Forgot your password?
typodupeerror
AI Hardware Slashdot.org

Arm Unveils New AGI CPU With Meta As Debut Customer 29

Arm unveiled its first self-developed data center chip, the AGI CPU, designed for handling agentic AI workloads. The new chip was built in partnership with Meta and manufactured by TSMC. Other customers for the new chip include OpenAI, Cloudflare, SAP, and SK Telecom. Reuters reports: The new chip, called the AGI CPU, will address data-crunching needed for a specific type of AI that is able to act on behalf of users with minimal oversight, instead of responding to queries as part of a chatbot. For years, Arm, majority-owned by Japan's SoftBank Group has relied only on intellectual property for revenue, licensing its designs to companies such as Qualcomm and Nvidia and then collecting a royalty payment based on the number of units sold.

"It's a very pivotal moment for the company," CEO Rene Haas said in an interview with Reuters. The new chip will be overseen by Mohamed Awad, head of the company's cloud AI business, and Arm has additional designs in the works that it plans to release at 12- to 18-month intervals. TSMC is fabricating the device on its 3-nanometer technology and is made from two distinct pieces of silicon that operate as a single chip. Arm plans to put it into volume production in the second half of this year but has received test chips that function as expected. In addition to the chip itself, Arm is working with server makers such as Lenovo and Quanta Computer to offer complete systems.
This discussion has been archived. No new comments can be posted.

Arm Unveils New AGI CPU With Meta As Debut Customer

Comments Filter:
  • by Misagon ( 1135 ) on Tuesday March 24, 2026 @04:09PM (#66059574)

    "AGI" is just a silly name chosen for marketing purposes.
    Not even close to "Artificial General Intelligence".
    Nothing to get agitated about.

    • by tlhIngan ( 30335 )

      In other words, it's just another fancy ARM SoC with a silly name. Like the trillions of other ARM SoCs out there.

      The big question I suppose is why they went with ARM and not say, RISC-V to make their SoC.

      • by Anonymous Coward

        Are you asking why ARM went with ARM to make their SoC?

        • ARM doesn't "make" anything, They license the ARM architecture to other people to use to fab processors. The article said this new processor was made by TSMC. What the article doesn't say is what makes it different from every other ARM SoC. Mind you, ARM SoCs are pretty good, and writing firmware for them pays my bills.
      • Because they have a software ecosystem that is well understood by just about everyone?

      • by rmav ( 1149097 )

        Because it is ARM. I mean, this SoC was designed by ARM, the company. Why should they use a competing ISA?

    • A bunch of stock-trading bots and quick-fingered suits probably just flinched and dumped a couple hundred billion into ARM stock though. See also: the company formerly known as Long Island Iced Tea.

    • by necro81 ( 917438 )

      "AGI" is just a silly name chosen for marketing purposes.

      Well, since ARM is trying to sell something, I guess that fits.

  • ...that had zero technical information about the chip design

  • The new chip, called the AGI CPU, will address data-crunching needed for a specific type of AI that is able to act on behalf of users with minimal oversight, instead of responding to queries as part of a chatbot.

    Are there end-users actually asking for this? AI chatbots are at least mildly amusing at times. But they hallucinate enough I wouldn't want to turn a current-gen AI loose to take actions on my behalf with little oversight. I wouldn't want to turn them loose to take actions with direct and constant oversight. This new push to have AI agents continually doing things without our interaction seems to be very cart before horse at the moment. It's like the entire AI push is utterly ignoring that testing cycles ex

    • You know, I've had girlfriends that hallucinated a lot, and they were still good in bed...
    • No. The "agentic" shite is handled fine by the multicore CPUs which tend to idle while the model burns through the "AI" cores anyway.

    • by gtall ( 79522 )

      I view it as a way to address investors' stupidity on AI-Hype and putting their name out there into the AI-Verse as relevant to a synergistic forward thinking future-centric Brave New AI-World. Marketing across the techno-babble universe will get hyperbolic over this. Marketing profits will flow. And then, in a fit of synergistic forward thinking future-centric hypergasm, they'll make it Quantum Ready.

      • I view it as a way to address investors' stupidity on AI-Hype and putting their name out there into the AI-Verse as relevant to a synergistic forward thinking future-centric Brave New AI-World. Marketing across the techno-babble universe will get hyperbolic over this. Marketing profits will flow. And then, in a fit of synergistic forward thinking future-centric hypergasm, they'll make it Quantum Ready.

        My buzzword bingo card is full!

    • by rmav ( 1149097 )

      Are there end-users actually asking for this? AI chatbots are at least mildly amusing at times. But they hallucinate enough I wouldn't want to turn a current-gen AI loose to take actions on my behalf with little oversight. I wouldn't want to turn them loose to take actions with direct and constant oversight. This new push to have AI agents continually doing things without our interaction seems to be very cart before horse at the moment. It's like the entire AI push is utterly ignoring that testing cycles exist for a reason. "Let it run wherever, what's the worst that could happen?" shouldn't be the philosophy on something being pushed as a world-changing technological advance.

      This is a chip for server farms, not for PCs.

  • by pele ( 151312 )

    Has anyone else noticed that somehow SAP is always somewhere in there in everything that is being done?

  • by presidenteloco ( 659168 ) on Tuesday March 24, 2026 @05:59PM (#66059774)
    Because when enough people start using a technical term incorrectly, that mistake becomes the "official, most correct" usage of the term in the natural language.

    Examples:

    CPU - you know, the big beige computer box under your desk, as opposed to the "monitor". Luckily somehow English has escaped from that misuse that was common in the 1990s and 2000s.

    "I could care less" - This now means "I couldn't care less" which actually made literal sense.

    "Meme" - you know, those silly viral images or animations passed around on socials or the Interweb: Example "Grumpy Cat"
                                    As opposed to its intended scientific meaning as "a unit of information with the property that it can induce its hosts to replicate it and thus carry it forward in time." Examples: the holy books of a religion.

    And now we have "AGI" which if we're not careful will come to mean "Agentic AI" or even, an ARM chip used for AI, as opposed to "Artificial General Intelligence".

     
  • by greytree ( 7124971 ) on Wednesday March 25, 2026 @09:33AM (#66060644)
    Please queue up to buy the new "Ford Mustang 1000mph Edition".

Power corrupts. And atomic power corrupts atomically.

Working...