On microprocessor giant Intel's (NASDAQ:INTC) July 27 earnings conference call, management provided some insight into when the company plans to release the first chips built using its 10nm manufacturing technology and what the ramp-up of such products could look like.
Intel CEO Brian Krzanich told investors the company intends to launch initial products on 10nm that are "simpler SKUs [stock-keeping units] at the beginning." Then, "toward the middle and back half of the year," the company plans to launch "high-performance, high-complexity SKUs."
Although I might be reading too much into this, I think these comments strongly suggest that Intel plans to launch products based on its second-generation 10nm architecture, known as Ice Lake, in the second half of 2018.
Do "complex SKUs" of Cannon Lake even exist?
Intel's first-generation 10nm architecture is known as Cannon Lake. The Cannon Lake parts are expected to come in only low-power variants for thin-and-light notebooks and convertible notebooks.
Both of those low-power variants are expected to be what Intel refers to internally as 2+2 configurations -- that is, they'll have two processor cores and GT2 graphics. The first number is the number of cores, and the second is the number following the GT, or "graphics tier."
The higher end of Intel's product stack is expected to be served by Intel's Kaby Lake Refresh and Coffee Lake processors. Kaby Lake Refresh is expected to come in a 4+2 configuration, while Coffee Lake is expected to come in 4+3e and 6+2 configurations for high-performance notebooks. The "e" indicates the presence of on-package eDRAM, enhanced dynamic random access memory, which is used to speed up graphics tasks.
Intel is not expected to build Cannon Lake-based products that will directly replace the Kaby Lake-Refresh or Coffee Lake parts. Instead, they are expected to co-exist, with the Kaby Lake Refresh and Coffee Lake parts handling the higher-performance and higher-power segments of the personal-computer market, with Cannon Lake-Y and Cannon Lake-U handling the lower-power, lower performance parts of the market.
So, if Intel is planning to release "high-performance, high-complexity" products built on 10nm in the second half of 2018, then I see one of two possibilities:
- Variants of the initial Cannon Lake-U and Y parts that perhaps run at higher frequencies than the parts that initially launch are what Intel considers "high performance" and "high complexity."
- Intel is planning to launch its second-generation 10nm architecture, known as Ice Lake, in the second half of 2018.
I don't think we can completely rule the first possibility out, but I think the second, based on Krzanich's statements and the increasingly challenging competitive environment is more likely. Krzanich did say that Intel continue to see "intense competition across [its] businesses."
If Intel doesn't release Ice Lake-based products for the high-performance notebook and mainstream desktop markets in the second half of 2018, the company will have to make d0 with its upcoming Coffee Lake-S for mainstream desktops, Coffee Lake-H for high-performance notebooks, and Coffee Lake-U for high-performance but lower-power notebooks until 2019.
That would be, from a competitive perspective, less than ideal.
Has Intel done this before?
There is a precedent for the plan that I've described -- it's exactly what the company did in 2015. Intel's first 14nm processor architecture, known as Broadwell, came out very late because of issues with the then-new 14nm manufacturing technology.
Since Intel didn't want the Broadwell delay to have a cascading effect on the company's product pipeline, it launched the Broadwell-U and Y parts for the notebook market -- and, surprisingly enough, even some Broadwell-H parts for the high-performance portion of the notebook market -- during the first half of 2015.
Then, in the second half of 2015, Intel launched its sixth-generation Core architecture, known as Skylake, across all its different segments, including Skylake-Y for fanless notebooks and convertible devices, Skylake-U for mainstream notebooks, Skylake-S for desktop personal computers, and Skylake-H for high-performance notebooks.
Intel might be planning something similar in regard to the Cannon Lake/Ice Lake transition. If Intel pulls it off, then it'll help to undo some of the damage the struggles with the 10nm technology did to Intel's product pipeline.
And for what it's worth, I'd be highly impressed with Intel.