What makes Cerebras chips different?

The Wafer-Scale Engine uses a whole silicon wafer as one processor, keeping compute and memory on the same chip to cut data movement, which Cerebras says speeds up token generation.

Where can developers use Cerebras inference?

Through Cerebras' own inference cloud, and from calendar year 2027 on Oracle Cloud Infrastructure, according to an October 2026 announcement.

Cerebras Systems news

AI chip and inference company1 story

Cerebras Systems is a Sunnyvale, California company, listed on Nasdaq as CBRS, that builds the Wafer-Scale Engine, a processor made from a whole silicon wafer with compute and memory on one chip.

A low office building with a red and grey entrance and a red Cerebras sign on the lawn at 1237 East Arques Avenue in Sunnyvale. — Oracle adds Cerebras inference and Intel Diamond Rapids Xeons to its OCI AI chip lineup

AI Chips & Infrastructure27 Oct 2026, 09:00 UTC

Oracle said on 27 October 2026 that OCI will offer Cerebras wafer-scale inference from 2027 and plans X14 instances on Intel's next Xeon, Diamond Rapids.

Summary

At Oracle AI World in Las Vegas on 27 October 2026, Oracle announced two more chip partnerships for Oracle Cloud Infrastructure. Cerebras fast inference, built on its Wafer-Scale Engine, is planned for OCI in calendar year 2027. Intel and Oracle plan a multiyear collaboration to expand Xeon capacity on OCI, including X14 instances based on the next-generation Xeon codenamed Diamond Rapids. Prices, capacity figures and the X14 launch date are not stated.

About Cerebras Systems

Cerebras Systems is a Sunnyvale, California company, listed on Nasdaq as CBRS, that builds the Wafer-Scale Engine, a processor made from a whole silicon wafer with compute and memory on one chip. It sells CS-series systems and a cloud inference service focused on fast token generation for reasoning models and AI agents.

Reviewed 28 Oct 2026

Facts at a glance

Type: AI processors, systems and inference cloud
Listing: Nasdaq: CBRS source
Headquarters: Sunnyvale, California source
Processor: Wafer-Scale Engine, compute and memory on a single wafer-scale processor source
Systems: CS-3 and CS-4 source
CEO: Andrew Feldman, co-founder

Questions people ask

What makes Cerebras chips different?: The Wafer-Scale Engine uses a whole silicon wafer as one processor, keeping compute and memory on the same chip to cut data movement, which Cerebras says speeds up token generation.
Where can developers use Cerebras inference?: Through Cerebras' own inference cloud, and from calendar year 2027 on Oracle Cloud Infrastructure, according to an October 2026 announcement.

Cerebras Systems news

Every story about Cerebras Systems

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