Cerebras wafer-scale AI inference is coming to Oracle Cloud in 2027

Cerebras Systems and Oracle announced on October 27 that they plan to make Cerebras fast inference available on Oracle Cloud Infrastructure (OCI) for AI labs, enterprises and developers building interactive, intelligent applications, with availability planned in calendar year 2027. The announcement, issued from Las Vegas, builds on Oracle's existing infrastructure and AI inference capabilities, including Dedicated AI Clusters in OCI Enterprise AI.

Why inference speed matters more now

As AI applications take on more complex, multi-step work, inference speed becomes critical to the user experience: each additional model call adds latency, which can slow agentic workflows that reason, check results and act across multiple steps. “People want AI that can do valuable work quickly,” said Andrew Feldman, co-founder and CEO of Cerebras. “Every step an agent makes requires inference, whether it's reasoning, checking its work, or taking action. The faster the inference, the more work AI can do without making users wait.”

The wafer-scale design

Designed for high-speed AI inference, the Cerebras Wafer-Scale Engine integrates compute and memory on a single wafer-scale processor. By reducing data movement and inter-chip communication, the design supports rapid token generation for long reasoning sequences and multi-step agentic workflows, where tokens are the pieces of text a model produces.

Oracle's view and the timeline

“AI inference demands choice,” said Mahesh Thiagarajan, president of Oracle Cloud Infrastructure. “As agents take on more complex work, customers demand compute that matches each application's performance needs. Bringing Cerebras' inference to Oracle Cloud Infrastructure will add another ultrafast and deterministic option to our broad compute portfolio.” The companies said Cerebras inference on OCI is planned to be available in calendar year 2027; the release carries a forward-looking disclaimer noting that features, timing and pricing may change, and it does not name deployment scale or pricing.