Arm

Running agentic AI systems at scale puts the CPU directly on the critical path - enabling complex orchestration while introducing new levels of concurrency and latency sensitivity that traditional inference patterns were never designed for. Google Cloud's Agent Sandbox on GKE, built on Arm-based Axion processors, is designed for exactly this. With support for up to 300 sandboxes per second per cluster and sub-second time to first instruction, it enables high-throughput agent orchestration alongside cost-efficient inference, complementing accelerators like TPUs on the same platform. Together we're redefining infrastructure with a more practical path to deploying agentic AI in production while balancing performance and cost efficiency. https://okt.to/7rcdZU

As agentic AI moves from experimentation to production, Google Cloud is taking a major step toward operationalizing agentic AI at scale. Today at #GoogleCloudNext they've announced the new TPU 8t and TPU 8i systems as well as the introduction of its Agent Sandbox on Google Kubernetes Engine (GKE), a purpose-built deployment framework designed built on Axion, Google’s Arm Neoverse-based CPU. ⚡️ These updates underscores a crucial shift toward purpose-built CPU architectures for next-generation AI workloads.

We're extending the Arm compute platform into production silicon with the Arm AGI CPU - purpose built for agentic AI. As Vince Jesaitis, Head of Global Government Affairs at Arm, outlines, this shift reflects how AI is reshaping the role of compute in the global economy. As AI infrastructure becomes foundational to economic growth, open, standards-based platforms will be critical to enabling global innovation at scale, supported by broad ecosystem participation and energy-efficient compute that can be deployed globally. https://okt.to/x9Tycl

Last week at #CadenceLIVE, Suraj Gajendra joined industry leaders from Cadence, NVIDIA, Dyna Robotics and Rivian to explore how the industry is closing the sim-to-real gap in physical AI. As AI moves from cloud to real-world systems, ensuring that what’s trained in simulation performs safely and reliably at the edge is critical. From compute platforms to software and simulation environments, it takes a full-stack, system-level approach and deep collaboration across the ecosystem. At Arm, we’re working with our partners to enable this continuous loop of real-world data, simulation and deployment, helping scale physical AI from concept to safe, real-world impact.

A first look at our new Seattle office 👀 We’ve recently opened our new Arm Seattle office, a space designed to bring our people, ideas, and technology closer together. Inspired by the concept of a “Living Circuit,” the workplace reflects how innovation happens at Arm: through connection, between teams, disciplines, and the broader ecosystem. From locally inspired artwork to spaces designed for collaboration and focus, the office brings together Pacific Northwest culture and cutting-edge compute in a meaningful way. Throughout the space, subtle details are layered in, from references to Arm’s origins and architecture to nods to the region, designed to be discovered over time and reflect how our technology and ecosystem have evolved. An exciting next chapter for the Seattle team as we continue to build the future on Arm.

Arm is heading to Google Cloud Next 2026 to showcase how cloud native and AI workloads are scaling on Arm. 🎉 Built on the Arm Neoverse platform, Google Axion processors enable strong price performance, improved energy efficiency, and the flexibility developers need to optimize across a wide range of workloads. If you are attending, be sure to visit the Arm booth to see live demos and connect with the team - we're looking forward to engaging with the community and sharing how the shift to Arm based infrastructure is accelerating innovation across the cloud ecosystem.

At #ArmEverywhere, Rene Haas shared why AI infrastructure is reaching a new limit. Agentic AI is pushing demand beyond human scale — with systems running continuously. Accelerators generate the tokens. CPUs keep everything moving, 24/7. As AI scales, that’s where the bottleneck is emerging. That’s why we built the Arm AGI CPU — designed to keep AI infrastructure running efficiently at scale. Discover more: https://okt.to/Ig9EGJ

We've received the Open Doors Award as part of the World 50 Impact Awards! 🎉 This recognition reflects our Inclusive Pathways strategy to expand access to STEM globally — helping open doors into technology for millions of learners through education, mentorship and early-career opportunities. 👏 Thank you to the schools, universities, nonprofits and partners around the world who make this possible — and to World 50 Group for the recognition.

Join us in San Francisco for Arm Create, a hands-on experience designed for developers working across AI, ML, and compute. 👨💻 Taking place alongside the AI Developer Conference by DeepLearning.AI, this in-person event brings together the community to explore how AI models run efficiently across cloud and edge environments on Arm. 🔹 Build AI experiences in minutes 🔹 See real-world demos across cloud and edge 🔹 Connect with Arm engineers and fellow developers 🔹 Enjoy drinks and bites throughout the evening 📅 April 28, 5-8 PM PDT 📍 25 Lusk St (8-minute walk from Pier 48) Attendance is free, with limited capacity. Request your invite: https://okt.to/TK1Xif

At HumanX, Will Abbey, EVP and Chief Commercial Officer at Arm, highlighted a defining shift for AI infrastructure: delivering more performance within fixed data center power envelopes. As agentic workloads emerge, that balance of performance and efficiency becomes critical. It’s exactly the kind of workload the Arm compute platform is designed for, and why we're expanding to include the Arm AGI CPU, built to deliver greater performance density at scale. 👀: https://okt.to/FLXBW6