A hint of MS Surface’s team looking at designing its own silicon

Avatar
28

“Then Microsoft Chief Product Officer Panos Panay stepped in to tell Bloomberg, “Microsoft plans to continue its expansion into chips, but is unlikely to focus on general-purpose chips like those from Intel, AMD and Qualcomm.

 

“However, Panos interestingly added, “Where we think we need to invest in silicon, we are absolutely going to. If there’s a need, we’re going to fill it.

 

“Translation: If Qualcomm can’t match Apple’s M1 Silicon prowess on ARM based Surface and partner devices, Microsoft will have to follow Apple’s path

Apple’s Craig Federighi and Johny Srouji Describe the Genesis of the M1 Chip while Microsoft Reveals their own ‘Pluton’ Processor – Patently Apple

Comments (29)

29 responses to “A hint of MS Surface’s team looking at designing its own silicon”

  1. Avatar

    ngc224

    Microsoft has a market cap of about $1.6 trillion, and is sitting on well over $100 billion in cash. Apple has significantly more than that.


    Intel has a market cap of about $186 billion, and they’re not getting it done. AMD, about $102 billion. Qualcomm about $165 billion.


    Apple saw the opportunity years ago. Only now is Microsoft starting to get it. They should not be dependent on these other companies.


  2. Avatar

    crp0908

    Panay's quote is just a subtle nod to Microsoft SQ1/SQ2, and Pluton.

  3. Avatar

    illuminated

    It is a bit weird seeing companies moving to proprietary chips.

    It is like the good old days are here again when every every company had its own CPU and OS.

    • Avatar

      jackwagon

      In reply to illuminated:

      Even then, a lot of the companies used similar CPUs; a lot of the 8-bit microcomputers used either 6502 or Z80, and a lot of non-x86 companies used the 68000 series before RISC hit the scene. A lot of the underlying hardware was different between different machines that used the same CPUs; for example, an Atari 800 and a BBC Micro would not have the same hardware past the CPU.


      That said, with many ARM-based processors, different processors (e.g. Exynos vs. Snapdragon) may have different hardware outside of the cores that are used, and operating systems will have to be configured accordingly (most likely compiled prior to shipment to be tailored to the specific hardware).

    • Avatar

      Paul Thurrott

      Interesting. Yeah, it sort of is, though at least there are only 2-3 choices now and not 17 or whatever.
  4. Avatar

    MutualCore

    In reply to bkkcanuck:

    Yes there are no PA Semi around to acquire in 2020. Apple always ahead of the curve.

  5. Avatar

    shark47

    The problem is unless they're already in the process of doing this, it'll be a couple years before we see a chip. I think we give Panos Panay too much credit.

    • Avatar

      Vladimir Carli

      In reply to shark47:

      A couple of years (at least) to see the first gen-1 chip. It took Apple ten years to get it where it’s now

    • Avatar

      will

      In reply to shark47:

      Agreed. I do not see them being 100% in on their own chips and if they are starting to talk about it I would guess a couple years off, and by them we will have M2 or M3 chips, maybe even and R1 chip from Apple.


      Intel and AMD are working on next gen stuff (well maybe more AMD right now) and with all of the device makers out there I do not see MS coming up with a 100% custom chip. Maybe something like they have done before with Xbox in that they have some say on a few items.


      However only Apple can sell Apple devices they have control over software to hardware. There is not a 3rd party like there is for Windows based systems.

  6. Avatar

    waethorn

    Um. Didn't they do this with "SQ1" and come back again with SQ2: Vohaul's Revenge?

  7. Avatar

    lvthunder

    In reply to bkkcanuck:

    Intel or AMD? AMD would probably work better though.

  8. Avatar

    BigM72

    Microsoft will get chips where it makes sense e.g. for hololens, AI ML acceleration (like Apple’s neural engine) but not general purpose CPUs

  9. Avatar

    wright_is

    Microsoft is more into FPGAs and the new integrated replacement for TPM.

    The RoI for designing their own ARM silicon just isn't there. It would need several billion in investment and they'd first need to get a team of expert chip designers and then design all the supporting bits - memory controllers, PCI controller, USB controller , graphic chip etc. Etc.

    That is a lot of work and you will probably have a lead time measured in years,before you have a viable product. Just look at apple,it has taken them nearly 10 years of continuous refinement and development to get it approaching mobile processors speeds,let alone desktop speeds (mobile as in laptop).

    • Avatar

      lvthunder

      In reply to wright_is:

      Why does it have to be an ARM chip? Has anyone seen a 5nm x86 chip? Maybe they run just as cool as the M1.


      Someone has to do it though. PC makers aren't going to be happy that they can't compete with Apple.

      • Avatar

        anoldamigauser

        In reply to lvthunder:

        Yup, no reason it has to be ARM.

        Obviously, an x86 chip would need to come from Intel or AMD. I would bet on AMD getting to 5nm before Intel, but they both have a way to go and Intel need to really up their game if they want to produce the chips themselves.

      • Avatar

        wright_is

        In reply to lvthunder:

        Because x86 is proprietary. They'd have to buy patent licenses from Intel and AMD, and they'd have to replicate all the work Intel and AMD have done over the last 30 years, before they could even hope to match what Intel and AMD currently have - don't forget that Ryzen is already at a 7nm and moving to 5nm with the next generation.

        ARM is the only one that makes any real sense (or openRISC, although that is still years behind, but is unencumbered, but first samples are still in the 30nm or more range), but that would still be a huge effort. It would make more sense for them to go to companies like Fujitsu and get their ARM processors for the desktop and server online than to design their own.

        If they designed their own chips, they'd either be too expensive, or they'd need to sell several hundred million units a year to recoup the RoI.

  10. Avatar

    madthinus

    Designing the chips is one thing. Getting them fabbed with leading edge technology is another. Apple can make chips at 5nm because of iPhone volumes. The can book the block of wafer starts with the likes of TSMC for leading edge designs. Microsoft does not have that volume. That alone might kill anything they try.

    • Avatar

      lvthunder

      In reply to madthinus:

      Do you think they are the only PC makers that would want these chips? How do you think HP, Dell, and Lenovo feel about not being able to make computers as cool (as in temperature) with the same performance as the Macs? I heard Apple has 1000 chip designers. That's not that many if you think about it compared to the size of some of these companies.

      • Avatar

        wright_is

        In reply to lvthunder:

        1,000 chip designers is a hell of a lot of chip designers, when you think about the total number world-wide. There aren't many unemployed chip designers out there - and most of them are probably unemployed for a reason.

        It isn't like writing software. Chip design is highly complex - I worked on Plessey PRISM software for chip design back in the late 80s and early 90s, the software was incredibly complex back then (and that was for custom chips, not general CPUs) and the actual chip designers were well above my pay-grade, when it came to smarts.

  11. Avatar

    winner

    I fully expect Microsoft's Apple envy to kick into high gear, and in 3-5 years they will bring out their own CPUs and then have trouble doing a successful migration.

    They always do this sort of thing when some other pseudo-competitor comes out with a great product: The Mac, the iPod, Google Search, and now Apple silicon.

  12. Avatar

    Vladimir Carli

    Microsoft moving into chip design and then transition to their own platform seems the perfect recipe for disaster. To compete with apple, it seems much more viable for microsoft to accelerate as soon as possible the move to cloud computing

Leave a Reply