It looks like Intel is playing favorites with Skylake Xeons in a way that will shatter the industry status quo. In SemiAccurate’s opinion this move is going to inflict a deep wound in OEMs and top tier customer relations and hurt Intel badly in the mid- and long term.
Lets say this straight up front, SemiAccurate thinks Intel’s latest move is going to cause them irreparable damage in public image, customer relations, and long term sales. What are they doing? Several trusted sources say that later today, likely at Supercomputing 16, the company will announce they have pulled in Purley aka Skylake-EP Xeons, to this year and will sell them to two key customers. So far this sounds like good news, next generation cash-cow server CPUs early is a positive thing.
The problems is the two key customers bit, Google and Facebook. Why is this a problem? They are the only ones getting them. If this sounds like a supply/ramping problem, it isn’t, they are the only ones allowed to buy them until launch in Q2/2017. To put the cherry on top of this blunder, some of the sources said that the non-favored two only found out they were shut out very recently with emphasis on very.
New generation server parts significantly reduce operating costs and without a huge price rise, they will also drop TCO more than a little. Since TCO is the only metric that these mega-providers care about, a new server generation is a must-have upgrade, the more of them they can get, the sooner they can get them, and the sooner they can put them in racks, the more profit the company makes. Server TCO is the largest slice of their costs by far, when you run many millions of units, a few dollars a month per adds up quick.
By now you likely get the problem. Google and Facebook were just handed a massive TCO advantage over their competition so they are probably jumping for joy. Amazon, Tencent, Alibaba, Baidu, and the other top players are much less chuffed by this selective blessing. Intel in effect just arbitrarily handed their bitter enemies about the single largest cost advantage in the game and gave them exclusivity for six or so months. Then there are the server vendors themselves who can’t get the best parts, Lenovo, Dell, and HPE among other are probably going to be unhappy too, but likely less so than those competing with Google and Facebook directly.
This is a torches and pitchforks moment people, to say this won’t end well is an understatement. SemiAccurate thinks this move by Intel will mark a profound inflection point in the semiconductor industry as a whole. Lets look at that in a bit more detail, specifically the short, medium, and long term effect, they are very different but none of them are good for Intel.
Note: The following is analysis for professional level subscribers only.
Disclosures: Charlie Demerjian and Stone Arch Networking Services, Inc. have no consulting relationships, investment relationships, or hold any investment positions with any of the companies mentioned in this report.
Latest posts by Charlie Demerjian (see all)
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018
- Intel can’t supply 14nm Xeons, HPE directly recommends AMD Epyc - Sep 7, 2018
- AMD reintroduces the Athlon name with two CPUs - Sep 6, 2018