Intel(INTC +7.06%) had originally planned to release a product known as Cannonlake built on its next-generation 10-nanometer manufacturing technology in 2016, but a recent report from BenchLife provided compelling evidence that Cannonlake has been delayed and that Kaby Lake, another 14-nanometer processor, will be the 2016 PC offering.
Now, according to a report from DigiTimes, Intel will be postponing the launch of Kaby Lake from "early 2016" to September 2016 for the consumer variant and to January 2017 for the enterprise-focused models.
If true, what does this mean for investors?
What is going on with 10-nanometer manufacturing technology?
If it is true that Intel will launch a third volley of 14-nanometer chips (Broadwell is the first, Skylake the second, and Kaby Lake the third) in late 2016, then this suggests that the 10-nanometer Cannonlake part will not arrive until mid-to-late 2017.
Given that Intel had said back in 2013 that it expected its 10-nanometer manufacturing technology to be production-ready by the end of 2015, it seems that Intel is facing some difficulties getting this technology ready for cost-effective, mass production.
Intel had originally said that its 10-nanometer technology would be good to go in 2015. Source: Intel
How does a potential 10-nanometer delay impact Intel's competitiveness?
The impact on Intel's competitiveness from a 10-nanometer delay needs to be evaluated on a per-segment basis. In the server market, this is a non-issue since most of Intel's 2016 and 2017 product lines for that market were likely to be on 14-nanometer anyway.
In the PC market, the situation is a bit more nuanced. For the majority of 2016, Advanced Micro Devices (AMD +7.24%) will likely be fielding products built on a relatively old 28-nanometer process, so Intel 14-nanometer products should have a significant manufacturing edge.
Where things will get interesting, though, is that for high-end desktop processors, AMD -- according to leaks from BenchLife -- plans to transition its products to FinFET manufacturing technologies in late 2016 with its Summit Ridge chip. Then, according to BenchLife, in the middle of the first quarter of 2017 timeframe, AMD plans to start ramping its integrated CPU and graphics parts also on FinFET technologies.
While I believe Intel 14-nanometer technology is more advanced than the 14/16-nanometer technologies from Samsung and TSMC, respectively, the gap should be significantly narrower than the gap between Intel 14-nanometer technology and any 28-nanometer process currently in production.
Additionally, with these 14/16-nanometer chips, AMD plans to bring out a new CPU core, code-named Zen, which the company says is a new architecture. Whether it will be competitive in terms of performance and performance-per-watt with Intel's best 14-nanometer designs remains to be seen.
That said, if Intel wants to maintain its market share dominance over AMD, it needs to get its latest architectures on the most advanced manufacturing technologies out there as soon as possible. If Intel launches its first 10-nanometer parts in mid-to-late 2017, then AMD could have a several quarter window of opportunity where the manufacturing gap between the two companies is not on the order of a full generation.
Finally, in the ultramobile market (tablets and phones), a 10-nanometer delay seems as though it will hurt Intel's competitiveness, particularly as Intel's mobile parts tend to lag its flagship PC processors in terms of manufacturing nodes (although Intel management has been vocal about wanting to close that gap).
With both TSMC and Samsung claiming that they will begin volume production of their respective 10-nanometer technologies by the end of 2016, Intel is unlikely to have any sort of "lead" in terms of timing. Now, if Intel 10-nanometer technology is significantly more advanced than what TSMC and Samsung bring to bear with their offerings, then Intel could have an edge assuming it can get competitive designs out in a timely fashion.
