Oracle Puts ChatGPT and Codex Inside Its Business Software, and Will Offer Google TPUs on Its Cloud

Oracle announced an expanded OpenAI collaboration at Oracle AI World on October 27, 2026, embedding ChatGPT and Codex across its applications, and said Google Tensor Processing Units will come to Oracle Cloud Infrastructure in 2027.

Product updatesOracleOpenAICodex

Oracle has announced a major expansion of its partnership with OpenAI, putting ChatGPT and Codex directly inside the business software that runs much of corporate back-office work. The announcement, made at Oracle AI World in Las Vegas on October 27, 2026, was paired with a second, striking piece of news: Google Tensor Processing Units, or TPUs, will be available on Oracle Cloud Infrastructure, or OCI, starting in calendar year 2027.

Taken together, the two announcements describe an Oracle that wants to be the place where frontier AI meets enterprise systems, and where customers can choose between rival chip architectures without leaving its cloud. For OpenAI, the deal is another step in a push into the enterprise market, landing one day after its key competitor Anthropic filed its public offering paperwork. For Google, it is the first time its custom AI chips will be offered on a hyperscale cloud platform other than its own.

Oracle chief executive Clay Magouyrk said OpenAI is transforming how enterprises solve business problems, from chat to coding to analytics to autonomous agents, and that the new product suite unites those capabilities with Oracle applications, data and cloud infrastructure. OpenAI president and co-founder Greg Brockman said more businesses are choosing OpenAI as it becomes easier to put AI to work through the tools and systems they already use, and that the expanded partnership puts frontier AI into everyday operations so teams can build software faster, strengthen security and take on more complex knowledge work.

Run your business through ChatGPT

The centerpiece is Oracle AI Workspace, a single user experience that brings OpenAI frontier models together with Oracle applications, enterprise data and cloud services. Through plugins in ChatGPT on Oracle AI Workspace, employees can access information, complete approved business workflows and take action alongside Oracle systems. It is designed to connect Oracle Fusion Applications, Oracle NetSuite, Oracle AI Database, OCI and other enterprise tools and data sources.

The idea, as Oracle and OpenAI describe it, is to set the standard for what they call the agent-ready enterprise: transforming Oracle applications for agents and bringing them into a unified experience in ChatGPT. People can describe business goals, and the plugins in ChatGPT will bring together the right Oracle tools and information to help them accomplish those goals. Oracle stresses that the workspace keeps AI inside the flow of enterprise work while maintaining the governance, approvals and controls that enterprise operations require.

Build applications with Codex, test security with frontier models

A second product, Oracle App Composer, targets application development. It lets employees turn ideas into enterprise applications without requiring coding expertise to build or maintain them. Users build applications with Codex, while Oracle manages deployment and operations on OCI. App Composer connects applications to approved Oracle data and workflows and provides access to OCI services, including OCI Kubernetes Engine and Oracle Autonomous AI Database. It also applies enterprise identity, permissions, networking, audit and application lifecycle controls, aiming to take AI-generated applications from idea to secure enterprise deployment at scale.

App Composer also supports APEXlang, a new Oracle technology that is meant to give customers the speed of generative development without producing opaque application code that is difficult to maintain.

A third offering, Oracle AI Security Mesh, is a managed service that uses OpenAI cyber models to help customers find and fix vulnerabilities in their applications and infrastructure across OCI, other clouds and on-premises environments, before attackers can exploit them. The service combines OpenAI frontier cyber models with Oracle testing tools, independent validation, testing infrastructure and operational safeguards managed by Oracle, so customers do not need to build and operate their own agentic defensive and offensive systems. Customers define where testing is authorized and control its scope, while Oracle manages safeguards intended to reduce the risk of outages or business disruption during testing.

A fourth product is aimed at smaller companies. The Oracle AI SMB Bundle is a free-to-start packaged offering that provides 12,000 dollars in Oracle cloud credits each year. It brings together ChatGPT Business for daily work, Oracle AI Workspace for accessing business applications and data through ChatGPT, Codex and Oracle App Composer for building apps, and OCI for deploying them. It also includes NetSuite QuickStart, which provides a foundation for managing core operations such as invoicing and cash flow. Terms, conditions and restrictions apply.

Google TPUs arrive on OCI in 2027

Separately, Oracle and Google announced that Google TPUs will be available through OCI starting in calendar year 2027, ramping over a multi-year period. That makes Oracle the first hyperscale cloud provider outside Google to offer TPUs. Oracle will provide the power, cooling and datacenter networking to deploy them at scale, while the chips are designed to support large-model training, including dense and mixture-of-experts models, as well as production inference for reasoning and long-context applications.

The announcement did not name a specific TPU generation, price, region or capacity figure, and those details matter for judging how competitive the offering will be against the Nvidia hardware that dominates AI training today. What is clear is the strategic signal: OCI will offer multiple leading AI compute architectures in one cloud. Mahesh Thiagarajan, president of Oracle Cloud Infrastructure, said bringing Google TPUs to OCI gives customers access to multiple leading AI compute architectures in one cloud.

The context is a cloud market in which the largest providers are racing to secure AI compute capacity and to differentiate on chip choice as well as model choice. Oracle already works closely with both OpenAI and Google on infrastructure, and the announcements at Oracle AI World extend that positioning from raw capacity into the applications layer: not just hosting models, but embedding them in the software where business decisions are made.

As with all product announcements of this kind, the usual caveats apply. Oracle notes that its announcement outlines general product direction, is not a commitment to deliver any material, code or functionality, and that development, release, timing and pricing may change at its sole discretion. The TPU timeline is a calendar-year 2027 start with a multi-year ramp, not an immediate launch, and the workspace, composer and security products will be judged on how they handle enterprise data controls in practice rather than on stage.

Still, the shape of the deal is notable. One of the largest enterprise software vendors is betting that the next interface for business software is a chat window powered by a frontier model, that employees will build applications by describing them to a coding agent, and that security testing will increasingly be done by AI agents operating under human-defined scope. And on the infrastructure side, a chip that was once a Google-only advantage is about to become a product any OCI customer can rent.