Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
Intel Businesses The Almighty Buck

Why Intel's Stock Just Dropped 10% After Reporting Earnings (techcrunch.com) 57

Intel shares are off around 10% in after-hours trading after the chip company reported its Q3 data. TechCrunch explains why: Investors had expected Intel to report an adjusted $1.11 in per-share profit, off around 22% from the year-ago period. They also expected it to report revenues of $18.26 billion in Q3, down a more modest 5% compared to the year-ago Q3. Notably, Intel beat revenue expectations with top line of $18.3 billion, and met earnings-per-share estimates of $1.11, on an adjusted basis. So, why are Intel shares sharply lower?

Quick consensus appears to point to weakness in the company data-focused business unit, the smaller of Intel's two halves (the other focuses on PC chips). Inside the data-side of Intel, its Data Center Group (DCG) had mixed results, including cloud revenue growth of 15%. However, at the same time, the DCG's "Enterprise & Government" business shrank 47% compared to the year-ago period, following what Intel described as "two quarters of more than 30 percent growth." Off that weakness, the resulting top line miss was sharp, with the market expecting $6.22 billion in revenue and DCG only delivering $5.9 billion.

Intel blamed COVID-19 for the weak economics conditions at play in the result. The company also highlighted COVID-19 when it discussed results from its internet of things business and memory operation, which declined 33% and 11% on a year-over-year basis, respectively. Perhaps due to COVID-19's recent resurgence in both North America and Europe, investors are concerned that the macroeconomic issues harming Intel's growth could continue. If so, growth could be negative for a longer period than anticipated. That perspective could have led to some selling of Intel's equity after the earnings report.

This discussion has been archived. No new comments can be posted.

Why Intel's Stock Just Dropped 10% After Reporting Earnings

Comments Filter:
  • i am shorting everything and move onto bitcoin - the monetary singularity
  • Since they're eating Intel's lunch on a lot of fronts right now.

    Or maybe it was Intel selling their NAND business to SK Hynix for 9 billion...

    • You'd think with every organization who used to have desktops frantically moving to laptops and having people work from home, there'd be a lot of chip sales out there. Like you say, AMD must really be eating their lunch for COVID to be a net negative for them.

      • by Junta ( 36770 ) on Friday October 23, 2020 @07:12AM (#60639306)

        The fact that they met overall estimates despite a huge gap on the server side speaks to exactly that: a surprise boost in laptop side chip sales but a drop in server side sales.

        Refreshing technology to better handle work from home was a big boost, but offset by a big weakness in selling server chips. Intel points to COVID-19 and to some extent they have a point *on the server side*, however I am seeing a lot of EPYC server sales. Intel's technical advantages over AMD are very niche areas (very large multi-socket systems, AVX512 (but barely enough to offset disadvantages), Optane DIMMs) and AMD has a much more straightforward set of advantages in general.

      • I'm not sure how rapidly AMD has made marketshare gains in absolute terms (though there are a lot more AMD-based systems showing up from vendors that historically either didn't bother with them or used them only for slum-tier systems that they designed and configured in ways that practically told you they didn't really want to sell them); but if nothing else the situation for Xeon margins must have shifted for the worse, and fast.

        Intel hasn't, historically, been shy about exploiting the fact that Xeons w
        • by raymorris ( 2726007 ) on Friday October 23, 2020 @09:42AM (#60639838) Journal

          Certainly you have a point about very high end systems, particularly single-threaded.

          That said, I find your user of the term "server grade" interesting. Servers are of course defined by the fact that they serve many users. Which makes them a natural fit for many cores. AMD has been in the lead for multi-core for a long time.

          For example I think it was 2012 when AMD launched the 8-core 3.6 GHz FX-8150 for $245. It wasn't until two years later that Intel had an 8-core CPU, for which they charged $999. Which meant that for a given budget, you could still get 16-32 AMD cores for the price of 8 Intel cores. For our servers, we needed to serve many users, not serve a single user with speed 1000X higher than the network can handle anyway. So for servers, lots of cores serving lots of users was what we needed, not one or two very fast cores.

          • by Junta ( 36770 )

            Probably best not to harken back to the Bulldozer days. In many respects Bulldozer was analogous to Netburst. Where Netburst was crazy high clock count (for the time) but crappy performance, Bulldozer was crazy high core count but crappy performance.

            Basically AMD started doing multicore and many other things as Intel was not yet out of Netburst hell. Then the vendors swapped, with Intel coming back to a reasonable architecture at more modest clock frequency and AMD going in on core count to the point of a c

            • You probably know more about different generations of processors than I do. All I know is:

              Our web servers, file servers etc are parallel workloads by their very nature. They want more cores, to serve more clients.

              The restriction is budget, performance per dollar, not the speed at which we could serve only one client, if there was only one client (not single thread performance). Single thread performance never mattered because a server was never pegged out by a single client, the issue was hundreds to thou

              • by Junta ( 36770 )

                Bulldozer's problem was that their concept of '2' cores were a hybrid, with some things shared and really single core and some things not, without very good mechanisms for the OSes or applications to be aware when they would actually be distinct or conflict with what *appeared* to be on a separate core. So the '8-core' was really something between a 4-core and 8-core, and a lot of workload basically had to ignore half the cores. More substantial than SMT/Hyperthreading, but not quite right to call them full

                • Thanks.

                  I remember at one time is was the other way around - Intel's multi-core shared a lot than AMD's did. So AMD was more like having X separate CPUs. I haven't had reason to keep up, though.

                  • A lot of the shared stuff was things like the FPU. I would guess things like serving up files and web pages, the cores on the Bulldozer CPU's were able to stay pretty independent of each other. I was things like games, graphics, video encoding where Bulldozer really started to fall apart.

                    • Yeah, bulldozers are great for highly parallel integer loads. I'm about to retire a 2P bulldozer machine for a single EPYC soon (although maybe we'll just go with a Threadripper, since I don't really need ECC for this thing). But honestly, the performance of the bulldozer is fine for what we do, it's just power hungry compared to the new stuff.

          • Main thing I noticed doing some limited performance testing at a web host (that had just started using Bulldozer Opterons (6200), but still had a sizable number of Phenom II era Opterons (6100):

            Each CPU package represents two NUMA nodes. Dual socket box has 4 nodes. Each node has two memory channels, so a minimum of 8 DIMMs are required for optimal memory configuration.

            Make sure all nodes have at least one DIMM installed, or you needlessly clog your HT links servicing memory access to the RAMless nodes.
            • Thanks for that info. That's useful and interesting.

              It may be noob level for a professional who specializes in nothing but architecting assembly of nodes. For those who not only spec the hardware, but are also the chief sysadmin, programmer, CISO, and CEO, that's something we could easily miss. It sounds like it can make a significant difference in performance.

              > This was around 8 years ago, and I don't believe that issue is still relevant.

              The MySQL and interleaving stuff isn't relevant, or I can forget

              • I believe they resolved the specific bug that caused MySQL to grind to a halt, but NUMA machines do still have certain design considerations that might warrant hiding the actual layout from the operating system - especially if the OS isn't apt to intelligently handle assigning processes to the nodes (i.e. treating it as any other generic multiprocessor system). I haven't dealt with this in recent history, but I suspect that it's a much smaller issue due to the efforts of developers to support machines with
                • A couple of articles on the MySQL NUMA issues - appears to have been partially related to how Linux handles paging out data under memory pressure, when a single node in a NUMA machine runs out.

                  https://blog.jcole.us/2010/09/... [jcole.us]

                  https://blog.jcole.us/2012/04/... [jcole.us]
                • > describes a machine in which the path between processing elements and memory are not consistent in bandwidth and latency (in contrast to a machine where memory access is entirely arbitrated by the chipset)

                  For anyone else reading this whose heart rate ticks up when they read these generalized terms, what it means is:

                  Each CPU core is connected directly to some RAM stick(s) / chips. It's fast when accessing that part of RAM.

                  The core is slow when accessing RAM assigned to a different core.

    • Comment removed (Score:5, Interesting)

      by account_deleted ( 4530225 ) on Friday October 23, 2020 @04:29AM (#60639014)
      Comment removed based on user account deletion
      • Without a new arch, intel is screwed. Without a new process, intel is screwed.

        It's spectacularly unlikely that they will pull both out of a hat at this point.

        Intel is screwed.

        Couldn't happen to a nicer bunch of frauds.

        • I don't know why any American would be happy to see Intel flounder. This is (was) a great American company. This is our lead in a vital industry going down the toilet. Just as our hostility towards foreign suppliers is going up. It's bad.
          • I don't know why any American would be happy to see Intel flounder.

            Because they are frauds who willfully reduced security for performance specifically for the purpose of stifling competition.

            This is (was) a great American company.

            Sure, decades ago.

            • I don't know why any American would be happy to see Intel flounder.

              Because they are frauds who willfully reduced security for performance specifically for the purpose of stifling competition.

              Yes, but Intel acted in this way because they could, because they didn't have significant competition. Do we really think that AMD is virtuous and would not act this way if given a virtual monopoly?

          • Tend to agree. I don't agree with many of Intel's tactics, but Intel up to about 2 or 3 years ago was the defacto leader in process. That crown has now shifted to TSMC. Located in a country that could easily become Chinese just like HK. Would the US or the west go to war over Taiwan? Maybe. But everyone better stockpile 5 years worth of chips before the war starts.
          • by sjames ( 1099 )

            What I would really like is for Intel to pull it's head out of it's ass, but that seems unlikely unless it faces an obvious existential threat. They've spent so long using their name and dirty market tricks to demand premium prices for substandard parts that it may well have to renew itself like a phoenix rising from the ashes in order to get back to competing on price and merit.

      • I'm actually not getting Netburst vibes at all and I'll tell you why. With Netburst they picked the wrong direction to develop architecture in. When you are at the cutting edge there isn't a guide to this stuff, you don't know what will work out and what wont. It seemed that the classic Pentium way of doing things was at the end of its development path and they needed a new way forward. The design teams thought Netburst was the way. Turns out it wasn't but they were able to fairly quickly change course. It

      • his company laughed at by pretty much every hardware reviewer on YouTube from Jays2cents to Gamersnexus.

        Nah, those fakers pretend that, but always come back to praise intel, since them and specially Linus, are deep in Intel pockets, thanks to the bribery program, which is running at full speed.

        • by Anonymous Coward

          Yeah, right, that's the Linus with his new Threadripper system, who's 'deep in Intel pockets'?

          Moron.

      • The high power usage is for the 10900k overclocked. If you look at the stock clock speeds, you'll see it only pulls 152 watts which is quite a bit less than the threadripper which pulls 230 watts at stock speed.

    • Selling the NAND business to invest in some other part of the company? Maybe good.

      Selling the NAND business to enable stock buybacks . . . ? Not so much.

      • That's the bit that worries me.

        I'm quite pleased to see AMD finally having a really good day vs. Intel, good on them, we could really use the competition; but I would hate to see every MBA at Intel treat the current reversal as an opportunity to just abandon the effort to engineer better next generation, continue to work on fab improvements, etc. and just go into white-collar chop shop mode and do a bunch of short-term spreadsheet engineering and selling things off for parts.

        We've lost far too many co
    • by thereddaikon ( 5795246 ) on Friday October 23, 2020 @07:28AM (#60639350)

      ....And then turned around and bought back $10 billion in stock. Because that's a great use of liquidity. /s

      Intel's stock price dropped because its apparent to investors that current leadership have no idea what they are doing. Everyone at the top is an account, nobody knows a single thing about silicon. Instead of solving their problems the finance guys are doing the only thing they know which is sell shit to make a quick buck. It looks good today but wont last in the long term.

    • The strange thing is AMD's stock dips whenever they announce a new better processor.

  • Comment removed (Score:5, Informative)

    by account_deleted ( 4530225 ) on Friday October 23, 2020 @03:22AM (#60638916)
    Comment removed based on user account deletion
  • And nothing about *reality*.

    Thst is why a healthy stock market *must* crash every 30 years or so
    To fix the disparity between belief and reality, when it becomes too much to be believable.
    (Or still crash, but *bigger*, like 2007, if some morons put a cork on that volcano to "prevent" it. )

    And why it is imaginary money,
    and it is a crime absolutely equal to theft, that somebody can pay my real actual work with that made-via-bullshit money!

  • Always hillarious (Score:5, Insightful)

    by Shaitan ( 22585 ) on Friday October 23, 2020 @04:50AM (#60639046)

    Seriously seeing these complex analysis and speculation over reasoning for movements in the market always cracks me up. The market is a bunch of ignorant people randomly responding to a delayed blip on the screen not some highly intelligent and competent animal responding to the nuance of a company's structure.

  • by Anonymous Coward

    News for Stockbrokers, Stuff that matters for Investors.

  • Comment removed based on user account deletion
  • Given COVID-19, one would think with the significant increase in telecommuting and the need for remote monitoring (we're installing sensors in everything so equipment can be remotely monitored), Intel's Data Center business and IoT business would have grown. I think the market is looking at other Cloud based companies and seeing their growth and confused why Intel's divisions that enable telecommuting missed the party.
  • "stock traders are a superstitious and cowardly lot".

    Ignorant, too.

  • I think this is only the beginning of shrinking data center sales for Intel. The hyper cloud guys have custom ARM chips for a fraction of the cost, and it's only going to eat up share from here on out. AMD will get a bit of a boost for a while, but it won't be long before they top out too in the DCs. We know for sure Amazon is moving a lot of machines to ARM, I would assume Google is too. MSFT won't be too far behind. And the Chinese guys definitely want to get off of Intel in case they get caught in some n

Technology is dominated by those who manage what they do not understand.

Working...