Oracle sets fall 2027 target for Arm AGI CPU on its cloud

Oracle cloud customers are set to get another processor option for running business applications and AI agents—but not until fall 2027. Arm said on October 28, 2026, that Oracle outlined plans at Oracle AI World to bring its AGI CPU to Oracle Cloud Infrastructure, known as OCI. The announcement puts a delivery target on their existing collaboration.

The relationship predates this timetable. On June 2, Arm announced at COMPUTEX that OCI was joining the AGI CPU ecosystem. At that point, Oracle executive Mahesh Thiagarajan described exploring how the processor could extend the benefits of Arm infrastructure to next-generation AI agents. The October announcement takes that earlier participation forward with a planned window for offering the processor to Oracle’s cloud customers.

The work behind an agent’s actions

Arm’s rationale starts with the services an AI agent needs to complete a task. Agents depend on applications and databases, then create additional CPU work as they take actions. A CPU, or central processing unit, handles the code and services involved in those steps. Arm identifies several activities behind that demand, rather than treating an agent’s work as a single operation.

• Executing code: running the instructions an agent uses to carry out work.
• Retrieving data and invoking tools: accessing information and calling services needed for a task.
• Coordinating tasks: managing the work across the applications, databases and services an agent depends on.

The intended market is not limited to new AI services. Oracle also plans to offer AGI CPU for general-purpose cloud workloads—the applications customers already run. The companies are positioning the same planned offering around both established computing needs and the extra work created by agents.

With Arm AGI CPU, we plan to extend those benefits across general-purpose cloud workloads and agentic AI, giving customers more high-performance compute options to scale their applications and build new AI services.

Mahesh Thiagarajan, President, Oracle Cloud Infrastructure

Sustained work within power and cooling limits

Arm describes AGI CPU as combining its Neoverse processor cores with high memory bandwidth—the capacity to move data to and from memory—and a design for sustained performance under load. Its stated goal is to complete more useful work within a data center’s power and cooling limits. That is an intended benefit, not a measured result for the planned Oracle service.

The October announcement does not give pricing, customer instance configurations or performance measurements for AGI CPU on OCI. Those details remain unresolved for customers evaluating the future offering. For now, Oracle has supplied a target season and intended workloads, rather than a service customers can provision today.

An addition to an existing Arm foundation

AGI CPU would enter a cloud portfolio that already uses Arm-based processors. Arm says OCI’s Ampere-powered offerings support enterprise applications, containers, web services and data processing. It also identifies NVIDIA’s Arm-based Vera CPUs as generally available on OCI for agentic AI and enterprise workloads. The planned AGI CPU service would broaden that portfolio, not introduce Arm computing to Oracle’s cloud for the first time.

There is an existing software foundation, too. OCI’s investment in Arm spans Oracle Linux, Java, MySQL and widely used cloud-native tools, according to Arm. The company points to that support as a familiar ecosystem for developers as their processor choices grow. It is part of the deployment case alongside the proposed hardware’s performance and efficiency.