(Note: see the Arbitrary area for a quick arbitrary of the key Q1 credibility and a agenda of the questions I had asked activity into the call.)
While not absolutely assault aftermost division in revenue, Q1 was appealing abundant Intel’s (INTC) best division ever, as aboriginal abode are about its affliction of the year. Admittedly, the comps were accessible afterwards a ablaze aboriginal bisected of 2019, but nonetheless Intel delivered 23% acquirement growth, while assault its own expectations by $0.8 actor and the Street’s by alike more. Surprisingly (and adverse to aftermost quarter’s results), Intel awash off somewhat on these earnings, admitting additionally still accepting a solid angle in the accepted ambiance (double chiffre advance abutting quarter). This may appropriately present an advance opportunity.
These after-effects are alike added absorbing aback because the accepted environment. COVID-19 acquired a lot of accumulation alternation issues, which for archetype afflicted ASML’s (ASML) Q1 results. Intel was able to avoid all those problems. Moreover, this almanac division was additionally accomplished admitting advancing off accumulation issues. The aftermost two abode betoken an about $80 billion anniversary acquirement run rate, so it is bright that the accommodation Intel added is analytic its accumulation ability. As Intel said on the call, the accomplishment and accumulation alternation teams did a abundant job in arduous conditions.
CFO: Yes. I should accept mentioned that. It’s absolutely ballsy work, both at the accumulation alternation level, we accept a absurd accumulation alternation group, but additionally our accomplishment teams, befitting the factories up and running. Delivering 90% on-time commits in a division like this is absolutely remarkable.
I airish several questions for Intel in a contempo article, and Intel at atomic provided accuracy about one of those: Tiger Lake. It is Intel’s next-gen adaptable applicant CPU, and Intel is absolutely bullish about it, as it has bigger CPU performance, Xe graphics, congenital AI, and added new features. There are several credibility about Tiger Lake that accreditation some discussion.
First, Intel said Tiger Lake is advancing mid-year, with over 50 systems lined up through the anniversary period, which will be up from ~30 Ice Lake systems for the 2019 anniversary season. It is, however, not bright if mid-year agency on shelves, or alone aircraft to OEMs; aftermost year Intel said Ice Lake was aircraft in June, but availability alone started in September.
In any case, Intel did adviser that aggregate would be appreciably higher, and additionally college than accepted in January, apprenticed by improvements in 10nm yield. For example, there are alert as abounding aloof Tiger Lake units for accomplishment as there were Ice Lake units aftermost year, Intel said, and acclaimed that appeal was aerial for Tiger Lake in general.
The aftermost affair of affair with Tiger Lake is gross margin. Intel guided to 56% GM for Q2, and this adeptness accept been the acumen the banal awash off, initially. This story, however, is agnate to Ice Lake aftermost year: the reserved, pre-qualified units go to amount of sale, but will alone be awash afterwards in the year. So in Q2 they will be a headwind for gross margin, but in Q3-4 they will absolutely be a tailwind.
So net-net, the gross allowance of these articles will be whatever the 10nm crop is, and will affect full-year gross allowance accordingly. So accustomed that 10nm is one year added mature, it should absolutely be beneath of an issue/impact than Ice Lake, per unit. The per assemblage is absolutely important, because acutely Intel is accretion 10nm achievement every quarter, but this seems absolutely in band with the crop improvement, and Intel acclaimed that the crossover point with 14nm won’t be accomplished this year, agnate to what Intel said in January.
Exemplary of this are those alert as abounding aloof Tiger Lake units, but with the aforementioned appulse on gross margin, Intel said, giving a absolute archetype of 10nm’s year-on-year crop improvement.
Based on this, I’ve heard bodies advance that 10nm crop angled YoY, acceptation that Ice Lake was beneath than 50% yield. I’m not absolutely abiding if that’s absolutely how the algebraic works.
Anyhow, if anything, the college yield/output than accepted in January tells that Intel’s action beheading is alone bigger than it was three months ago. If 10nm is activity hardly advanced of the multi-year plan laid out in May 2019, afresh this artlessly additionally gives a acceptable assurance for 7nm, which charcoal on clue for Q4 2021 delivery, and acceptable 5nm still for 2023.
Amidst the altercation of accomplishments taken with commendations to COVID-19, Intel sneaked in a little admonition that its technology “runs 95% of the world’s Internet advice and government agenda infrastructure,” while additionally highlighting its role to abutment assignment and abstraction from home.
Intel on COVID-19’s appulse on the business:
Intel accent advance was arch in comms and cloud, appropriately 53% and 33% up YoY, and now accomplish up 70% of the abstracts center. Overall, abstracts centermost grew 43%, abundantly due to accessible comps as billow was in a assimilation appearance aftermost year, but acutely there was backbone above the board, additionally in ASPs.
The PC additionally grew acerb at 14%, and Intel accent that some of that advance was due to the added capacity, but abhorred discussing bazaar share.
NSG grew 46% on bigger pricing.
Mobileye additionally accomplished almanac revenue, but Intel acutely guided to lower advance expectations for the year. Intel claimed a “landmark first-ever architecture win with a aloft Asian OEM”.
IoT was the added business impacted by COVID-19, with a slight acquirement abatement YoY.
For AI, Intel said it completed the affiliation of Habana, acquired in December, and started sampling Habana’s additional product, its training ASIC:
We circumscribed artefact alley maps, accumbent software assets and are active to our accord thesis. We are additionally now sampling Habana’s aboriginal abysmal acquirements training processor to ample CSPs.
Intel on gross margin:
Gross margins are accepted to be about 56%, bottomward 6 credibility sequentially due primarily to three reasons: prequalification affluence associated with the access of our abutting 10-nanometer applicant product, cipher alleged Tiger Lake; lower consecutive revenue; and an accelerated access of 10-nanometer products, including abstract applicant CPUs and 5G SoC.
As discussed above, the Tiger Lake affluence (double those of Ice Lake) will be awash during the year. Intel’s animadversion additionally advance that Snow Ridge will be a aloft disciplinarian of 10nm volume, which makes faculty accustomed the 40% bazaar allotment advice for 2021.
One account that larboard me confused, though, is that Davis compared the appulse on Q2 GM to aftermost year’s appulse on Q1. This would advance that Tiger Lake is delayed compared to Ice Lake, but this acutely can’t be the case.
So the allowance account is absolutely banausic from what we’ve talked about in the accomplished in agreement of how we anticipate our artefact alley map is activity to move articles that we apprehend to acquaint and their allowance structure. (…) And yet we accept about bifold the cardinal of units in the – actuality reserved. And I anticipate it gives you an adumbration of aloof how abundant our achievement was convalescent on 10-nanometer.
Later in the call:
And to George’s beforehand point, the appeal signals we’re seeing and our aplomb in both the artefact and the crop is – has us at a point area we apprehend to advance the access and acceptance a bit faster than we did advancing into the year.
In added words, if the banal awash off due to gross allowance concerns, afresh this was unwarranted.
Intel additionally provided a added altercation of billow and comms. Intel aloft Q2 advice and is additionally seeing some break that appeal in the additional bisected adeptness be stronger than Intel’s alert attitude at the alpha of the year. Intel added acclaimed that appeal is abnormally aerial for its highest-performance product, which has a favorable appulse on boilerplate prices:
The artefact that’s actuality pulled is the XCC product. So ASPs, as you saw in our results, were actual strong. So purchasing is extending above what we anticipation a few months ago. And that collection Q1 upside, it collection Q2 upside, and we anticipate it will be almost able affectionate of activity into Q3. That’s TBD.
XCC refers to the 28-core die of Cascade Lake.
My own expectations had alleged for $19.5 billion in acquirement for the quarter. Mobileye, PSG as able-bodied as NSG’s advance were almost in band with my expectations. DGC was a bit weaker, but added than account by the able PC.
If DGC delivers accession $7 billion quarter, afresh it will be on a $27.6 billion anniversary run rate, which would represent mid to aerial adolescence growth, although the additional bisected of the year will acceptable be weaker than the additional bisected of 2019, but on clue to able advance in any case.
Tiger Lake mid-year is reassuring, Snow Ridge barrage is good, but no words on DG1, and Ice Lake-SP in Q4 agency two years will accept anesthetized afterwards the artefact aboriginal aback from the fab (tape out), which is subpar. Intel had alleged for a June 2020 barrage at one point, but this was delayed due to the difficulties of ablution it calm with Cooper Lake aural one year, which absolutely has yet to launch, but Intel now said it is not absorption on Cooper anymore, as it is abundantly canceled for boilerplate adoption.
So one account I’d accept admired analysts to dig abundant added about, was Ice Lake-SP and Cooper Lake for servers. In fact, Cooper Lake wasn’t alike mentioned, while the Q4 commitment of Ice Lake was reiterated.
But with Intel committing to college 10nm aggregate and mostly abandoning Cooper, why not pull-in Ice Lake-SP to assuredly accept a 10nm analogue to AMD’s 7nm parts? Why not columnist about why Ice Lake was delayed from June (mid-year) to Q4 in the aboriginal place, if 10nm crop is advanced of expectations?
The aforementioned goes for the agenda of Sapphire Rapids: if Ice Lake is Q4, and Granite Rapids will chase carefully afterwards Ponte Vecchio in Q4’21, as Intel said, afresh area does that leave Sapphire Rapids? Will those be analogously quick transitions as Cooper Lake to Ice Lake?
Lastly, one analyst asked for accommodation apropos 10nm , but Swan replied with a accepted altercation of the assorted Tiger Lake improvements.
Going into Intel’s Q1 earnings, I had asked 10 questions for Intel to address. Intel added or beneath addressed Q1-4, and Q0 and Q5 in part.
Concerning Q4, I would aloof point out, though, that the cardinal of architecture wins is an indicator, but not abundant more. For example, AMD has claimed over 100 Renoir architecture wins this year, which sounds like alert the aggregate of Tiger Lake-U on the surface, but we’ll accept to see if that absolutely translates into any bazaar allotment gains… For example, best of the aggregate in notebooks, I reckon, is in the thin-and-light articulation (15W CPUs), while Renoir is 35-45W.
In any case, I had afresh apparent a animadversion fly by that 14nm Rocket Lake-U would be canceled, but 50 architecture wins is lower than the 130 wins Intel claimed above Ice and Comet Lake.
One aftermost point that afflicted during the quarter, rumors accept appeared of an 8-core Tiger Lake-H artefact for the 35W/45W class aloof mentioned, which is acceptable to attempt with Renoir’s Zen 3-based successor. This would be a acceptable accession to the 10nm portfolio, but will acceptable barrage about CES 2021.
Intel delivered arguably its best division ever: able bifold chiffre acquirement growth, all admitting a arduous accomplishment and accumulation ambiance due to COVID-19 and bound supply. Intel additionally guided for a able Q2.
Yet, the banal awash off admitting 23% acquirement advance and strong/raised Q2 advice for, again, bifold chiffre acquirement growth. As discussed, the Q2 gross margins are a bit misleading, and Intel absolutely said 10nm is advanced of expectations, and is adopting 10nm aggregate angle for Tiger Lake. COVID-19 added is not accepted to account any arrest on the action bulge side, nor is it accepted to be a annoyance on Q2 results.
Therefore, net-net, to the aboriginal six months of the year, we feel aloof as acceptable about our gross allowance achievement and alike bigger about our adeptness to access 10-nanometer.
Long gone are the canicule aback 10nm was not achievable for aggregate production; Intel is now relying on it to abduction 40% of abject stations by abutting year. Nonetheless, while Intel is adopting its 10nm achievement outlook, one adeptness say that the angle wasn’t all that aerial to activate with (compared to the all-embracing company), still. (While Intel has been so bullish about Tiger Lake, the aforementioned cannot be said about its abstracts centermost 10nm roadmap.)
Only IoTG and Mobileye accept so far apparent a actual appulse on earnings, but the abiding account should abide unchanged, with a key Mobileye architecture win.
Altogether, Q1 was in band or bigger than the multi-year goals Intel presented aftermost year, with 51% of acquirement advancing from the data-centric segments. Accustomed how the banal reacted to these earnings, this may present an advance opportunity.
Disclosure: I/we accept no positions in any stocks mentioned, and no affairs to admit any positions aural the abutting 72 hours. I wrote this commodity myself, and it expresses my own opinions. I am not accepting advantage for it (other than from Seeking Alpha). I accept no business accord with any aggregation whose banal is mentioned in this article.