7th  February, 2026    

VCF 9 and the Model Context Protocol: Building the Highway for AI-Native Operations 

VMware has been the backbone of enterprise infrastructure for more than two decades, and one of its enduring strengths has been the consistency and depth of its automation and integration interfaces. As enterprises move toward AI-driven operations, this foundation becomes even more critical. With the introduction of Model Context Protocol (MCP) and the release of VMware Cloud Foundation (VCF) 9, Broadcom is not merely aligning VMware with this shift — it is positioning VMware as an AI-native private cloud platform.

VMware Cloud Foundation (VCF) 9 was unveiled at VMware Explore 2024 as a major evolution of the private cloud. Broadcom has positioned VCF 9 as an AI-native platform with integrated VMware Private AI Services. By incorporating native support for the Model Context Protocol (MCP), VCF 9 makes AI governance, operational context, and ecosystem integration first-class capabilities. The VCF Services Platform acts as a centralized broker, securely exposing infrastructure services to AI systems in a standardized, governed manner. In effect, VMware is now delivering the 'highway system' for AI-driven operations as a built-in capability, rather than leaving customers to assemble the complex integration layers themselves.

Importantly, MCP does not replace VMware’s existing APIs, SDKs, or automation tools such as PowerCLI. Instead, it formalizes how these capabilities are discovered, described, and consumed. vCenter, NSX Manager, SDDC Manager, and ESXi already expose rich REST APIs and SDKs. MCP sits above these interfaces, defining what tools exist, which operations they support, what inputs are valid, and what outputs look like — making VMware infrastructure understandable and safely consumable by AI agents without breaking enterprise guardrails.

A useful way to think about MCP is as traffic infrastructure rather than a new vehicle. The APIs and automation tools remain the cars. MCP defines the lanes, traffic signals, and checkpoints that keep traffic flowing safely. Crucially, MCP does more than provide the “pedals” to execute actions — it provides the dashboard. By exposing logs, health signals, alerts, topology metadata, and performance metrics as standardized context, MCP allows AI systems to understand the state of the environment before acting, rather than making blind or risky decisions.

This emphasis on context is what makes MCP especially powerful in enterprise environments. Infrastructure decisions should never be driven by isolated commands alone. An AI agent needs to know whether a host is already under resource contention, whether an NSX component is degraded, or whether a proposed change might violate an operational policy. By combining action interfaces with real-time operational context, MCP enables informed automation instead of brute-force execution.

Native MCP integration through the VCF Services Platform is also significant from a governance and supportability perspective. VMware environments are typically bound by ITIL processes, strict role-based access control, and compliance requirements. MCP enforces structure by design. Every AI-initiated action maps back to a supported VMware API, runs under controlled identities, respects platform permissions, and produces auditable outcomes. This makes AI-driven operations not only powerful, but defensible and supportable in regulated enterprise environments.

At the same time, VMware’s approach remains open rather than closed. While VCF 9 delivers native MCP capabilities, the ecosystem continues to support custom-built and partner-provided MCP servers and tools. Organizations can extend MCP to cover bespoke workflows, integrate third-party platforms, or introduce additional policy layers — provided these extensions are rigorously tested, validated, and aligned with VMware-supported interfaces. Even when leveraging community or partner solutions, execution must remain within VMware’s supported boundaries, relying on certified APIs and tooling to preserve stability, security, and operational support.

This openness is further amplified by VMware’s mature automation ecosystem. PowerCLI modules, vSphere Automation SDKs, NSX APIs, Terraform providers, Ansible collections, and long-standing GitHub repositories already encapsulate years of operational best practices. However, community-driven scripts and non-official integrations should be treated as source material rather than turnkey solutions. They must be reviewed, tested in non-production environments, security-validated, and operationally hardened before being elevated into MCP tools. MCP does not discard these investments; instead, it elevates validated automation into structured, AI-consumable capabilities.

From a practical adoption standpoint, MCP in VMware environments should begin with low-risk, high-value use cases. Read-only operations such as inventory discovery, capacity analysis, health validation, and configuration drift detection are natural starting points. Over time, carefully controlled write operations — including VM lifecycle management, maintenance workflows, and network validation — can be introduced with policy enforcement and approval gates. High-risk actions remain intentionally constrained, ensuring AI augments human operators rather than bypassing them.

Ultimately, VCF 9 represents more than a platform upgrade. By embedding MCP natively through the VCF Services Platform, VMware is delivering on its Private AI vision — ensuring AI is no longer an external add-on bolted onto infrastructure, but a first-class consumer of the platform itself. For architects and operations teams, the message is clear: build on VMware’s native MCP foundation, leverage existing automation investments, and move toward intelligent operations without sacrificing trust, control, or compliance.