At the Arm Everywhere event, Arm CEO Rene Haas introduces Arm’s first production silicon, the Arm AGI CPU, a company first as Arm moves beyond IP and compute subsystems to building complete chips for AI data centers.
In his keynote, Rene outlines why Arm is expanding into silicon and how the rise of agentic AI is changing the requirements of modern data centers. As AI workloads move beyond simple queries to continuous, multi-step tasks, demand for CPU compute, memory bandwidth, and power efficiency is increasing.
Meanwhile, Mohamed Awad, EVP of Cloud AI at Arm, explains the core features and benefits of the Arm AGI CPU in detail, outlining its three key principles: performance, scale, and efficiency. He details the architecture, including high core density, memory bandwidth, low-latency design, and rack-level deployment, as well as Arm’s broader roadmap for AI data center platforms.
The event also features perspectives from Meta and OpenAI on the scale and infrastructure required to support AI systems serving billions of users, and why system-level efficiency is becoming a primary constraint.
0:05:15 Arm’s history and platform evolution
0:12:25 The shift from traditional AI workloads to agentic AI
0:14:09 Why CPUs remain critical in AI data centers
0:17:32 Power, performance, and scaling constraints in modern infrastructure
0:21:18 Introducing Arm AGI CPU
0:22:40 Software enablement across the Arm data center ecosystem.
0:45:49 Arm AGI CPU architecture and design approach
1:18:54 Arm’s long-term roadmap for AI infrastructure
