Oracle Adds Google's Gemini to Fusion Apps and NetSuite; Shares Jump 8.4%

Oracle is putting Google's Gemini models inside the enterprise software its customers use to run their businesses, deepening a partnership with a company it competes against directly in the cloud infrastructure market. Oracle shares rose as much as 8.4% to $127.64 following the announcement, according to Bloomberg.

The company said Thursday it is expanding its Google Cloud partnership to bring Gemini into Oracle AI Agent Studio for Fusion Applications, with the models also powering embedded AI features across Oracle Fusion apps and NetSuite.

Two Gemini models are involved. Gemini 3.1 Flash-Lite is positioned for low-cost, high-efficiency tasks. Gemini 3.5 Flash is aimed at heavier reasoning work and creative output, including video and presentation generation. Both join a roster of models Oracle already makes available to customers, including options from Cohere and Meta.

NetSuite alone counts more than 44,000 customers, according to the company, giving Google a direct channel into a large base of existing enterprise software users without having to sell its models separately.

"Organizations need the flexibility to choose the AI model best suited to each problem," said Chris Leone, Oracle's applications-development chief. He said Oracle Fusion converts that model reasoning into action "through governed workflows, approvals, and transactions."

A rival becomes a supplier

Oracle and Google compete for cloud infrastructure customers, yet their partnership has expanded from raw compute to embedded agentic AI. Notably, Oracle chose Google's models for this layer of its application stack rather than models from OpenAI, despite Oracle's roughly $300 billion infrastructure commitment to that company. Pricing for the Gemini integration within Oracle's apps has not been disclosed.

The move reflects a broader pattern among large enterprise vendors: rather than committing to a single foundation model, companies are offering multiple models and letting customers pick one per task. Oracle's decision to add a direct cloud competitor's models to its own product line is an unusually explicit version of that strategy.

Skepticism from analysts

Not every observer is convinced the integration solves enterprise AI's harder problems. "It doesn't necessarily look as glittery as it sounds. There are challenges under the hood," Gartner analyst Balaji Abbabatulla told The Register, describing the plan as something to approach with caution.

The central concern is accountability. If an autonomous agent embedded in a workflow makes an incorrect decision at machine speed, errors can compound before a human notices. It remains unclear who bears responsibility when that happens — Oracle, Google, or the customer deploying the agent. Oracle points to its monitoring and audit tooling as a mitigation, but Abbabatulla said no vendor has fully answered the liability question.

What this means

The stock reaction suggests investors reward Oracle for diversifying its AI supply chain rather than betting everything on one partner, even as the company carries a heavy debt load tied to its data center buildout and its OpenAI commitment. For Google, the deal is a distribution win: access to tens of thousands of NetSuite and Fusion customers without having to compete for them directly. But the unresolved liability question — who answers for a bad decision made by an autonomous agent — is the real test of whether multi-model enterprise AI is durable or just a good headline. Oracle's governance and audit tools will need to prove themselves in practice, not in press releases.