One thing that has been truly frustrating about the delays that Intel (INTC -1.55%) has faced in bringing its 10nm technology into mass production has been the stagnation in the company's processor innovations as a result.
After it released a processor known as Skylake back in the second half of 2015 built using its 14nm manufacturing technology, Intel was supposed to introduce a new processor design known as Cannon Lake using its 10nm technology in 2016.
Cannon Lake would've brought improvements across the board to Intel's desktop and notebook processors thanks to an enhanced processor core, new graphics engine, support for more power-efficient memory technologies, and possibly the integration of new components, too.
Unfortunately, since Intel could not bring its 10nm technology into mass production on schedule, its chip teams went with an alternate plan: For two subsequent processor generations, Intel would improve the performance and power characteristics of its 14nm technology and use those to build faster, more efficient versions of the Skylake processors. This strategy worked -- for the most part. Intel delivered better processors last year and this year with these enhanced versions of 14nm, and products based on Intel's second 14nm enhancement -- marketed as 14nm++ -- are expected to carry the bulk of Intel's business through 2018.
Cannon Lake-based chips are still expected to make it to market, but Intel appears to have cancelled the versions aimed at data centers, desktops, and high-performance notebooks; only parts targeted at ultra-low power notebooks are expected to launch this year. Unfortunately, Intel missed out on a critical opportunity. Since it is likely that the 14nm-based products after Skylake were added to Intel's product plans very late in the game, Intel didn't have time to update key aspects of the chips. For example, Intel's third- and fourth-generation 14nm products, known as Kaby Lake and Coffee Lake, respectively, use the same basic CPU, graphics, and memory controller technologies that were found on the original Skylake processors.
The upshot is this: Intel could've delivered far more interesting third- and fourth-generation 14nm products to the market had they been planned well in advance. Fortunately, I strongly suspect that Intel won't make this mistake during its 10nm product generation.
Three generations, three new chips
It's common knowledge that Intel's first two 10nm-based products -- Cannon Lake and Ice Lake -- will both feature improvements to key subsystems such as graphics and CPU. Moreover, during the summer, I published proof that Intel's third-generation 10nm processor, known as Tiger Lake, will have a new graphics processor compared to the one in Ice Lake.
This tidbit of information is a pretty strong indication that Intel planned Tiger Lake far enough in advance that it could have its chip technology development teams make improvements to the key subsystems within the chip.
If I'm right -- and I think the odds are in my favor on this one -- then Intel should deliver solid improvements in Tiger Lake's capabilities compared to Ice Lake's. At a minimum, this should mean faster, more efficient CPU and graphics capabilities, but in today's world -- where computing tasks are increasingly being offloaded to specialized units -- it's often not enough just to make the existing technologies on the chip better.
Intel could very well add new technologies, such as a dedicated artificial intelligence processor to help make personal computers smarter without hurting battery life.
The point is this: While Intel will likely end up selling chips based on the same architecture for over three years (mid-2015 through the end of 2018) due to poor planning with respect to its transition to its 10nm technology, I don't think this phenomenon will repeat.