Intel Announces a Major Reorg

Posted on July 27, 2020 by Paul Thurrott in Uncategorized with 30 Comments

Microsoft's Big Bets on ARM are Really About Intel (Premium)

In the wake of its 7 nm disaster, Intel today announced that it would reorganize to accelerate its releases and improve focus and accountability. And it looks like the firm has found a fall guy for its problems.

Effective immediately, Intel will separate its Technology, Systems Architecture and Client Group (TSCG) into the following teams, whose leaders will report directly to the CEO:

Technology Development. Led by Dr. Ann Kelleher, this team will now lead Intel technology development focusing on 7nm and 5nm processes.

Manufacturing and Operations. Led by Keyvan Esfarjani, this team will continue Kelleher’s work driving product ramp and the build-out of new fab capacity.

Design Engineering. This team will be led temporarily by Josh Walden, as Intel says it is “conducting an accelerated global search to identify a permanent world-class leader.”

Architecture, Software and Graphics. This team will continue to be led by Raja Koduri and is responsible for driving the development of Intel’s architecture and software strategy, and dedicated graphics product portfolio. It will also continue to invest in software capability as a strategic asset and further build-out software engineering with cloud, platform, solutions and services expertise, Intel says.

Supply Chain. This team will continue to be led by Dr. Randhir Thakur and will ensure that supply chain is a competitive advantage for Intel.

“I look forward to working directly with these talented and experienced technology leaders, each of whom is committed to driving Intel forward during this period of critical execution,” Intel CEO Bob Swan said in a prepared statement. “I also want to thank [chief engineering officer] Murthy [Renduchintala, who is leaving the firm] for his leadership in helping Intel transform our technology platform. We have the most diverse portfolio of leadership products in our history and, as a result of our six pillars of innovation and disaggregation strategy, much more flexibility in how we build, package and deliver those products for our customers.”

Join the discussion!

BECOME A THURROTT MEMBER:

Don't have a login but want to join the conversation? Become a Thurrott Premium or Basic User to participate

Register
Comments (31)

31 responses to “Intel Announces a Major Reorg”

  1. martinusv2

    Intel had to shake things up. Hope this will help them. I eager to see what Raja will do with the graphics department. If it will be competitive against nVidia / AMD.

  2. blue77star

    It is not all doom and gloom.

  3. glenn8878

    7 nm won't save them since the momentum is behind ARM chip design. Maybe the issue is CISC chips needs a complete overhaul to make them low power. Plus, Intel in bungling their 5G modem for Apple doesn't make them seem any wiser or competent.


    Microsoft on the other hand has shown complete incompetence with their ARM strategy. Even if they want to port to the better chip, they couldn't.

  4. truerock2

    I see several issue:


    1. Because Intel had no competition for years, Intel spent years trying to make their CPUs run slow so that they could stretch out product life cycles and maximize return on investment for each product life cycle. I always felt this would lead to disaster. Building your whole corporation strategy on NOT making better products will not end well.
    2. Intel apparently has no problem designing CPUs. Intel's current CPUs are quite good given the 14nm lithography and probably 10nm lithography going forward. The problem Intel has is introducing a new generation of lithography. That is not rocket-science. It is not even a "difficult computer science problem". It is a matter of employees working very hard and very carefully to implement new equipment. Based on my experience, California is rife with lazy, careless people... as opposed to other geographic locations like Taiwan (for example).
    3. Intel never had a good RISC CPU strategy and Intel needs a good RISC CPU strategy. Intel always eschewed RISC because it is difficult to differentiate a RISC CPU from competitor products and also the profit margins are small. Intel needs to get over that mental hurdle. Strategizing up from a small and simple product helps a corporation build better products that are more complex. It is more difficult to be a complex-product organization and try to downsize to small and simple (e.g. Intel Atom CPU). This wasn't such an important strategic issue until the iPhone was created and the world for small-efficient CPUs became so important. Apple CPUs and ARM CPUs could hardly be called RISC anymore. They have become large and bloated CPUs - which is fine if implemented carefully and all that RISC experience from the past is helping Apple and ARM design good "complicated" CPUs.
    4. Intel cannot blow-off CPU market segments. Server, Desktop, Notebook, Tablet, Smart Phone, IOT, etc... are all important to being competitive. There is a chain of design issues that each level feeds from the other levels. If Intel thinks it is going to ignore one area it hurts all of the other areas. Wanting to concentrating on high-margin areas (e.g. grapics processing units) is understandable - but, thinking you are going to blow-off low-margin products just because they are low margin does not work. This is always a problem for giant corporations because the small, strategically important products don't show up in the financials... their numbers are a round-off to 0.
    5. The x86 has become completely uncontrolled in its support of legacy architecture. Supporting legacy architecture keeps a lot of Intel customers continuing to purchase Intel CPUs - but, Intel absolutely needs to start a program of eliminating old x86 architecture.
      • truerock2

        In reply to RobertJasiek:

        x86 is a CISC architecture... not a RISC architecture. I'm not sure what your point is.

        I added point #5 to my comment... maybe that addresses your question?


        The latest Arm CPUs have over 60 opcodes - so, you could question whether the latest ARM CPUs are RISC, I suppose. Except when the latest Intel x86 CPUs probably has something like 2,000 opcodes.

        • RobertJasiek

          In reply to truerock2:

          RISC is fine in theory but will such a CPU run all x86 applications?

          • truerock2

            In reply to RobertJasiek:

            Obviously not...

            again, I'm not quite sure what your point is.


            One of Intel's (and also Microsoft's) biggest problem is that it relies on its legacy technology support as a strategy to retain customers. It is a short-sighted strategy that leaves Intel with a bloated, poor quality product.


            So, sure if Intel wants to sell x86 CPUs with support for 2000 opcodes most of which nobody uses and Intel can make a lot of money doing that... OK - but, what is the longer term strategy? Their product just keeps getting more bloated with obsolete technology and will inevitably be an inferior product. Somehow Intel needs an exit strategy. They could copy Apple if they can't think of anything else.


            Maybe, Intel could just start dropping support for legacy technology that hardly anyone will use in the future?

            • RobertJasiek

              In reply to truerock2:

              If less than 0.5% of x86 applications use rare opcodes, then by all means abandon them immediately. If, however, a major RISC redesign means that more than 50% of x86 applications won't run, then a RISC CPU would be for an alternative market - not for x86 operating systems and their softwares. Unless there would be an automatic compiler (not interpreter / emulation) from x86 (incl. x64) executables (not source codes) to RISC executables.

            • glenn8878

              In reply to truerock2:

              Is Intel the problem or Microsoft? If these 2000 opcodes are really not used, then what's the reason for emulation being such a drag when used in ARM? What if Intel unilaterally begins to dismantle these 2000 opcodes especially since much of the time the CPU isn't being used and Windows still runs sluggishly. CPUs should be fast enough to process these opcodes as software. This is the essence of emulation. Microsoft should already be working on porting these unused opcodes into software in compilation or other tricks. However, this isn't the problem really since the alternative is AMD who has no problem making better chips although it means parity at lower prices and not necessary a big enough bump in performance and small gains in low power.

  5. jfgordon

    I wonder how is the timing of this announcement related to Apple's? Of course, both Apple's shift and Intel's problems are years in the making. However, is Intel's announcement a reaction to Apple going public with its transition plan? Or Apple's announcement a preemptive move to come out on top of the industry given these problems on Intel's camp?

  6. jaboonday

    "This team will be led temporarily by Josh Walden, as Intel says it is 'conducting an accelerated global search to identify a permanent world-class leader.'"


    Imagine being told you're good enough to temporarily run this org, but apparently not good enough to be a "permanent world-class leader." I HATE it when companies do this.

  7. RobertJasiek

    @mattbg, no, Intel does not have another 5 years time. AMD already has the better gaming CPUs. Endconsumers need better mobile CPUs now. Non-gaming desktops need much better energy efficiency or Apple cuts percentages there, too, with silent and fast enough devices.

    Intel was 10 years late starting to develop lower TDP CPUs and then lost 5 years for 10nm, which still provides worse battery life than 10th generation 14nm CPUs.

  8. solomonrex

    As small as Apple's market share is, they have a large share of profitable high end Intel chips, and this sudden collapse can't be just a coincidence.

  9. Pierre Masse

    In which department will they develop the equivalent of ARM processor? In Design Engineering? Maybe they are waiting to find an ARM person to lead it. During that time there is a chair to fill. An armchair.

  10. siv

    I am amazed that Intel never managed to come up with a low power equivalent of the ARM chip that actually worked. Maybe they have to go RISC and that is the only way to get low power and performance?

    • mattbg

      In reply to Siv:

      With a company as big as Intel that promotes mainly from within, I wouldn't be surprised if politics carried a lot of blame. For a company based on x86 with a stronghold to maintain on that side, are they going to give the necessary priority to something that may or may not be the future and, if successful, could challenge the foundation of the company?


      I think the focus on 7nm is a bit overblown - most people care about the end product perhaps even more than the end performance result, and that says nothing about the clock speed, what the transistor length is, how many of them there are, or how tightly they are packed into a given space. The desktop is in decline and replacement cycles are longer, so they have time. But the end products will suffer if they don't sort themselves out in the next 5 years or so.

  11. jhambi

    Yea all this combined with spectre / meltdown crap of recent has tarnished their image for me. One glimmer of excitement I had for Intel was Clear linux, but it seems they are abandoning that as well.

  12. jbinaz

    Something, something, deck chairs on the Titanic?

  13. toukale

    To be honest I am selfishly rooting for Intel not to get anything productive done until 2023 just to give AMD time to pickup some market shares in the pc space to help bring much competition to that space.

    • solomonrex

      In reply to toukale:

      You don't need to root for them. With this admission, it's a lead pipe cinch lock. AMD will expand increasingly into servers, workstations and work laptops, and ARM will take over mobile laptop type devices (they already dominate tablets and chromebooks). And perhaps MS's move into the cloud will finally accelerate in PC shops, as stubborn IT shops run out of hardware options.


      Are you skeptical? OEMs and MS will be spending a lot of time now on AMD (short term) and ARM (long term) and that will impose switching costs (or more if they get into ARM design). Intel's shortages will be real, AMD isn't big enough yet, TSMC has limited bandwidth (and worse if China invades/disrupts) and Apple is ramping up.


      Intel's business model was based on cutting edge fabs, without that, the same revenue will not be there, which will mean further losses for them personally, but also all of their partners. And obviously the loss in reputation was already started and might be fatal now.

  14. codymesh

    this is, unfortunately, 3 years too late.


    also a missed opportunity for Intel to spinoff their fab.

Leave a Reply