The Register Home Page

back to article PC chip plunge wipes out a quarter of Intel’s value

Intel’s share price sank by almost a quarter on Thursday after the company revealed a deep dip in sales of PC processors. The rout began after the silicon-slinger posted Q3 results that included some nice numbers, such as a 21 percent year on year revenue jump to $16.5 billion, helped by 78 percent growth for datacenter and AI …

  1. Anonymous Coward
    Anonymous Coward

    What Innovation?

    This seems the expected outcome of discontinuing flash, accelerator and high speed networking products to focus on a legacy CPU business just before the AI datacenter buildout.

    Maybe Oracle will buy Intel like they did with Sun Microsystems.

    1. coredump Bronze badge

      Re: What Innovation?

      Perhaps they used innovative methods to sack all those people over the last few quarters.

      ...but I tend to doubt it.

      1. Anonymous Coward
        Anonymous Coward

        Re: What Innovation?

        They didn't have to innovate sacking solutions that hard...the market decided it didn't want Sun Microsystems based architecture. Sun was always good for throughput but rubbish for latency and as the internet sped up, latency became more important.

        I remember when Sun Microsystems were, sort of, the rage...you'd see pairs of them all over the place running databases, some places used them for VPN servers because the cryptographic capabilities meant that IPSec etc could achieve much higher throughput and concurrency...but when it came to upgrading/replacing those boxes...folks just moved to x86_64, because the single thread performance was just lightyears better...and we've been there ever since.

        The other thing that happened was Open Source. It became massively easier to get enterprise level functionality without an enterprise level "call for a quote" bullshit.

        So yeah, it wasn't malice that got people sacked at Oracle...it was Corpo thinking that killed off the jobs.

        1. Stuart Castle

          Re: What Innovation?

          I remember, back in the 90s, ordering a CD rom for a Sun workstation. It was one of the low end sparc stations, so used IDE rather than SCSI. Bear in mind that the drive was a bog standard PC CD drive with a Sparc/Sun compatible firmware, how much do you think they charged? I was thinking maybe £200, a little above the average PC drive cost at the time. Nope. Nowhere near. They charged over £600, and I still had to fit it. £600 for a basic, beige, 2 speed CD rom drive. Thankfully, it wasn't my money, my employer paid for it.

    2. FirstTangoInParis Silver badge
      Terminator

      Re: What Innovation?

      > Maybe Oracle will buy Intel like they did with Sun Microsystems.

      And that really will be the death of them.

    3. Sproggit Silver badge

      Suspect that is Unlikely...

      I have no inside knowledge on this, but would point out that the reason that Oracle purchased Sun Microsystems is because, at the time of purchase, *ALL* the biggest Oracle customers were running on Solaris. More than that, perhaps, Oracle had tuned their product to run on Solaris so that - in the words of one of my DBA mates from that era, "it goes like the clappers on Solaris"... You can read a little bit about the steps Oracle took to do that here:-

      https://www.oracle.com/technical-resources/articles/solaris/sol-why-os-matters.html

      but the integration of Oracle RDMBS on to Solaris was legendary - for example you would ask your SA to please not create conventional partitions for Oracle, and instead let Oracle write it's own partition structure to disk. Why? Because their structure was optimized for their DB processes - and eeked out more performance.

      I don't think we can make the same claim of Oracle-on-Intel... The moment we start talking about Oracle running on non-SPARC silicon [the Sun microprocessor family], then Oracle have no allegience to one vendor over another... and in terms of raw performance, it's probably fairer to compare e.g. Xeon and Threadripper, since these would be the most likely platforms for enterprise DB customers.

      All of which is a long-winded way of saying that I don't think the incentives Oracle had to buy Sun apply to Oracle buying Intel, or even a stake in Intel. Strictly from Oracle's perspective, AMD/Threadripper is probably a better bet...

      1. VoiceOfTruth Silver badge

        Re: Suspect that is Unlikely...

        Solaris also had working reliable clustering. I can't remember what the alternative was for Linux on Intel, but it was supplied by Oracle. DBAs preferred Solaris as it worked and they were familiar with it.

    4. katrinab Silver badge
      Alert

      Re: What Innovation?

      The price of credit default swaps (insurance against a company going bust) suggests that Oracle is in very great danger of going bust.

      I think Intel's biggest problem is actually increased competition from ARM-based CPUs. I'm typing this reply on a laptop with a Snapdragon CPU in it.

      1. Steve Davies 3 Silver badge

        Re: suggests that Oracle is in very great danger of going bust.

        I don't think that there will be much in the way of regret around here if that happens.

        in the words of the late Patrick Stewart in Star Trek, 'Make it so'.

  2. Yet Another Anonymous coward Silver badge

    Vertical Integration

    By owning its own Fabs Intel is uniquely positioned to fail to deliver at all points on the supply chain from wafers to CPU simultaneously

    Having sold off its memory and storage business because there is no way to make money selling memory and storage.

    1. Anonymous Coward
      Anonymous Coward

      Re: Vertical Integration

      Intel pioneered DRAM chips ... the 1103, back in 1970~1971.

      https://en.wikipedia.org/wiki/Intel_1103

      And now there's not enough (sufficiently-cheap) DRAM to support PC CPU sales.

      Irony.

    2. Eric 9001 Bronze badge
      Boffin

      Re: Vertical Integration

      It's not hard to make money selling memory and storage, considering that all manufactures have been very hesitant to compete by reducing prices to anywhere near the manufacturing cost - so if you got fabs that can make either, you can print money.

      Sure, the 5 billion dollar fab needs to get paid for, but if a billion large capacity RAM or NAND chips get sold and $10 of profit was made on each, the business turns out ludicrously profitable (the profit margins are usually higher than that).

      While the "Optane" product line seems to have been oddly expensive to manufacture and thus very expensive (but people bought it anyway, as it was at least twice as fast), the main reason why the products failed was seemingly because of Intel's handcuffing schemes (for example, they offered 3D XPoint as main memory, but instead of making it just work as RAM and provide software support for it, they shackled it to a specific server chipset and "surprisingly", it sold poorly).

  3. Blue Screen of Bleurgh Bronze badge

    I am reminded of that great Gordon Gekko quote from "Wall Street" Seems rather apt here when you strip away the veneered BS.

    "Greed, for lack of a better word, is good. Greed is right. Greed works. Greed clarifies, cuts through, and captures the essence of the evolutionary spirit."

  4. Like a badger Silver badge

    Jam tomorrow

    Have Intel been taking lessons from the British government?

  5. Missing Semicolon Silver badge

    Investment

    Almost like if cutting off all investment in making stuff chokes off the income from selling stuff. Whodathunk?

    1. Anonymous Coward
      Anonymous Coward

      Re: Investment

      Yeah but that's long term thinking, you're not thinking like an MBA...you need to understand, an MBA only intends to be around for 2 years, until they can be "poached" by another large business...by poached I mean intentionally fired so they can claim their golden parachute. If you don't follow the MBA playbook you only get one insanely high payout...or even zero...if you want both of the massive windfalls in one 2 year sitting at a large business you must follow this process:

      0. You need to be basically anonymous, someone people have never heard of before, but somehow has an illustrious industrial history hopping around mid tier jobs for 20 years...it's important that you never stay in a job longer than 18 months to 2 years because if you do, the game is up, people will know that you know nothing.

      1. Takeover when a really good CEO is retiring / has died / has suddenly become a basket case. Just as a new, popular product launch is happening.

      2. Take credit for the results from this product cycle because of the murky water around the product launch.

      3. You will naturally be criticised for this, so it's important that now find a way to increase profits in a way that requires no expertise yet makes the shareholders believe that you are an industry stalwart. In preparation, you produce Q3 finances that show a downturn in revenue, you blame your predecessor...you proudly declare "ITS TIME FOR CHANGE / INNOVATION / THE SPIRIT OF XYZ TO RETURN...whatever".

      4. Put out loads of press releases about products in development. If those don't garner any attention...orchestrate a "leak". This is purely for hype purposes...these products will never exist, so you never have to produce any actual benchmarks or comparisons with anything...just bar charts with "2x faster" etc...it's important that if you do put a fake comparison chart out that you never cite what you're comparing to. It's just 2x faster. Put up a pre-order for a phantom product.

      5. Layoff thousands of staff before the end of Q4...sometime it has minimal press coverage, yet maximum impact on the employees...around about Christmas will do. Now you wait for the Q1 results.

      6. Q1 results arrive, through sheer genius, hard work and exceptional leadership, you have increased profits by 18%, you have achieved 100,000 pre-orders for a product that doesn't exist that you have spent zero R&D on yet. Accept your profit linked bonus.

      7. You're now in a position where you have claimed a huge bonus, you have a phantom product in the pipeline, you are understaffed...but extremely profitable. Your eyes turn to the exit...you need to get out before the shit hits the fan, but you want the golden parachute...to achieve this you need to find a way to get fired, but not be labelled as incompetent.

      8. You take an interview on a semi well known internet publication. Somewhere that has a history, but is more popular and widely known than a trade magazine, but not really mainstream. A mediocre outlet like The Register. You say something offensive...like shitting on your predecessor, pick the wrong pokemon, admit to liking systemd...something that will get the nerds enraged but will make no sense to anyone out there buying the products.

      9. Wait for the PR department to pick up on it. They live and breathe on Reddit these days and live in a bubble...so when they see the outrage and vitriol on Reddit...they fire it up the chain to the board. The board are even more out of touch and think you're a massive threat...they "fire you" by allowing you to resign and give you your golden parachute...you leave with you professional record intact, nobody in the general public is any the wiser around your mild transgression and you're ready to "bail out" another company.

      Remember kids, use this knowledge wisely...it won't work for everyone...you need to permanently exist in the realms of "plausible but difficult to prove"...it's important that you never actually directly get involved in producing anything...you need to be close enough to something to be strongly associated with it, but distant enough that if there are problems they're nothing to do with you.

      I.e. it was all your work, but never your idea.

      You need to dodge accountability and responsibility like an Olympic gymnast crossed with a Spaniel and time your moves to occur roughly 2-3 months before the shit hits the fan so it can't be pinned on you.

  6. Tron Silver badge

    AI is killing the sector.

    Only the maleficent seven believe they are doing well out of it, but they are just funding each other on a crazy circular binge to force people to buy stuff that they do not want and are now starting to resist.

    Datacentres are not a solution. Those who really want to keep using AI will do so on prem for security. The cracks are starting to show.

    1. Danny 14

      Re: AI is killing the sector.

      sort of. it is a good time to be selling RAM and storage. If only intel had a business model selling them....

    2. Geoff Campbell
      Pint

      Re: AI is killing the sector.

      "the maleficent seven" - bravo! Well played, I say!

      I heard a financial analyst refer to "the S&P 493" the other day, and I'm still chuckling now.

      GJC

    3. Anonymous Coward
      Anonymous Coward

      Re: AI is killing the sector.

      I agree...but local AI won't seriously take off until NVIDIA loosens it's death grip on the GPU market or a new market for AI accelerators emerges.

      I'm amazed we haven't seen any ASIC or FPGA tech arise yet...when crypto mining was in it's early years, there was a huge race to push hashing rates forward and we got from CPU mining to ASIC/FPGA mining pretty quickly...I know, I know...it's not the same thing etc etc...but the fact that we went from general purpose compute to purpose built compute in a very short span of time was testament to the innovation going on, it took 4 years to go from general purpose compute to specialised compute...AI seems much, much slower for some reason. We're about 8 years in at this point...it all seems very fishy. The massive RAM shortages etc seem very convenient for NVIDIA...#justsayin.

      1. Claptrap314 Silver badge

        Re: AI is killing the sector.

        The thing is, ASIC and FPGA were already very much a thing. The move by crypto to these was precisely because hashing really is so special that dedicated hardware can give you a big win. (Pun intended.)

        Having 0 direct exposure to the relevant code, but with a fair amount of knowledge of some of the relevant factors, I expect that FakeI is doing a lot of sparse matrix multiplies. Or maybe not so sparse. That involves a lot more random memory access than hashing does. It's also what CPUs (and GPUs?) are made for. Since memory access has been THE bottle neck for a few decades, again, CPUs are already optimized for this class of work.

        1. Anonymous Coward
          Anonymous Coward

          Re: AI is killing the sector.

          For sure, but FPGA and ASIC (old as they are) are usually last ditch efforts to get the most out of energy...but the trade off is they are extremely limited once implemented (ASIC for sure, FPGA, less so)...they are generally single purpose chips...

          All we need for a bit of a leap forward towards ASIC or FPGA for AI is a bump in storage performance...remove the need for masses of VRAM and stream models directly from storage...it's already possible with some of the inference tools out there right now...but most people don't have access to fast storage for various reasons...but you can bet there are some players out there that already have scaled storage that they barely use that can be put to work for local AI, bypassing the need for huge amounts of RAM.

    4. Anonymous Coward
      Anonymous Coward

      Re: AI is killing the sector.

      I think the maleficent seven are fighting it out, purely because they know that not being on top is an existential threat to them. They're doing it because they have to, not because they want to.

      I doubt any of the big players (outside of OpenAI and Anthropic) really care all that much about AI...they're fighting for survival.

  7. breakfast Silver badge

    The bubble quivers

    Could this and Alphabet dropping, along with NVidia pulling on some deals that indicate weakening demand, be the first turbulence that leads the the correction we've all been waiting for? If I was working around the markets I'd be double and triple-checking my escape strategies at this point.

    1. Geoff Campbell
      Pint

      Re: The bubble quivers

      "Slowly at first, then all at once". Originally about politics, I think, but applies equally to the end of IT bubbles.

      GJC

    2. Tron Silver badge

      Re: The bubble quivers

      I thought that IBM's boss stating to investors (not geeks) that 'open weight models are now irresistibly good and cheap on-prem' was the death knell for the giant datacentre model. The datacentre investment is a huge chunk of this, and anyone locked into it may be about to see some Titanic/iceberg scale synergy.

      The collapse of the datacentre model might take down enough for the markets to sour on AI in general. Nowhere outside hacker groups is seeing serious benefit or ROI. Titans of industry (as they see themselves) don't like being scammed by irritating tech bros, taking all their cash and making them dependent on third parties to do absolutely everything.

      This may be less a 'correction' as tipping point for a very rapid deflation.

      Those with skin in the game might consider that if they wait too long, stock markets will suspend trading and apps will freeze up. Better to be on the first chopper out of Saigon than to be trying to get on the last.

      1. doublelayer Silver badge

        Re: The bubble quivers

        It wasn't. IBM's boss told investors that the things that IBM sells will sell well because the things IBM doesn't sell aren't necessary. Ask Anthropic or Spacex and they'll tell you that frontier models are advancing faster in quality and efficiency, which means that however people run them, they will still have the market for companies investing the most. Ask CoreWeave or AWS and they'll tell you that LLM users are heavily biased to large, unquantized models and sure, you can build your own GPU clusters, but you either build too few and have to ration who gets to run LLM queries or you build too much and have expensive equipment sitting idle. IBM's statement provided no news about technology or purchasing trends.

    3. Anonymous Coward
      Anonymous Coward

      Re: The bubble quivers

      No probably not, corrections rarely happen...people have been waiting for a correction in the housing market for damned near 30 years...growth has slowed down.

      Tech bubbles like AI tend to follow an S curve. Slow growth followed by an explosion followed by a plateau.

      The way an AI "crash" is likely to occur is that growth will simply stop...the market won't shrink, demand won't disappear...there just simply won't be any more massive growth for a while.

      To a certain extent, we're already seeing it...datacentre projects being put on hold, less hiring, investment slowing down, mergers/acquisitions happening...the signs are all there for an "S" curve...the signs are not there for a full blown snap backwards.

  8. Geoff Campbell
    Facepalm

    Well, yeah.

    Here's a clue - do something to make RAM and storage affordable again, and people will go back to upgrading their PCs, with nice shiny new Intel CPUs. There's no point buying a new motherboard and CPU if you can only afford half the RAM and storage your old PC already has.

    Advanced, high-level consultancy, I know. My invoice is in the post.

    GJC

    1. Steve Davies 3 Silver badge

      Re: and people will go back to upgrading their PCs

      Over ar Microsoft they are saying,

      "Hold my Beer"

      as they frantically work out how to introduce more Hardware restrictions for Windows 11 so that people will be forced to buy new Intel Powered computers.

    2. Irongut Silver badge

      Re: Well, yeah.

      Intel could make RAM free, I still wouldn't buy one of their shitty CPUs.

      Why buy a second rate CPU manufactured by a second rate chip fab when you can buy an AMD CPU manufactured by TSMC - the fab Intel's own CEO says are the best.

      1. Geoff Campbell
        Pint

        Re: AMD

        I just *knew* that would trigger the AMD fanboys :-)

        Yeah, OK. This is coming to you from an AMD PC, as it happens. I also have several Intel PCs, and a bunch based on various ARM chips. Guess what? They all work just fine.

        GJC

      2. Anonymous Coward
        Anonymous Coward

        Re: Well, yeah.

        Hey, easy there...

        Intel has some of the best used CPU deals on the market. When they launch new stuff, it's really boring and uninspiring...but 4-5 years down the line, the price will have dropped on it so much on the used market that it becomes incredible.

        I just switched my server stack to Xeon Gold 6248 and Platinums in the same generation...I got to re-use all my RAM (plus install a little bit more) and the performance bump is insane. I did it for less than £1,000...I've got high performance compute for days at a very cheap price point...single threaded, they're about equivalent to a 5800X...but use significantly less power and have significantly more cores...about 4 times the memory bandwidth as well...actually insane.

        Brand new, these CPUs would have easily been £15k each...you can pick the CPUs up by themselves for around £35 each now...sure the RAM is expensive, but when you're saving that much on the CPU with equivalent performance to a CPU that is still considered "decent" by modern standards, who cares?

        The only people getting fucked by the RAM prices are people trying to build new machines with newly released components. You can find equivalent old Intel stuff that will do as good a job (or better) for next to fuck all.

    3. Bran Muffin

      Re: Well, yeah.

      Has anyone else seen Sam Altman's interview where he decries private ownership of computers (and all that lovely memory they contain)? Its the most tone-deaf, ridiculous, self-serving nonsense you will ever read or listen-to.

      1. coredump Bronze badge

        Re: Well, yeah.

        ... until the next one.

  9. Groo The Wanderer - A Canuck Silver badge

    IBM v 2.0... *LOL*

  10. Irongut Silver badge

    Strong progress on Intel’s manufacturing processes that will bring it closer to parity with TSMC

    In the future we will be ALMOST as good as the competition! But not as good. And not yet.

    Truly a glowing recommendation from the company's own CEO. No wonder the stock market hated it.

  11. Bitsminer

    "our OEM partners are aggressively managing inventory..."

    Intel-speak.

    They stopped ordering product.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon