Every successive processor generation presents its own challenges to all chip makers, and the ramp of 14 nanometer processes that will be used in the future “ #Skylake ” Xeon processors, due in the second half of this year, cut into the operating profits of its Data Center Group in the final quarter of 2016. #Intel also apparently had an issue with one of its chip lines – it did not say if it was a Xeon or Xeon Phi, or detail what that issue was – that needed to be fixed and that hurt Data Center Group’s middle line, too. Still, despite a slowdown in spending among enterprises for new compute capacity, Data Center Group still turned in a record year for revenues and profits, thanks to growth in #cloud, #hyperscaler, #telecom, and service provider spending on compute capacity, even if the growth rate was only about half of the 15 percent sustained target rate that the company set for the Data Center Group a few years back. With Intel having such dominant share of the server market – Xeon and Xeon Phi chips commanded 89.2 percent of server revenues and 99.3 percent of server shipments in the third quarter of 2016 – it is no surprise at all that Intel now rises and falls directly with what is going on among large enterprises, cloud builders, hyperscaler, telcos and other service providers, and small and medium businesses. There just is not that much growth that can come from taking share away from #IBM System z mainframe and Power Systems and the remaining bases of Itanium, Sparc, Sparc64, Alpha, and other processors and the ARM server base is still nascent and is no opportunity for Intel to get new sales from. In the quarter ended in December, Intel’s overall sales were $16.37 billion across all of its product lines, up 9.8 percent from the year-ago period; net income was $3.56 billion, down 1.4 percent. This drop in earnings was not just due to the 14 nanometer ramp with the Skylake Xeons, but also due the ramp of #3DNAND flash memory chips as well as for Optane 3D XPoint memory for SSDs. There were also some costs related to restructuring charges for the layoffs of 15,000 employees that Intel announced last summer as it moved investments away from its Client Computing Group, which makes chips for PCs, laptops, and tablets, and towards new technologies such as 3D NAND flash, 3D XPoint memory, silicon photonics, rack-scale computing, and Omni-Path switching. Intel’s bottom line was squeezed by the ramp of 10 nanometer processes for “Kaby Lake” Core processors, due at the end of this year and their follow-ons, “Cannon Lake,” which will ultimately be re-etched as Xeon server chips, and the work Intel is doing on 7 nanometer chip manufacturing. And finally, Data Center Group’s profits were also impacted by an unspecified intellectual property cross-licensing and patent deal that Intel did in the quarter with an unnamed communications player. In the final quarter of last year, revenue for platforms at Intel in the Data Center Group – meaning processors, chipsets, motherboards and in some cases systems – came to $4.31 billion, up 7.3 percent, while Other revenue (which is not specified but which should include Omni-Path networking, RackScale architecture, and other stuff) amounted to $362 million, up 23.1 percent. But because of all of these issues above, operating income for Data Center Group fell by 13.5 percent to $1.88 billion. It is a bit strange that Intel did not announce the intellectual property deal when it happened, but the size of the deal may not have been, in and of itself, material to the company’s financials. As for the chip issue in the Data Center Group, Intel’s chief financial officer, Bob Swan, did not elaborate, except to say that this had a greater impact on the numbers and that there were higher than expected failure rates on unspecified components shipped to some high-end customers and that it had found a fix for the issue and set up a reserve to cover the costs for replacement parts. Neither of these issues, said Swan, would affect the books in 2017. The big issue for Intel in 2016 was the shift from the 22 nanometer “Haswell” Xeon processors to the 14 nanometer “Broadwell” parts, which provide a bump in performance and price/performance but which also, at least at first, incur higher costs as the manufacturing process ramps. The real problem is that enterprises, which now account for less than half of the revenues within Data Center Group, are cutting back on spending, and this might be a little disconcerting but it is absolutely predictable given the current global political and economic climate and given that enterprises are moving more and more workloads to public clouds or using services from hyperscalers and cloud builders where they might have otherwise built them and hosted them – and much less efficiently – in their own datacenters. Sales of chippery to cloud builders (which includes what we at The Next Platform call hyperscalers as well as public cloud companies) rose by 24 percent in Q4, according to Swan, and telecommunications and service provider companies both grew at close to 30 percent in the period as well But sales of products that ended up in enterprises or government agencies both fell by 7 percent in the period. The growth on one side is enough to fill in the gap on the other, but it is not enough to keep Intel at that target of 15 percent sustained growth for Data Center Group.
https://www.nextplatform.com/2017/01/27/skylake-xeon-ramp-cuts-intels-datacenter-profits/
No comments:
Post a Comment