EY has published an analysis comparing the Model Context Protocol (MCP) with enterprise data platforms as foundations for AI-ready data strategies. The central question for CIOs: should integration effort concentrate on a new protocol layer, or on the governed data platforms already running the business?

The piece lands as enterprises move from AI pilots to production workloads that demand reliable, governed access to corporate data. MCP has emerged as a standard mechanism for connecting AI models and agents to tools and data sources. Enterprise data platforms — warehouses, lakehouses and governance stacks — carry the semantic, quality and access-control burden that AI applications inherit.

What is MCP, and why are CIOs hearing about it now?

MCP defines a common way for AI systems to discover and call data sources and tools. Its appeal is speed: a single protocol that lets models reach systems without bespoke integrations for every pairing of model and source. Vendors across the stack have moved to support it.

For enterprises, that translates into a potential reduction in integration overhead — the work that typically stalls AI deployment timelines.

Where do enterprise data platforms still win?

Data platforms remain the system of record for lineage, access policy, quality controls and semantics. AI agents that bypass these controls create audit and compliance exposure. EY's analysis frames the decision not as either/or, but as a question of which layer owns which responsibilities.

The practical pattern: platforms govern and curate; MCP-style interfaces expose that governed data to models and agents.

What questions should CIOs ask before committing?

Key decision points:

• Does the existing platform expose data through interfaces AI agents can consume securely?
• Where will access control and auditability be enforced — platform, protocol layer, or both?
• Which workloads genuinely need the flexibility MCP offers versus stable, governed pipelines?
• What vendor support exists across the current stack for protocol-based access?

The bottom line for enterprise IT leaders

Organizations that treat MCP and data platforms as complements — protocol for connectivity, platform for governance — position AI workloads to scale without duplicating controls. Those that skip the governance question risk faster integrations paired with unmanaged data access.

EY's full analysis is available on the firm's site.