intel directly stated it has identified the problem, has a fix, but the ramp is now set for 2019 rather than the mythical 2H/2018 target. If you know anything about semiconductor manufacturing, a new spin of a device takes about three months to pass through the fabs for production wafers. A high priority test lot can be done in possibly two months, but the manufacturing takes a lot of time, there are no shortcuts. If what Mr Krzanich said was true, Intel has the ability to run about four tests of their ‘fix’ and still hit a January 1, 2019 ramp start, assuming no overlapping test cases which they are undoubtedly doing.
Why is this problematic? It sure looks like the bit about, “defined improvements” is more of the blind panic guesswork SemiAccurate moles are describing and far less of the fix Intel wants you to believe it is. We will state for the record that there is no way a known fix would take eight months to implement, period. Do recall that is for the start of the ramp, not production parts for sale, add 4-5 months to the tally for that.
There is nothing stated that we can call out as intentionally inaccurate but taken together they both do not logically parse and, for some people, point to something that isn’t the case. If people ‘misinterpret’ that on their own, it isn’t Intel’s fault, right? The real problem is putting this messaging up against the real schedules which Intel doesn’t dare talk about in public, then things look really grim.
Lets start out with a question from Stacy Rasgon, again from the same call, asking of the 2019 release was first half or second half. Brian Krzanich replied, “So I’m just going to correct you. You said that supposedly we have the solutions. We do understand these, and so we do have confidence that we can go and work these issues, Stacy. Right now, like I said, we are shipping. We’re going to start that ramp as soon as we think the yields are in line. So I said 2019. We didn’t say first or second half, but we’ll do it as quickly as we can based on the yield.”
The above is classic PR speak for “2nd half”, so the 10nm ramp just moved from 2H/2018 to 2H/2019, not January 1, 2019 as they probably hoped you would believe. This is not a minor slip, again. Also note the confidence level BK portrays, it isn’t the strong, confident tone you would expect for a solved problem. Better yet is what they are telling partners.
Note: The following is for professional and student level subscribers.
Disclosures: Charlie Demerjian and Stone Arch Networking Services, Inc. have no consulting relationships, investment relationships, or hold any investment positions with any of the companies mentioned in this report.
Latest posts by Charlie Demerjian (see all)
- HyperX ships it’s 60 millionth enthusiast memory module - Oct 15, 2018
- Bittware/Nallatech water cools 300W of Xilinx FPGA - Oct 12, 2018
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018