VeeamON 2025: Entra ID Backup and the MCP Integration Nobody Expected

VeeamON 2025 just wrapped in San Diego and there were two announcements that caught my attention more than anything else: Veeam Data Cloud now backs up Microsoft Entra ID, and Veeam is integrating with Anthropic’s Model Context Protocol to let AI systems access data stored in Veeam repositories.

Entra ID Backup Is Overdue and Critical

If you manage Azure infrastructure, Entra ID is the identity backbone of everything. Users, groups, roles, applications, service principals: it’s all there, and if it gets corrupted or misconfigured, nothing works.

VDC now supports full Entra ID tenant backups including audit and sign-in logs, with granular restore for users, groups, admin units, roles, and applications. There’s automated scheduling and a change detection feature that compares your live tenant against your backup to flag inconsistencies.

As someone with Azure certifications who’s watched partners struggle with Entra ID misconfigurations, this is a big deal. The number of times I’ve seen a partner accidentally delete a service principal or modify a conditional access policy and have no way to roll it back: that scenario now has an answer.

From a support perspective, the change detection feature is what excites me most. Being able to compare current Entra ID state against a known-good backup turns a “something changed and we don’t know what” investigation into a quick diff. That’s the kind of tooling that reduces mean time to resolution significantly.

The Anthropic MCP Integration Is Wild

This one genuinely blindsided me. Veeam announced that it’s building capabilities to let AI systems securely access data stored in Veeam backup repositories through Anthropic’s Model Context Protocol.

Let me translate what that means: your backup data becomes a data source that AI agents can query. Instead of backup data sitting in cold storage waiting for a restore scenario, it becomes accessible context for AI-driven analysis, compliance queries, and operational intelligence.

The infrastructure implications are significant. MCP is essentially a standardized API that lets AI systems access external data sources. Veeam implementing MCP means they’re treating backup repositories as a structured data platform, not just a recovery target. That’s a philosophical shift. Backup data has intelligence value, not just insurance value.

I’m speculating here, but this could enable use cases like: “Show me all configuration changes across our Entra ID tenant in the last 30 days” or “Identify which VMs had the most disk churn last quarter” without writing KQL queries or parsing logs manually. The AI agent queries the backup data through MCP and synthesizes the answer.

VDC Is Becoming a Platform

Step back and look at the VDC trajectory over the last 18 months:

• February 2024: BaaS for M365 and Azure. Basic backup as a service.
• November 2024: VDC Vault, Storage-as-a-Service with integrated provisioning.
• April 2025: Entra ID backup and AI integration via MCP.

Each release expands what VDC is. It started as a backup service and it’s becoming a data platform. The architecture underneath has to support increasingly diverse workloads, increasingly complex access patterns, and now an AI integration layer that treats backup data as queryable knowledge.

The team building VDC is solving cloud infrastructure problems at scale, and each announcement makes that more visible. The intersection of infrastructure, automation, and cloud-native architecture is exactly where Veeam is heading, and it’s increasingly where I want to be too.

The CrowdStrike Partnership

Also worth mentioning: Veeam announced a strategic partnership with CrowdStrike with two new integrations for Falcon LogScale and Next-Gen SIEM. The pattern is clear. Veeam is positioning itself at the intersection of data protection and security operations, which means deeper integrations, more APIs, and more platform complexity to manage.

I keep saying it — interesting times at Veeam. But yeah, actually interesting this time, not just corporate-blog interesting.