One way that microprocessor giant Intel (INTC -0.61%) has aimed to maximize the profitability of its core personal computer processor business has been to very aggressively segment its products. The idea is to build products with the right cost structure and performance level at each price point.
Over the last several years, Intel has put a lot more effort into its lowest-cost personal computer processors. These are based on the company's low-power Atom processor cores, so they're slower than its mainstream Core i-series processors. However, they are also significantly cheaper to build, so Intel can sell them quite cheaply, making low-cost desktops and notebooks much more viable.
In 2013, the company introduced Bay Trail-M (for notebooks) and Bay Trail-D (for desktops) for such low cost PCs, based on its 22-nanometer Silvermont processor. In early 2015, the company followed those up with Braswell -- a product manufactured on the company's 14-nanometer technology. And now, in late 2016, the company is in the process of rolling out Apollo Lake, which is a second generation product manufactured in the company's 14-nanometer tech.
So, a question worth asking is, "what comes after Apollo Lake?"
And the next Atom processor for low-cost PCs is called...
According to a leaked Intel slide that appears to have originated from a poster on the AnandTech forums, the next generation Atom processor is called Gemini Lake. It's the direct successor to Apollo Lake and it's expected to arrive during the fourth quarter of 2017.
Very few additional details are given in that slide, but with a code-name in hand I did a little bit of digging and came across a key detail.
Gemini Lake to be manufactured in 14-nanometer manufacturing technology
Per a LinkedIn profile of a former Intel employee, Gemini Lake will be manufactured in Intel's 14-nanometer manufacturing technology.
Given that Gemini Lake is targeted at low-cost personal computers, and given that the company's next generation 10-nanometer manufacturing technology could be quite expensive for a while as it will be quite new in late 2017, a mature 14-nanometer manufacturing technology seems like the appropriate one to build this product on.
It is likely, though, that instead of the same 14-nanometer technology used to manufacture Braswell and Apollo Lake, Gemini Lake will use the new 14-nanometer+ technology that Intel says it's using to build its seventh generation Core processors. This improved manufacturing technology alone should allow for improved performance and power efficiency relative to the prior generation Apollo Lake.
It's not clear at this time whether the company is also planning any substantial architectural updates, but I wouldn't be surprised to see Intel upgrade several of the key intellectual property blocks on the chip (such as the graphics/media subsystem and the image signal processor). It would also be nice to see an updated processor core, but that may be a little too much to expect.
Intel's manufacturing strategy is coming into focus
In a previous column, I took a close look at what Intel's strategy vis-à-vis manufacturing technology adoption appeared to be. In a nutshell, it seems that in order to preserve margins, Intel is only transitioning its personal computer products to new manufacturing technologies in areas where it feels that it needs to.
For segments of the market that Intel thinks that it can still remain highly competitive even on an older manufacturing technology (through performance enhancements to those more mature technologies coupled with architectural changes to the products themselves), it seems to be planning to stick to those older technologies.
With these details of Gemini Lake now known, it would appear that a large portion -- likely the majority -- of the company's processor shipments from the second half of 2017 until at least the third quarter of 2018 will be on 14-nanometer technology rather than on the newer 10-nanometer technology.
The good news is that the 14-nanometer technology should be very healthy over that period, which should help keep manufacturing costs down and profits up. The bad news is that this seems to imply that Intel's 10-nanometer technology, though it has already been significantly delayed (giving Intel's engineers more time to work on it), isn't expected to be healthy enough to support a bigger portion of the company's product shipments until, perhaps, the second half of 2018.