Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
Businesses Hardware

Qualcomm To Acquire NUVIA: A CPU Magnitude Shift (anandtech.com) 25

Today, Qualcomm has announced they will be acquiring NUVIA for $1.4 billion -- acquiring the start-up company consisting of industry veterans which originally were behind the creation of Apple's high-performance CPU cores. AnandTech reports: NUVIA was originally founded in February 2019 and coming out of stealth-mode in November of that year. The start-up was founded by industry veterans Gerard Williams III, John Bruno and Manu Gulati, having extensive industry experience at Google, Apple, Arm, Broadcom and AMD. Gerard Williams III in particular was the chief architect for over a decade at Apple, having been the lead architect on all of Apple's CPU designs up to the Lightning core in the A13 -- with the newer Apple A14 and Apple M1 Firestorm cores possibly also having been in the pipeline under his direction.

NUVIA had been able to recruit a lot of top industry talent from various CPU design teams across the industry, and had planned to enter the high-performance computing and enterprise market with a new server SoC with a new CPU core dubbed "Phoenix." NUVIA particularly had made aggressive claims about how their design would be able to significantly outperform the competition both in raw performance and power efficiency once it came to market -- usually such claims are always to be taken with scepticism, however due to the members of the design team and talent having proven themselves in the form of Apple's very successful CPU microarchitectures, there's a lot more weight and credibility to them compared to other start-ups.

Qualcomm now acquiring NUVIA gives them the possibility to take advantage of the start-up's early work in the server space, possibly reinvigorating the company's ambitions in the server space, and giving them a second shot at the market. It's to be noted however that in today's press release about the acquisition there had been no mention of server or enterprise plans. Furthermore, the move also has larger repercussions in the consumer space, with Qualcomm claiming that NUVIA CPU designs are expected to be deployed in flagship mobile SoCs and next generation laptops, as well as other industrial applications such as digital cockpits and ADAS.

This discussion has been archived.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

Qualcomm To Acquire NUVIA: A CPU Magnitude Shift

Comments Filter:
  • by DrMrLordX ( 559371 ) on Thursday January 14, 2021 @05:20AM (#60942298)

    Qualcomm already terminated their own efforts at producing an ARM server CPU. They've mostly abandoned that market. Buying Nuvia does seem to indicate that they're going to take Nuvia away from the server market and instead use their R&D efforts to bolster Qualcomm's existing product portfolio.

    • I don't see any reason why they can't be considering expanding in that market, but also want to use their prowess to improve their cellphone SoCs. It's generally considered positive to have multiple reasons to do something (in this case, buy a company.)

      • > I don't see any reason why they can't be considering expanding in that market

        I suspect Apple is uninterested in the server market beyond their own consumption, so I can see interest on the part of other vendors. Amazon seems like the most likely candidate, but no overt moves yet. Someone's going to go into this space, I guess if you're at Qualcomm it's a matter of "why not?"

        Sadly the most obvious vendor for this market is Sun, who I think would be perfectly happy to move from SPARC to ARM if it meant m

        • I do not believe Oracle has any more SPARC servers in development since the SPARC M8 CPU based systems were released three years ago.
      • Google, Microsoft, and Amazon either have or will produce their own in-house ARM server CPUs using reference Neoverse designs licensed from ARM (until such a time that those license may become unavailable or simply onerous to acquire; see nVidia acquisition). Outside of Ampere, all the companies (including Qualcomm) have thrown in the towel on trying to sell commodity ARM server/workstation products under the assumption that more companies will just design their own server SoCs in-house.

    • by Pinky's Brain ( 1158667 ) on Thursday January 14, 2021 @06:54AM (#60942456)

      Qualcomm have been unable to compete with ARM and if a cloud provider wants to create an ASIC from a standard ARM core they are big enough to do it themselves. There was no money in it.

      If NUVIA has a significant performance advantage over standard ARM cores why wouldn't they sell or license them in more markets?

      • Qualcomm have always licensed for $$ and $$ for the end products sales price. That is why Apple and Samsung ran away. China is improving fast, and understands any release must excite and be close to Apple, and the end game is really about getting the best fab contractor available. ARM is mature, and useless addons are not what the users want. Apple gets it, more cache, more memory, and X battery life, with X86 simulation.
        • Apple uses their massive margins to go as far with vertical integration as possible, to prevent a competitive hardware marketplace in the run up to their monopoly (handed to them by Microsoft's continued failure and Google being an advertising company which will never be able to deliver a truly premium product which managed to ensnare most of their potential competitors). Apple have learned from Microsoft, the competitive hardware market place gave them the leg up when they needed it and Microsoft floundere

      • Qualcomm doesn't compete with ARM. They license the same core designs from ARM as everyone else that has tried or is trying to produce ARM-based server silicon. There's no reason why Qualcomm couldn't have licensed Neoverse (like Amazon and Ampere) to produce their own server products. Instead they bought Nuvia.

        • They don't compete with ARM, but NUVIA's future core was marketed as full custom and aimed to compete with Neoverse.

          Without a full custom core which can outcompete Neoverse, Qualcomm didn't have a convincing deal for hyperscalers. Hyperscalers have the scale to do SoC development in house and the margins they'd be willing to pay for that service would be low, if NUVIA outperforms Neoverse there's a lot more money in it for Qualcomm. That's the difference pre and post NUVIA acquisition.

          • ARM doesn't manufacture based on their own designs, so it's not like Qualcomm had to outperform Neoverse to actually make money with it. Ampere is doing just that. Or at least, they're trying.

            In any case, Qualcomm already nixed Centriq and they didn't bite on Neoverse, despite the fact that ARM specifically introduced the Neoverse platform so that companies like Qualcomm wouldn't have to go with a custom core+interconnect to roll out server SKUs. I still think Qualcomm has zero confidence in the future o

    • Arm was never a good choice for servers simply on account of the pipeline and branch prediction. But thats a very different scenario to desktop and mobile where the Arm seems to thrive.

      Qualcom is looking to build a forte to take on the M1 in phones and Arm laptops. Which is great, having real competition will put a firecracker up apples butt that it cant rest on its (currently rather impressive) laurels.

      The Arm desktop revolution is coming. But I suspect Intel/AMD is gonna be ruling the server room for some

    • Their interest in the company may be more for the technology and personnel rather than the product. Qualcomm might be planning to leverage that server tech into their other mobile offerings.
  • Pro move by Apple (Score:3, Insightful)

    by Anonymous Coward on Thursday January 14, 2021 @05:48AM (#60942340)

    1. convince your employees to leave.
    2. they design a 'server chip'.
    3. company gets bought out by Qualcomm.
    4. switches focus to mobile chips.
    5. Apple sues Qualcomm for breach of IP.
    6. Apple licenses 5G IP at discount as part of settlement.

  • Is this like a Quantum Leap, only for the 2020s ?

    • Is this like a Quantum Leap, only for the 2020s ?

      Ziggy estimates a 90.38% chance that this is indeed the case, however it may take longer for quantum computing technology to truly develop to the point of being useful and economical.

  • I swear at first I read "Qualcomm to acquire NVIDIA" and was mind-blown.

"If the code and the comments disagree, then both are probably wrong." -- Norm Schryer

Working...