In early 2017,chipmakerAdvanced Micro Devices(NASDAQ:AMD) released the first products based on its all-new Zen processor architecture. Zen represented a huge step up from the company’s prior processor cores, which were all derived from its ill-fated Bulldozer architecture. It delivered a huge increase in performance and power efficiency, allowing it to re-enter the high-performance desktop and data center processor markets.
One thing that really helped the Zen-based products to be as competitive as they are in the marketplace (particularly in the gaming desktop segment) is that the competition — in this case,Intel(NASDAQ:INTC) — was caught flat-footed.
Image source: Intel.
In 2015, Intel released a processor family known as Skylake. Intel’s original plan was to introduce a successor to Skylake, known as Cannon Lake, in 2016 and then the successor to Cannon Lake, known as Ice Lake, in 2017. In 2018, the successor to Ice Lake, now known to be called Tiger Lake, was expected to launch.
However, thanks to manufacturing struggles (Cannon Lake, Ice Lake, and Tiger Lake are all planned to be manufactured in the company’s highly delayed 10-nanometer technology) as well as poor planning on the part of Intel’s product groups, Cannon Lake, Ice Lake, and Tiger Lake are still nowhere to be found.
Instead, Intel is selling processors based on the very same Skylake architecture that debuted in 2015, with the only performance enhancements coming through tricks like adding additional processor cores to its chips as well as boosting operating frequencies.
Intel’s stagnation in processor architecture ultimately meant that AMD’s Zen processor core — which is roughly on par with Intel’s pre-Skylake processor architecture, known as Broadwell, in capabilities — came within spitting distance of Intel’s best. This, coupled with aggressive pricing and good products based around the Zen cores, allowed AMD to capture significant share in certain market segments previously dominated by Intel.
Recovery will be difficult for Intel
Over the long term, it will be difficult for Intel to open up anything close to the kind of competitive advantage that it had over AMD in the past. Not only has Intel’s processor innovation cadence been set back by years, but AMD won’t sit still.
AMD, for example, intends to release enhanced versions of its Zen processor cores in the years ahead, with the company having publicly committed to launching Zen 2 in 2019 and Zen 3 in 2020.
Intel, too, should finally start bringing new processor core innovations when it introduces its Ice Lake family of processors either late in 2018 or early in 2019, but at this point Intel and AMD should be innovating in lock-step.
The fact that Intel’s processor technology stagnated for several years allowed AMD’s Zen to enjoy a much stronger competitive positioning when it came to market than it would have if Intel had kept to its prior product introduction cadence. It could also mean that as Intel finally clears its product backlog and brings those chips to market, they merely serve to maintain the relatively small competitive edge that Intel currently has.
Another issue for Intel with respect to its competitive positioning relative to AMD is that AMD can rely on third-party chip manufacturing companies to build its chips. AMD’s main partner is Global Foundries, whose track record is admittedly spotty, but the company has indicated that it also plans to tap chip manufacturing leaderTaiwan Semiconductor Manufacturing Companyfor future chip production.
Should Intel’s manufacturing woes persist in the years ahead and should, at a bare minimum, TSMC continue its impressive track record of execution, AMD may actually have access to a more predictable road map of manufacturing technologies than Intel’s own chip designers do.
Ultimately, Intel’s continued manufacturing stumbles severely compromised the company’s ability to bring out new processor designs that improve performance, add features, and generally allow it to maintain the significant processor performance lead over AMD that it had enjoyed for over a decade.
Moreover, while it’s easy to blame Intel’s manufacturing group for stumbling as it has (and, believe me, I do), the company could have invested in bringing its new processor innovations to its currently shipping 14-nanometer+/14-nanometer++ technologies to try to maintain a clear lead in processor technology.
Instead, Intel seemingly waited until the last minute to build suitable back-up products to the delayed 10-nanometer chips, which likely forced it to reuse the aging Skylake architecture to minimize risk and ensure timely product launches. Intel’s product groups should have invested in more aggressive contingency plans.
Intel’s execution here has been nothing short of awful.