AI Interoperability Becomes the New Battleground for Enterprise Control
Enterprise software partnerships are being reshaped by AI interoperability standards, but not in the liberating way many CIOs hoped. As large vendors race to become the orchestration layer for agentic AI, they are packaging interoperability as a feature while embedding new forms of vendor lock-in risks. Technically, the landscape is maturing: open protocols make it possible for AI agents to call data and capabilities across multiple platforms. Commercially, however, each major provider wants to own the runtime, governance, and roadmap for cross-vendor agents. That tension is redefining SAP–Oracle competition and forcing enterprises to interpret conflicting messages about openness, integration, and control. The result is a quieter, contract-driven struggle over who monetises AI workloads and who dictates the guardrails for automation that cuts across ERP, CRM, analytics, and development environments.
SAP’s Joule Studio 2.0: Open by Design, Contained by Policy
SAP is positioning Joule Studio 2.0 as a hub for AI interoperability, promising agents that connect across SAP and non-SAP systems. Native support for Model Context Protocol and A2A hints at genuine cross-platform capabilities, enabling agents to span hybrid landscapes and ingest real-time data from multiple sources. Extensibility is marketed as a core principle: customers can add tools, workflow steps, or custom code and hook into non-SAP applications. Yet SAP’s new API policy reveals a different dimension. Analysts argue it is structured to control access to not just data but business capabilities, especially for third-party AI platforms that want to orchestrate agents over SAP processes. Combined with SAP’s partnership to embed Anthropic’s Claude models inside its own Business AI Platform, the strategy looks less like neutral openness and more like a carefully curated walled garden wrapped in interoperability language.
Oracle–Samsung: Operational Gains Today, Tighter Gravitational Pull Tomorrow
Oracle is pursuing a contrasting route, illustrated by its collaboration with Samsung Electronics on software standardisation for semiconductor operations. By supplying its Java SE Universal Subscription, Oracle is promising Samsung engineers structured access to the latest security patches, enhanced technical support, and a more secure, standardised development environment. The pitch is about operational gains: simplified IT operations, streamlined licence management, and reduced operational risk as Samsung’s foundry and device businesses scale globally. Unlike SAP’s overt focus on AI interoperability, Oracle’s strategy here emphasises proprietary reliability over open experimentation. Yet the dynamic is similar: once development, security, and compliance processes are deeply bound to a subscription stack, switching costs climb. The more value Oracle creates in these embedded environments, the harder it becomes for enterprises to move critical workloads or agent runtimes elsewhere without significant disruption.
Openness vs Tight Integration: Why Both Paths Increase Vendor Lock-In
Enterprises now hear two narratives about AI integration. One celebrates openness: standards-based connectors, model-agnostic agents, and bi-directional data sharing across clouds and applications. The other champions tight integration: deeply optimised, secure environments where AI is pre-wired into core operational systems. In practice, both paths can increase vendor lock-in. Open APIs and interoperability standards are often mediated by licensing terms, rate limits, and premium tiers that determine what cross-vendor agents can actually do in production. Highly integrated offerings, meanwhile, make it economically and operationally painful to decouple AI workflows from the platform that hosts them. The technical barriers to interoperability are falling, but control over runtime costs, governance mechanisms, and roadmap priorities remains squarely in vendor hands. For IT leaders, the real integration question is less “Can we connect?” and more “Whose rules govern the connections we rely on?”
How Enterprises Should Negotiate Freedom in the AI Agent Era
As SAP, Oracle, and peers race to become the default home for cross-vendor agents, enterprise software partnerships must be evaluated through a contract and governance lens, not just a technical one. Analysts warn that choosing where to build the first major cross-vendor agent effectively sets the pricing and control structure for the entire AI estate for years to come. To preserve freedom, enterprises should demand clarity on API usage rights, surcharges for third-party AI environments, and ownership of audit trails and telemetry. They should also avoid assuming that a single provider must be the permanent centre of gravity for agentic AI. Instead, architectures and contracts should be designed for portability: clear exit paths, data and model abstraction layers, and negotiated rights to run agents from multiple platforms. In the AI era, interoperability promises are useful—but negotiable control is what actually protects enterprise autonomy.