Oracle and Cerebras announced on October 27, 2026, that they plan to offer Cerebras AI inference through Oracle Cloud Infrastructure (OCI) during 2027. The planned service targets AI labs, enterprises and developers building interactive applications and multi-step agents.
What Oracle and Cerebras plan for OCI
The companies describe the offering as another compute option within OCI. Oracle’s existing AI infrastructure includes Dedicated AI Clusters in OCI Enterprise AI. The planned Cerebras service would join that broader portfolio.
This would be a separate OCI offering from the Cerebras-powered OpenAI API tier in NeoTeo’s earlier report on OpenAI’s Ultrafast tier.
Why the companies are targeting multi-step AI work
Inference is the stage when an AI model generates an answer or other output. An agent that reasons, checks its work and takes action may make several model calls in sequence. Cerebras and Oracle say each additional call can add latency, the delay before a response, which can slow interactive applications.
The companies therefore point to workflows where users or software need repeated responses as a target for the planned service. The announcement names AI labs, enterprises and developers building interactive, intelligent applications; it does not describe a general-purpose replacement for OCI’s existing AI infrastructure.
Cerebras’s architecture and the 2027 plan
Cerebras says its Wafer-Scale Engine integrates computing and memory on one wafer-scale processor. The company says that design reduces data movement and communication between chips, and connects it to rapid inference for long reasoning sequences and multi-step agents.
Cerebras and Oracle’s announcement sets planned OCI availability during calendar year 2027. Bloomberg reported a more specific plan for deployment to begin in the first half of 2027. The companies’ announcement says the development and timing of the described functionality may change and does not commit them to deliver it.
