AMD continued to enjoy strong growth in Q2 2018

AMD released its second-quarter 2018 financial results yesterday evening. The company posted revenue of $1.76 billion, up 53% year-on-year, and made $153 million in operating income on that haul compared to a $1 million operating loss a year ago. Net income was $116 million, compared to a $42 million net loss a year ago, and gross margin rose three percentage points to 37%.

  Q2 2018 Q2 2017 Change
Revenue $1.76 billion $1.15 billion Up 53%
Operating income $153 million -$1 million Up $154 million
Net income $116 million -$42 million Up $158 million
Gross margin 37% 34% Up 3 percentage points

The Computing and Graphics business brought in $1.09 billion of revenue, a 64% year-on-year increase. AMD says that result comes from strong Radeon sales and the continued momentum of its Ryzen CPUs. Investors looking for signs of AMD's exposure to the cryptocurrency market will want to note that sequential revenue declined 3% from last quarter primarily because of a decline in crypto demand. The division made $117 million in operating profit for the quarter compared to $7 million a year ago.

While AMD's CPU average selling price declined in part thanks to a more competitive Intel lineup and the consequently lower suggested prices of some second-generation Ryzen CPUs, GPU ASPs increased thanks to products both for the consumer market and data center.

The Enterprise, Embedded, and Semi-Custom division raked in $670 million, up 37% year-on-year and 26% sequentially, thanks to increases in semi-custom and server product revenues. That sequential figure suggests Epyc products are beginning to have a small but material impact on the server CPU market. The division made $69 million in operating profit, too, compared to $16 million a year ago.

The All Other portion of AMD's balance sheet posted an operating loss of $33 million compared to $24 million a year ago.

For the next quarter, AMD expects revenue of about $1.7 billion, plus or minus $50 million. If that range proves correct, the company will enjoy 7% year-on-year growth for the quarter. AMD also expects gross margin to increase to about 38%, buoyed by Ryzen and Epyc CPUs and counterbalanced by lower GPU sales into the cryptocurrency market.

Comments closed
    • K-L-Waster
    • 1 year ago

    Can it be? An AMD earnings report *without* one time charges turning the ink from black to red?

    Y’know, it’s possible they may actually have a future in this whole semi-conductor business. Whoda thunk it?

    • jihadjoe
    • 1 year ago

    Great!

    Now all I need is for them to release a CPU on DDR5 coz I’m not planning on blowing any cash on DDR4. GamersNexus reported that AMD really wants to beat Intel to market with it so hopefully that’ll be next year with Zen2.

      • Puiucs
      • 1 year ago

      You will most likely have to wait for Zen 3 for that (or maybe Zen 2+ at the earliest, but highly unlikely).

    • Krogoth
    • 1 year ago

    2019 is going to be the year of AMD. They pretty much got open season on the Enterprise/SMB markets. Upcoming Epyc refresh (based on Ryzen+ architecture) is going to be very compelling for clients they are in an upgrade cycle.

    Intel only has Skylake-X family until Q1 2020 when Cascade Lake should land.

      • freebird
      • 1 year ago

      I definitely agree with the 1st sentence, even though that will probably get me down voted. The Cascade Lake family was stated to be Late 2018(read December), but probably Q1 or maybe even Q2 2019 for most servers to show up. Intel mentioned on their conference call that “…the next generation of Xeon will ramp in the early part of 2019.” I expect Intel to raise the TDP in the Enterprise to 250w to try and keep up with 7nm Epyc.

      Cascade probably won’t offer much of a clock speed increase without a TDP increase. It’s real claim to fame is 3DXP (Apache Pass) Support.

      Cascade is early 2019, Server Ice Lake (10nm) is planned 1Q 2020. Intel stated in earning call client 10nm Holiday 2019 and Server following shortly after…

      Who knows where the TDP limit may end in the HEDT. TR2 is upping the TDP limit to 250w, with motherboards supporting at least 300w or more. Intel realizes it is in a real fight now so I expect them to maybe even go 300w TDP.

      It is looking to be a very interesting year…2019

    • smilingcrow
    • 1 year ago

    If Intel’s massive struggles with 10nm are indicative of how difficult things may be generally below the 10/7nm barrier, AMD has the advantage of working with 2 Fabs.
    If I was Apple and considering dropping x86 for custom ARM cores the fact that Apple have access to multiple Fabs might be the tipping point more so than fear of Intel’s design stagnation.
    But with AMD also producing x86 chips and with them having access to 2 Fabs, by sticking with x86 Apple have access to at least 3 Fabs AND 2 CPU design teams competing against each other.
    So ARM for desktop makes less sense unless they see synergy between having MacOS and iOS on the same ISA.

      • blastdoor
      • 1 year ago

      I think the synergy that might matter more is the synergy of controlling the whole Mac stack, from silicon to OS. But for that synergy to matter, Apple would need to have a vision of what they want to accomplish with the Mac. That vision is a little murky. I could *imagine* the Mac becoming a high-end workstation for AI/ML and AR/VR development work, and there are hints of interest in that from Apple — but only hints.

      In the absence of a clear vision for what they would do with such silicon-software integration, I think you’re probably right — better to stick with x86.

      Edit:
      Synergy, synergy, synergy!

        • chuckula
        • 1 year ago

        This post needs to use the word synergy at least three more times before I synergistically upthumb.

          • Redocbew
          • 1 year ago

          I wonder if it’s even possible to synergistically downthumb. Synergy is always good, right?

            • smilingcrow
            • 1 year ago

            Sin + Energy = Trump! So all good, right?

        • ronch
        • 1 year ago

        Also, would designing their own CPUs really make a big difference? They may have succeeded in designing better ARM cores than stock ARM or Qualcomm cores but in x86 they are up against Intel, and Intel’s design teams are practically second to none.

        Also, depending on how Zen 2 shapes up against Intel they may just design the logic block that they need for whatever vision they want to accomplish and have AMD custom design the rest of the chip with Ryzen 2 or Navi or some other future CPU or GPU. AMD already has most of the Lego™ blocks.

          • Redocbew
          • 1 year ago

          If it’s hard to build an x86 CPU at 10nm, then it’ll be hard to build an ARM CPU at 10nm. Apples and Oranges.

            • smilingcrow
            • 1 year ago

            A significant difference now is that with AMD being competitive again Apple don’t have to choose between x86 from one Fab only, Intel’s, but also have access to the multiple Fabs that AMD use.
            So going ARM might well reduce the Fabs they have access to unless Intel offered to fab them for Apple.

    • kuttan
    • 1 year ago

    Chuckula’s decade of sponsored “AMD hate” efforts gone down the drain.

      • BurntMyBacon
      • 1 year ago

      Really? I thought he more or less achieved [i<]his[/i<] desired result.

        • kvndoom
        • 1 year ago

        Those who understand Chuck know that downvotes and upvotes have equal weight.

        No matter which thumb you click, he wins.

          • K-L-Waster
          • 1 year ago

          He’s only motivated by the ABS value…

            • chuckula
            • 1 year ago

            I even take it a step further with RMS values to emphasize the amplitude of the peaks.

      • chuckula
      • 1 year ago

      What are you talking about. I just made a bunch of money this morning.

      Incidentally, since you obviously have a 6 or 7 figure position in AMD, you shouldn’t have any trouble buying me a “failed” 9900K part if it is able to beat a 2700X in 20% of TR’s official benchmark suite.

      I mean, considering the 2700X is the single greatest achievement of AMD’s existence and the 9900K is a failure I think it’s pretty obvious that Intel can’t win 20% of the benchmarks.

        • Srsly_Bro
        • 1 year ago

        By switching to gecko insurance?

          • Shobai
          • 1 year ago

          Take it to the next level: [url=https://calgarysun.com/news/canada/alberta-man-legally-changes-sex-for-cheaper-car-insurance-i-didnt-feel-like-getting-screwed-over-any-more/wcm/6bb18269-c5a7-4e85-9c3e-4571df59d654?ref=hvper.com&utm_source=hvper.com&utm_medium=website<]Alberta man[/url<]

    • srg86
    • 1 year ago

    This happy Intel user is still happy to see this. A strong AMD pushes Intel to make better products.

    It would need AMD to use Intel memory controllers and chipsets for me to go back to buying AMD CPUs though. The last time that happened was the K6/Pentium MMX era.

      • MOSFET
      • 1 year ago

      Well, not trying to convince you, but Ryzen is SOC so there isn’t much in the chipset except extras, and have you seen Ryzen memory benchmarks? Even dual-channel at 2400 they are chart-topping compared to Intel’s dual-channel procs (AIDA64 anyway).

    • blastdoor
    • 1 year ago

    Meanwhile, Facebook is down about [b<]20%[/b<] this morning. The world has gone topsy turvy!

      • chuckula
      • 1 year ago

      You think that’s impressive? HOLD MY BEER!
      — Intel after the mining market collapses.

    • maxxcool
    • 1 year ago

    I hope so, they just got my money for a 16/32 core workstation .. enjoying it immensely thus far for log parsing and vm-tuning.

    • Unknown-Error
    • 1 year ago

    Keep it up team RED!

      • chuckula
      • 1 year ago

      Don’t say the word “red” when we’re talking about quarterly results!

      Too soon man, too soon.

        • BurntMyBacon
        • 1 year ago

        Fun fact: Fresh brewed tea is more painful in your nose than you might think.

          • ermo
          • 1 year ago

          You’re holding it wrong.

      • srg86
      • 1 year ago

      I remember when AMD were team green. They seemed to change to red after the ATI buyout/merger.

        • K-L-Waster
        • 1 year ago

        The GPU branding colours woulda been pretty confusing….

    • Chrispy_
    • 1 year ago

    I run an Intel-powered company for the most part.

    I have simplification and management reasons to stick with Intel, as well as good (lucrative/economic) relationships with some key Intel-biased hardware vendors.

    And yet, here I am buying EPYC and Ryzen. Despite antitrust and anti-competitive rulings against Intel it’s still an uphill struggle for them. For me to be buying them at a corporate level even with the huge disadvantage, they must be doing [i<]something[/i<] right! I just wish they'd hurry up with Navi!

      • techguy
      • 1 year ago

      I wish we had been able to go with Epyc in our latest round of server hardware refreshes. Sadly, our server vendor of choice (Dell) has yet to produce a blade design which offers the option to use Epyc. Hopefully by the time we’re ready to refresh again they will have remedied this situation. Given our 4 year refresh cycle I would certainly expect that to be the case.

        • Srsly_Bro
        • 1 year ago

        Should have dropped a “you better” on your boss or maybe forward him the letter to Intel to change his mind in the future.

    • chuckula
    • 1 year ago

    Profits are on the Ryze!

      • Fursdon
      • 1 year ago

      I see a profit margin a ryzen.

      • Srsly_Bro
      • 1 year ago

      #ryzeup

      • Ninjitsu
      • 1 year ago

      AMD truly is the house of the ryzen Su(n).

        • DragonDaddyBear
        • 1 year ago

        Intel may be bad company but it’s no reason to move to Paradise, NV.

    • ronch
    • 1 year ago

    First of all, now with the money coming in they should start designing the architecture that will eventually succeed the fundamental Zen design.

    Secondly, FIRST.

      • jovian_litany
      • 1 year ago

      Give them time, hopefully there is not as much hubris with Ms. Su as there was in the Jerry Sanders Era.

      • Tom Yum
      • 1 year ago

      I think you can certainly see increased R&D spending which is an important step towards that. Revenue is up $600mil but operating profit is only up $150 mil despite margin increasing, indicates substantial growth in operating costs. Nothing about AMDs business can easily explain that, except for R&D increases with a splash of higher marketing budget on the side. Both important for a sustainable business.

      • Srsly_Bro
      • 1 year ago

      Nice job, bro.

      • freebird
      • 1 year ago

      Zen 2, 3 and 5 are not enough for you?
      [url<]http://www.legitreviews.com/zen-5-talked-amd-ryzen-processor-tech-talk_204536[/url<] (Yes, Chuckula, all I was saying that they already are working on more... but unlike you, I don't have a baby about something when somebody posts a link for proof.) BTW, for those that didn't listen to the Earning call, Lisa Su was specifically asked about where 7nm Zen 2 (Rome) was manufactured : (this is the part that sent Chuckula into is tizzy) Question from Matthew D. Ramsay - Cowen & Co. LLC And I think some folks have asked a couple of questions around process node, and you're obviously sampling already with Rome. So maybe could be explicit about where you're manufacturing that? Lisa Su stated: "..so as I said earlier, we are working with both the TSMC and GLOBALFOUNDRIES in 7-nanometer. As for the 7-nanometer Rome that we're currently sampling, that's being manufactured at TSMC." So that shoots the hole in the theory that AMD will NOT use both TSMC & GF for 7nm Zen 2. At least as far as EPYC is concerned. 3rd edit: wow... some people didn't like THAT news... 3 down votes in 3 minutes...

        • chuckula
        • 1 year ago

        Yeah wait… you literally required a diaper change the other day because Intel didn’t provide some extremely specific timing data about Optane DIMMS that were being shown in a live public demonstration and that actually exist right now.

        You literally said they “CANNOT” tell us this information because those Optane DIMMs… that were literally running live on stage in working systems… are somehow not designed properly or some other FUD.

        But all AMD has to do is have some random guy name-drop the fact that AMD is working on a future CPU architecture and we’re all supposed to fall down worshipping?

        I’d be far more interested if AMD [b<]wasn't[/b<] working on some followup architectures.

          • freebird
          • 1 year ago

          Is Chuckula gonna Cry over my posts now…?

          How sad. 🙁

          Quote my WHOLE POST why don’t you…Captain FUD. I didn’t post any FUD, but you seem to like to make baseless accusations and misquotes. I linked to the Anandtech article where they listed all the questions asked of Intel and all the non-answers Intel gave.

          I questioned why Intel couldn’t answer some simple questions (they talked about shipping for revenue in 2018 in the article I linked) and then you decided to drag 7nm Vega 20 into it…

          Grow up.

            • Srsly_Bro
            • 1 year ago

            I think he requires another diaper change. Someone is a stinky boy.

          • freebird
          • 1 year ago

          Oh, I see you edited your comment to put more unspoken words in my post… So go ahead and quote some of the “extremely specific timing data” that I asked for…

          I only posted Questions asked by the audience you dufus.

            • chuckula
            • 1 year ago

            No, I edited my comment to remove a trivial spelling typo and the exact same comment is the same as the original.

            But since you are so enraged about Intel existing let’s make friendly bet: Your 16 core RyZen 2 had better not only beat an 8 core 9900K in TR’s gaming frame time analysis, but it better do so by a bigger margin than the 9900K — you know that chip with 3 year old cores — beats a 2018 RyZen 2700X.

            I’m actually being easy here. I was going to require AMD to provide an objectively superior AVX-512 implementation in RyZen 2 compared to Ice Lake but since AMD is generally always allowed to have the training-wheels on in order to have its products considered to be “miraculous” I made it easy on you.

            • freebird
            • 1 year ago

            No one here enraged, but you my friend…

            • chuckula
            • 1 year ago

            Good, then we’ll reinstate the requirement for RyZen 2 to have full AVX-512 support since you’re so confident.

            • NTMBK
            • 1 year ago

            This is a great example of MAD, or Mutually Assured Downvoting.

            Nobody wants your pissing match cluttering up the comments. Kindly both shut up and take it to personal messages.

            • freebird
            • 1 year ago

            Good thing Chuckula said I needed a diaper change that should cover all the pissing… 😀

            That’s why a upvoted your comment.

            • Srsly_Bro
            • 1 year ago

            I also suggested you needed one and it seems like you haven’t been changed yet.

            • freebird
            • 1 year ago

            Are you offering to do it? I may need adult care some time in the future.

        • derFunkenstein
        • 1 year ago

        It takes many moons to design and refine a CPU architecture. If they’ve been working on Zen 5 for a year (let’s just guess that) then they need to start NOW on whatever is going to replace it a year later. This isn’t rocket surgery.

          • BurntMyBacon
          • 1 year ago

          [quote=”derFunkenstein”<]This isn't rocket surgery.[/quote<] What university do I have to attend to learn that? (O_o) Also, what do they use for anesthetics?

            • psuedonymous
            • 1 year ago

            [quote<] Also, what do they use for anesthetics?[/quote<] Nothing, they just snip the nozzle extension right off and launch (as happened for COTS1)

            • ronch
            • 1 year ago

            University of Pyongyang.

          • freebird
          • 1 year ago

          Yeah, but you can’t get the cart too far before the horse either… some of their designs are based on where and what manufacturing process will be available 3/4 years down the road. I’m sure they have plenty of ideas, but only Zen 3 is slated for 2020 on 7nm+; everything after that is still up for grabs. Will GF have 3nm, when will they have it? They are planning on skipping 5nm. TSMC stated they will have 5nm & 3nm. GF is looking for a partner (at 3nm) to lower the cost burden.

          AMD is lucky in a sense… Lisa Su stated in the conference call that they bet heavily on 7nm, which seems to be working out. Intel looked like it was hoping 10nm and Ice Lake would be fixed by now or at least early 2019 (which would’ve been very competitive with Rome), because Cooper Lake crawled out of the lake like Swamp Creature to appear on the latest roadmap.

          [url<]https://www.youtube.com/watch?v=lSmVPxPwR7I[/url<]

        • Goty
        • 1 year ago

        People are downvoting you because ronch is talking about the successor to Zen as an architectural philosophy (for lack of a better word) whereas the nth iteration of Zen is still… well, still Zen.

          • freebird
          • 1 year ago

          That makes even less “cents”. How much money does Intel have and what iteration of CORE are they on??? 9th if the rumor is correct? Changing the base architecture of something that is working well does not always result in a return on investment. Just ask Itanium and Bulldozer how they are doing.

            • Goty
            • 1 year ago

            Intel is on the 9th iteration of this architecture and now they’re bound to process advances to improve performance in significant ways. I would think both companies would like to avoid that scenario in the future, especially considering the challenges currently facing the fabs with each successive process shrink.

            • freebird
            • 1 year ago

            uArch(s) don’t grow on a tree for a reason. Very costly to develop and validate. Much easier to get low hanging fruit of multiple generations with tweaks to different blocks (memory controllers, cache, branch prediction, etc.) Yes, a some point stuff like Quantum compute or some other radical change will be needed, but as we saw with Bulldozer uArch… it put AMD on life support for a few years.

            • Goty
            • 1 year ago

            Exactly, it’s an expensive, difficult, [i<]time-consuming[/i<] process, which is why ronch suggested that AMD should be working on one already.

            • freebird
            • 1 year ago

            Yeah they always have ideas, I’m sure they have ideas about nano-tubes and quantum computing, but they can’t pour a huge amount of R&D into it until the manufacturing tech proves promising either. So what is wrong with what I said… the Zen uArch JUST arrived. 3 or 4 iterations will care them to next one, which yes they are already thinking about.

            • Waco
            • 1 year ago

            Why would anyone think a totally new uArch is needed any time in the near future (say, within 5-8 years)?

            Now that they have a good base architecture they’ve got a looooong way to go before they need a new core from scratch. The next decade won’t be decided by who has the best core, it’ll be decided by everything around the core.

      • Waco
      • 1 year ago

      They’re deep into designing the future chips, they’re just not telling you about them. I’m not at liberty to say more than that, but no chip vendor of this size isn’t looking out 5-10 years for future architectures.

        • ronch
        • 1 year ago

        I’m sure they are already designing it and they’ll only start talking about it about 2 years before expected release. I’m just saying what is probably obvious to them, and excitingly mysterious to us.

          • Waco
          • 1 year ago

          Oh they absolutely are deep into it. Roadmaps into the 2025 range is the furthest I’ve seen, but they’re definitely planning out pretty far.

Pin It on Pinterest

Share This