AIToday
AI Business & IndustryITmedia AI+Published: Nov 20, 2026, 10:01 JST

Arm's data-center share hits 50%, says Arm Japan chief

Send on LINE
Arm's data-center share hits 50%, says Arm Japan chief

3 Key Points

  1. What happened

    Arm Japan President Takayuki Yokoyama said Arm's share at major cloud providers' data centers reached 50% in 2025, up from 0% eight years earlier.

  2. Why it matters

    As AI shifts from chatbots to agents that repeatedly call tools and databases, much of that work runs on CPUs, where Arm designs chips.

WHO IT HITSThis lands on cloud providers designing data-center servers and on smartphone and device makers, who can buy Arm's pre-integrated CSS designs to shorten chip design-to-shipment time.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

Arm's traditional business is licensing circuit designs (IP) to chipmakers, and its designs are used in almost every smartphone model. The shift Yokoyama describes is about where AI work happens: as AI moves from answering questions to completing tasks, agents repeatedly call tools, pull data from databases and coordinate with other agents. Yokoyama says much of that general-purpose processing runs on CPUs rather than GPUs.

The economics of Arm's model are also changing. Instead of supplying designs component by component for chipmakers to assemble, Arm's CSS bundles multiple designs with optimization and validation already completed, which the company says can sharply cut time from chip design to shipment. In a test simulating an AI agent planning and booking an anniversary dinner, Arm says processing time fell 24% versus the prior generation.

Yokoyama ties the edge push to cost, privacy and latency: relying on a data center round trip can feel slow and raises risk when AI handles preference and behavior data. Whether Arm converts its data-center momentum into the same position on devices is the open question its new mobile package is meant to answer.

FAQ
What does Arm's CSS for Mobile 2 offer device makers?
It bundles multiple designs with optimization and validation already done, and its C2-Ultra CPU with SME2 speeds small language models on devices by up to 70%.
Why is AI processing moving to edge devices?
Yokoyama cited usage cost, privacy and response speed, saying sending everything to data centers is not always best as AI handles personal data.
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleMeta reevaluating 2030 net zero goal as AI buildout strains clean energy