Key Takeaways
- The expanded Oracle and Google Cloud partnership embeds Gemini models directly inside Oracle Fusion Applications and NetSuite, moving an outside AI provider from an infrastructure option into the core application layer.
- Buyers need to treat AI as an ERP selection criterion and ask which company's models power a platform's AI features and who controls the roadmap behind them.
- ERP buyers evaluating a single finalist are now effectively evaluating two vendors, the software company and the AI provider behind its automation features.
- Ignoring the risk of AI ERP vendor lock-in during selection can leave an organization bound to a single AI provider's pricing and release schedule.
AI is moving from an optional infrastructure choice to a built-in layer of the ERP applications.
Case in point: The expanded Oracle and Google Cloud partnership embeds Google's Gemini models directly inside Oracle Fusion Applications and NetSuite.
In other words, the model has moved from an optional infrastructure choice into the application layer itself.
Today, we are exploring what this shift means for ERP buyers and why AI as an ERP selection criterion now deserves the same scrutiny applied to functionality and total cost of ownership. We will also cover how to negotiate around AI ERP vendor lock-in before a contract is signed.
The 2026 Top 10 AI-Enabled ERP Systems Report
Panorama’s experts share their insights on the ERP systems and AI platforms shaping the next phase of ERP evolution.
What the Oracle and Google Cloud Partnership Actually Changes
Embedding an outside AI model inside business software is not new. Microsoft has built Dynamics 365 Copilot on OpenAI's models since 2023, and SAP announced in May 2026 that Anthropic's Claude would extend its Business AI platform and power agents inside Joule. Oracle's own cloud infrastructure already gave customers a choice of outside models, including Cohere and Meta's Llama.
What changed for Oracle is where the outside model sits. Earlier integrations, including Oracle's own infrastructure-layer offering, made an outside model something a developer or administrator had to configure deliberately. Oracle is now embedding Google's Gemini 3.1 Flash-Lite and Gemini 3.5 Flash models natively inside Fusion Applications and NetSuite, so the model shows up as default agent and automation capability built into the application itself.
(Oracle confirmed the details in its own official announcement, and Google Cloud has published its own technical notes on the underlying Gemini 3.1 Flash-Lite release.)
Why AI Depth Deserves a Place in ERP Selection Criteria
Selection teams have spent two decades evaluating ERP platforms on module depth and total cost of ownership. Identifying the best ERP software meant comparing core financial and supply chain functionality side by side, and the depth of AI capability rarely factored into the decision.
Fast forward to today: ERP buyers are essentially evaluating two vendors (the ERP vendor and the AI vendor), even though only one name appears on the contract. Four questions now belong in every ERP selection process:
- Model strategy: Is the AI model in the infrastructure layer or in the core applications?
- Data portability: If the underlying AI provider changes, can the organization's historical data and trained configurations move with it?
- Pricing exposure: Does AI usage carry its own consumption-based pricing layer that can shift independently of the core ERP license?
- Roadmap dependency: How much of the vendor's AI roadmap depends on a partner's release schedule rather than its own?
Buyers already know how to run this kind of interrogation. The same scrutiny applied during an ERP or SCM software evaluation, now belongs in every AI feature demo.
The Vendor Lock-In Risk Hiding Inside AI Roadmaps
AI ERP vendor lock-in describes a narrower risk than the licensing lock-in ERP buyers have negotiated against for years. Traditional ERP contracts lock in the database and the customizations built on top of it. This new layer binds the organization to the AI models and data pipelines a vendor's agents depend on, along with the custom logic built on a specific provider's application programming interface.
An organization that builds custom Fusion agents around Gemini today adds a dependency on Google's pricing and roadmap alongside Oracle's — the same tradeoff that already existed with Fusion's other third-party model options (Cohere, Meta). Switching model providers later will likely require some reconfiguration of the agent, though Oracle hasn't published details on how much rework that involves.
Vendors rarely disclose this dependency on their own, so buyers benefit from raising the question directly during due diligence.
Expert Insight
Our ERP system selection consulting team has found that organizations skipping a formal AI capability assessment during due diligence usually discover their AI ERP vendor lock-in only after a renewal forces a pricing conversation. Raising these questions earlier gives an organization real leverage to negotiate before signing.
How to Build Defenses Against AI ERP Vendor Lock-In During Selection
1. Require Written Model Sourcing Disclosure
Ask each ERP finalist to name, in writing, which AI features run on models the vendor built and trained itself and which run on a partner's model, such as Google's Gemini inside Fusion Applications and NetSuite. A verbal assurance during a demo is not something a finalist scoring matrix can rely on once budget approval depends on the answer.
2. Negotiate Data Portability and Exit Terms Before Signing
Ask what happens to trained AI configurations and historical outputs if the organization switches ERP platforms or if the ERP vendor switches AI partners. This protects the organization from a renewal negotiation that surfaces AI ERP vendor lock-in for the first time.
3. Bring In Independent Oversight Before the AI Layer Goes Live
Once a platform is selected, an experienced ERP implementation consultant can validate that the AI agent layer meets the requirements defined during selection, rather than leaving finance or operations teams to discover integration gaps only after go-live.
Learn More About ERP Vendors' AI Strategies
The Oracle and Google Cloud partnership will not be the last time a vendor moves an outside AI model from its infrastructure layer into its core applications. AI as an ERP selection criterion deserves a permanent place in every RFP.
Panorama's independent ERP selection consultants help organizations build these questions into the RFP and vendor scoring from the start. Contact us below to learn more.
FAQs About ERP Vendors' AI Strategies
How does the expanded Oracle and Google Cloud partnership change ERP AI evaluation?
Buyers evaluating Fusion Applications or NetSuite should ask which capabilities run on Oracle's own models and which run on Google's Gemini, since the answer affects pricing and long-term roadmap control.
Why does AI as an ERP selection criterion matter for buyers beyond IT?
AI capability now shapes forecasting and reporting across nearly every ERP module, and its depth affects daily operations as much as core financial functionality does. Scoring AI capability formally during finalist evaluation surfaces gaps before the contract stage rather than after go-live.
What questions belong in an RFP about a vendor's AI strategy?
Finalists should confirm how AI usage is priced and what happens to trained configurations if either party changes providers, since these answers rarely appear unless a buyer asks directly.
Does AI ERP vendor lock-in affect organizations that already run Fusion Applications or NetSuite?
Yes. Existing customers now have Gemini-powered features arriving inside applications they already pay for, often without a separate purchase decision. AI ERP vendor lock-in does not wait for a new selection cycle to appear. Organizations already live on these platforms should audit which workflows depend on the embedded model and raise pricing and portability questions at the next contract renewal.
Should organizations delay ERP selection until AI roadmaps stabilize?
Waiting is rarely the right choice, since ERP replacement cycles run five to ten years and vendors will keep updating AI partnerships throughout that time regardless of when a buyer selects. A stronger approach is building AI evaluation criteria into the selection process now and revisiting model sourcing at each renewal.
