For Arm, from 🤫 hussh
One architecture, on both ends of the agent.
Arm is under nearly every phone a private agent lives on, and under Graviton, Grace and Axion in the cloud. 🤫 Private Agent One runs on both ends: the device in the person's hand, and the efficient servers of a garage and warehouse grid. Performance per watt is your moat and our cost advantage. We'd love to make the private agent an Arm-native workload, device to data center.
What Arm and 🤫 make possible.
Nearly every device our agent runs on is Arm, and KleidiAI accelerates on-device AI with no code changes.
We will build KleidiAI and SME2 into Agent One's on-device inference.
A private agent accelerated on every Arm phone and PC, with a royalty on each.
Graviton, Grace and Axion are all Neoverse, the efficiency leaders for the AI grid.
Our garage and warehouse grid standardizes its CPU tier on Neoverse.
Royalties across our device and grid footprint alike.
You have moved into your own silicon with the Arm AGI CPU for agentic AI infrastructure.
We bring a real private-agent workload for exactly that market.
Agentic AI on Arm, proven from device to data center.
Everywhere, efficient and one architecture end to end.
Arm designs CPU architectures and IP, and earns license fees plus per-chip royalties that recur for the life of every design. A uniquely durable model. Its fiscal year ended March 31, 2026.
Up 23% year over year, a third straight year above 20%, split about 53% royalty and 47% license.
Cumulative Arm-based chips, the largest installed base of any CPU architecture.
Apple, Qualcomm, MediaTek and Samsung SoCs are all Arm-based.
Arm's figure for top-hyperscaler CPUs, led by Neoverse in Graviton, Grace and Axion.
Arm's first in-house production chip, announced March 2026, for agentic AI in the data center.
Scalable Matrix Extension 2, delivered through Lumex and KleidiAI.
Reach and power efficiency at every tier. The ground a private agent runs on.
One architecture, phone to data center
The same family spans phones, edge, cars and cloud servers. One software target, end to end.
Power efficiency
The moat behind mobile and the case for the data center. Performance per watt makes a low-cost grid possible.
The largest ecosystem
The biggest developer and software ecosystem of any CPU architecture. Code written for Arm runs across the fleet.
Compute Subsystems
Pre-integrated, validated subsystems that shorten time to silicon and carry higher royalty rates. A growth driver in FY2026.
Neoverse in the cloud
Graviton, Grace and Axion are all Neoverse. Arm cites about 50% of top-hyperscaler CPUs and nearly all data-center DPUs.
Armv9 and SME2
Today's architecture, with matrix acceleration on the device. Higher royalties and real AI performance.
Where Arm is going.
Every one of these priorities has room for a private agent the person owns.
AI everywhere on Arm
Rene Haas calls agentic computing the next phase, built on high-performance, power-efficient computing at global scale.
Neoverse in the data center
Neoverse royalties doubling year over year, with about half of hyperscaler CPUs and nearly all data-center networking.
Into silicon
The AGI CPU takes Arm from IP and subsystems to its own production chips for agentic AI.
On-device AI
Lumex, SME2 and KleidiAI: real-time assistants, translation and personalization on the device, with no per-app porting.
Efficiency leadership
Performance per watt as the through-line, from the phone to the server rack.
Lumex, KleidiAI and the Arm AGI CPU.
What Arm shipped and announced most recently, and where Agent One plugs in.
Lumex CSS
Announced September 2025: Armv9.3-A C1 CPUs, SME2 and the Mali G1 GPU, with up to about 5x faster on-device AI and about 3x better energy efficiency than the prior generation.
KleidiAI
Libraries that bring SME2 acceleration to apps with no code changes, built into PyTorch ExecuTorch, Google LiteRT and Microsoft ONNX Runtime.
Arm AGI CPU
Announced March 2026, with broader availability in the second half of 2026, for agentic AI in the data center. Meta is the lead customer.
Neoverse V-series
The cores behind Graviton5, Grace and Axion. The efficient CPU tier for the AI factory, with NVLink Fusion.
Buy together. Build together. Sell together. Take risk together.
Arm runs on both ends of our north star: the device the agent lives on and the efficient servers behind the grid. Everyone gains, and nothing is taken from the person.
Build: KleidiAI inside the agent
SME2 and KleidiAI accelerate on-device AI across Arm phones and Copilot+ PCs, with no code changes.
We will build KleidiAI and SME2 into Agent One's on-device inference, so the private agent runs accelerated on every device with no porting.
A private agent that runs on the device, consent-first, and an Arm royalty on every device it runs on.
Build: the grid on Neoverse
Graviton, Grace and Axion are all Neoverse, the efficiency leaders.
We will standardize the 🤫 Puppy and Warehouse One CPU tier on Neoverse: one software target across device and data center.
Arm takes load off a sold-out cloud at low cost, and we get the most efficient CPU tier there is.
Buy: an Arm footprint, end to end
Arm earns a royalty for the life of every design that ships on its architecture.
We commit our device and grid buying to Arm-based silicon: Lumex-class SoCs on the device, Neoverse in the grid. Every 🤫 node ships as an Arm design.
Recurring royalties across our whole footprint, and one architecture for us, device to data center.
Sell: private, efficient AI on Arm
Arm's story is edge AI and cloud AI on one efficient architecture.
Together we tell the story of private, efficient edge AI on Arm: Puppy One nodes on Neoverse, Agent One accelerated on Lumex-class silicon.
One story that serves both companies' edge and cloud narratives.
Take risk: the agentic reference workload
The AGI CPU targets agentic AI infrastructure, Arm's new market.
We would love to design with you early, with a real private-agent workload across device and grid.
Arm gains a genuine agentic workload. We gain first-class acceleration.
The people building Arm.
Rene Haas
CEO
Leads AI everywhere on Arm and the move into production silicon: efficient computing for the agentic era.
Jason Child
EVP and CFO
Leads the royalty and license model and the economics of the AGI CPU business.
Richard Grisenthwaite
EVP and Chief Architect
Leads the architecture roadmap: Armv9, SME2 and the foundation an agent runs on.
Chris Bergey
EVP, Edge AI
Leads mobile and edge: Lumex, KleidiAI and on-device AI, where the private agent lives.
Mohamed Awad
EVP, Cloud AI
Leads data center, Neoverse and the AGI CPU, the grid tier of what we build.
Performance per watt, the spine of the AI factory.
Arm's efficiency is what makes a low-cost edge grid possible, from the phone in the person's hand to the server in the garage. One architecture, owner-operated, all the way through. And 🤫 Agent One is free for life to every federal and state government employee and every past or present service member who helps build it.