Intel Supply in Q4: “Output Capacity up, Supply-Demand Still High”by Dr. Ian Cutress on September 26, 2019 5:05 AM EST
Recent reports from media outlet Digitimes have stated that Intel’s high demand issue, and its inability to fully supply that demand, will continue into Q4. The report states that Intel’s OEM partners and the supply chain are still short of the supply they are requesting from Intel and its distribution network. We reached out to Intel for comments.
Intel’s high demand has been of particular note since mid-2018. Since the discovery of hardware vulnerabilities such as Spectre and Meltdown, and the fixes that reduced overall performance of a large number of the installed server base, many of Intel customers have been increasing the size of their server deployments in order to re-match their original capacity. This issue caused a sharp up-tick in demand of Intel processors, and Intel has driven newer architectures that try to minimise those performance deficits (with an overall performance uplift when the new architecture is factored in). As a result, Intel moved some of its fabrication capacity away from its future 10nm process and back onto its 14nm in order to meet demand.
The consequence of this is record revenues for Intel – the company shifted a lot of production into its high core count and high-cost parts. The CEO of Intel, Bob Swan, addressed the issue in its recent financial update:
“We're also making steady progress increasing CPU supply. Through our investments, focused execution and tighter customer collaboration, we expect our PC CPU supply will be up mid single-digits this year while we expect the PC TAM to grow slightly. We'll continue to work with our customers to meet their required product mix and ramp additional capacity to ensure we are not a constraint on their growth.  We lost a little bit of share in the second quarter, particularly in CSG at the lower end small core primarily due to supply constraints. So -- and our expectation is that we'll begin to work our way back in the second half of the year given the capacity we've put in place to have more supply and meet our customers' demands.”
The recent report from Digitimes seems to put some cold water on Intel being able to meet all of its 14nm CPU demand by the end of 2019. With the recent launch of Comet Lake in the mobile space, Cascade Lake in the enterprise and workstation space, and the future launches of similar product lines, users and Intel’s partners are expecting a strong supply of CPUs with the generational update.
We reached out to Intel to see if there was any update to Digitimes’ commentary. An Intel spokesperson stated:
“We continue working to improve the supply-demand balance for our PC customers. In the first half of 2019, we saw PC customer demand that exceeded our expectations and surpassed third-party forecasts. We have added 14nm output capacity and are ramping volume on 10nm with systems on shelf for holiday. While our output capacity is increasing, we remain in a challenging supply-demand environment in our PC-centric business. We are actively working to address this challenge, and we continue to prioritize available output toward the newest generation Intel® Core™ i5, i7 and i9 products that support our customers’ high-growth segments.”
This essentially partially confirms Digitimes’ report. While Intel is increasing its output capacity, with a focus on 10nm as well (Intel reports two fabs on 10nm at this time), the focus on 14nm will be on high-growth segments, which for Q4 is likely to remain the high-end processors. Partners looking at Core i3 and lower performance cores might have to extend their lead-times for CPU supply yet again. It will be interesting to see when Intel will be able to re-reach parity between demand and supply.
Post Your CommentPlease log in or sign up to comment.
View All Comments
Dolda2000 - Thursday, September 26, 2019 - linkWhen making flawed products makes the demand for them increase rather than decrease, can anyone really doubt Intel's monopoly position? Sure AMD seems to have good products these days, but realistic alternatives to x86 can't come soon enough.
brakdoo - Thursday, September 26, 2019 - linkThis has probably little to do with Intel's or AMD's supply planning or product quality.
Many OEMs are building inventory before the next rounds of tariffs are in place. They need to get those chips to china ASAP so that they build laptops and servers and ship them before dec 15th. Only a small portion of the PC/Server supply chain has been moved out of china.
This kind of demand is unpredictable and unlike last year it's not their fault (unless you expect them to oversupply the market when there is regular demand).
BurntMyBacon - Thursday, September 26, 2019 - linkDemand for AMD products is particularly high right now AND Intel is pulling record revenues. I don't think the nominal market size is expanding at a sufficient rate to support both. The overstocking prior to tariffs you mention seems to be the most reasonable theory.
lemans24 - Thursday, September 26, 2019 - linkPlease reply with figures that show that AMD is now pulling record figures because last quarter they were down in revenue globally, which to me is the only revenue worth looking at
Intel999 - Thursday, September 26, 2019 - linkAMD's first half of 2019 revenues were driven down by the collapse of the crypto boom and the final year of console sales. AMD suffered the disappearance of, at least, $600 million in consumer GPU and chips for console through June 30th.
At the same time their CPU sales have been running 20 - 30% over 2018 levels in Q1 and Q2. This is the only market that Intel and AMD currently compete. Intel's desktop units were down 7% and server units were down 10% in the most recent quarter. Laptop was up 1% and those volumes probably made up for the 7% drop in the much smaller desktop segment.
The record figures for AMD start in the current quarter where the 3000 series of Ryzen and their first material consumer GPU launch took place. AMD is expected to post between $1.85 and $2.0 billion in revenues.
zepi - Wednesday, October 2, 2019 - linkAMD:s own guidance released in July was $1.8B:
"For the third quarter of 2019, AMD expects revenue to be approximately $1.8 billion, plus or minus $50 million"
jgraham11 - Thursday, September 26, 2019 - linkwell actually probably, tech server companies are disabling, HyperThreading on Intel servers because of the currently known CPU vulnerabilities (Meltdown, Spectre variant 1, 2, 3, and 4, Zombieload, Foreshadow, etc), but also the ones they still don't know. Because of loosing half the performance that you originally purchased, Microsoft, Amazon, Google now have to buy twice the processing power just to break even with what they previously had.
PeachNCream - Thursday, September 26, 2019 - linkFrom what I recall reading, hyperthreading doesn't result in a doubling of performance. Performance benefits are workload dependent and in some cases there is regression when its enabled. Shutting it down won't result in the loss of 50% of a given processor's output.
abufrejoval - Friday, September 27, 2019 - linkHT only ever yields 10-20% increased speed a) when there were sufficient threads/tasks ready to run b) no real CPU cores available for scheduling instead.
With server utilization rarely ever near 100%, the difference for lack of hyperthreading won't be anywhere near 50%, unless you're talking to the hyperscaler's lawyers trying to sue Intel for damages.
Intel likely oversold the potential of HT and it's probably only fair that it should bite them when the lawyers now oversell the damage of disabling it.
But the technically savvy shouldn't be misled quite so easily.
Maxiking - Thursday, September 26, 2019 - linkBy faulty you mean stating your product can reach certain frequency and then it does who it can't like in the AMDs case?