Oracle adds Cerebras inference and Intel Diamond Rapids Xeons to its OCI AI chip lineup

What Oracle, Cerebras and Intel announced

Cerebras and Oracle announced plans on 27 October 2026 to make Cerebras inference available on Oracle Cloud Infrastructure for AI labs, enterprises and developers. The service is planned for calendar year 2027. Cerebras says its Wafer-Scale Engine puts compute and memory on a single wafer-scale processor, which cuts data movement and speeds up token generation for long reasoning chains and multi-step agents. The companies say the partnership builds on Oracle's existing inference options, which include Dedicated AI Clusters in OCI Enterprise AI.

The same day, Intel and Oracle announced a planned multiyear collaboration to expand Xeon computing on OCI. It covers current and future Xeon platforms, including planned OCI X14 offerings based on the next-generation server processor codenamed Diamond Rapids. The companies say they will coordinate roadmaps and processor capacity planning for OCI's expansion.

What each chip is for

The two deals target different parts of an AI workload. OCI president Mahesh Thiagarajan called Cerebras another ultrafast and deterministic option for agents that must respond quickly. Cerebras CEO Andrew Feldman said every step an agent takes needs inference, so faster inference lets AI do more work without making users wait.

Intel's pitch is that CPUs run the applications, data movement and tool calls around a model. Diamond Rapids is planned to bring enhanced Advanced Matrix Extensions with FP8, FP16 and BF16 support for inference software, plus Advanced Performance Extensions. Intel and Oracle say this suits embedding generation, reranking and smaller-model inference next to the services agents use. Intel executive Kevork Kechichian also pointed to hardware-based confidential computing. Current Xeon-based OCI offerings already pair Intel CPUs with built-in AI acceleration, flexible CPU and memory configurations and Oracle Acceleron networking, so customers can scale agent execution separately from the accelerators that serve larger models.

Why it matters

With these two deals, OCI now has announced plans for NVIDIA GPUs and Vera CPUs, Google TPUs, Cerebras wafer-scale systems and Intel Xeons, all within one week. Oracle is positioning OCI as the cloud where customers pick a chip per workload rather than a single accelerator family. The earlier announcements are covered in NVIDIA and Oracle's Vera Rubin expansion and Google TPUs on OCI. More hardware news is in AI Chips & Infrastructure.

What to watch next

Neither release gives prices, capacity, regions or a launch date for X14. Both are described as plans and carry forward-looking statement warnings, so the 2027 timing for Cerebras on OCI is a target, not a commitment. Benchmarks against NVIDIA-based OCI inference are also not stated.

Questions people ask

When will Cerebras inference be available on Oracle Cloud?

Cerebras and Oracle say Cerebras Inference on OCI is planned to be available in calendar year 2027. Pricing is not stated.

What is OCI X14?

Planned Oracle Cloud Infrastructure instances based on Intel's next-generation Xeon processor, codenamed Diamond Rapids. A launch date is not stated.

Why does Oracle offer so many AI chips?

Oracle says AI inference demands choice and that customers want compute matched to each application's performance needs.

Sources

1. Cerebras and Oracle Team Up to Bring Cerebras AI Inference to Oracle Cloud Infrastructure (Cerebras Systems, 27 Oct 2026)
2. Intel and Oracle Plan Expanded Collaboration to Support Enterprise and Agentic AI on Oracle Cloud Infrastructure (Intel, 27 Oct 2026)