Clay Magouyrk, Oracle’s co-chief executive, joined OpenAI chief executive Sam Altman on stage in Las Vegas for a keynote at 12:40pm ET Tuesday, as Oracle set out a broader plan to put OpenAI’s tools into the software businesses already use.
At Oracle AI World, Oracle announced an expanded collaboration with OpenAI that will bring ChatGPT and Codex into its enterprise applications, with the aim of making databases more actionable. Magouyrk said OpenAI was changing how companies tackle business problems, spanning chat, coding, analytics and autonomous agents.
The first product, Oracle AI Workspace, will allow employees to use ChatGPT plugins to find information and carry out approved workflows across Oracle Fusion Applications, Oracle NetSuite, Oracle AI Database and Oracle Cloud Infrastructure (OCI). Oracle App Composer will enable employees to build applications with Codex.
Greg Brockman, OpenAI’s president and co-founder, said more businesses were choosing OpenAI as the company made it easier to use artificial intelligence through the tools and systems they already had.
Oracle paired the software announcement with a wider hardware pitch. It said Google’s Tensor Processing Units, OpenAI’s Jalapeño chip and Cerebras hardware would become accessible through its cloud infrastructure. Google’s Tensor Processing Units will be available through OCI in calendar year 2027, the first occasion on which Google’s custom chips will run on another hyperscaler’s cloud.
The company also plans to offer Cerebras’ artificial intelligence inference, the computing used when a model produces responses, on OCI. Oracle will begin installing the equipment in the first half of 2027, and availability of Cerebras AI inference on OCI is planned for calendar year 2027.
Andrew Feldman, Cerebras’ co-founder and chief executive, said every action an agent takes requires inference, whether it is reasoning, checking its work or acting. Cerebras’ Wafer-Scale Engine puts compute and memory on one processor, and the Oracle offering is aimed at AI labs, enterprises and developers building interactive applications.
Magouyrk also named Cerebras and SambaNova as providers of low-latency inference options. Oracle did not disclose the amount of Cerebras hardware it intends to deploy, and said timing, features and pricing may change at the discretion of Oracle and Cerebras.
Mahesh Thiagarajan, president of Oracle Cloud Infrastructure, has watched OpenAI’s GPT-6 Astra model being trained at Oracle’s Stargate data centre campus in Abilene, Texas. He said customers want computing capacity suited to each application’s performance needs as agents take on more complex work.
Oracle also introduced AI Security Mesh, which uses OpenAI cyber models and is intended to help customers identify and fix vulnerabilities across OCI, other clouds and on-premises environments. Its Oracle AI SMB Bundle is free to start and includes $12,000 in Oracle cloud credits each year.
Investors responded swiftly after the announcements: Cerebras stock rose by up to 10%, while Oracle shares were up nearly 5% on Tuesday.
Comments (0)