Microsoft has retired the Microsoft Deployment Toolkit. Two separate accounts of the shutdown — one from The Register, one from TechRadar Pro — agree on the substance and on the reaction: a free, twenty-plus-year-old tool that thousands of IT departments still run in production was ended with no transition period, and the people who ran it are not happy about the replacements they were handed.
Here is what actually changed, and what the organizations left holding custom task sequences can realistically do about it.
What Microsoft did
MDT dates to 2003. Microsoft announced its deprecation in December 2024, with October 2025 named as the original discontinuation deadline. The retirement itself was confirmed in January 2026 — and it was immediate. Microsoft’s own wording, as quoted by TechRadar, was that it “is announcing the immediate retirement of Microsoft Deployment Toolkit (MDT).”
The consequences are narrow to state and wide in effect. Existing deployments will continue to function. Everything supporting them stops: no further fixes, updates, security patches or enhancements; no compatibility guarantees with future Windows releases; and download packages that may be removed from official distribution channels. There is no direct upgrade path to anything.
The Register noted that when Microsoft was asked to explain the abrupt retirement, no reasoning was provided.
Configuration Manager environments have specific cleanup to do. Per the guidance TechRadar cited: “The MDT Integration with CM and Standalone is no longer supported with Configuration Manager,” and “Customers should remove MDT TS steps, followed by removing MDT integration, to avoid TS corruption.” Leaving the integration in place is not a neutral decision — it puts live task sequences at risk.
Why administrators are angry
The complaints in both write-ups were not nostalgia. They were about properties the replacements do not have.
One commenter quoted by The Register claimed “there are security flaws that Microsoft chose not to fix. Instead they retired the product immediately.” Another offered a blunt three-part theory of why MDT was never going to survive: “It is free, it does not invade the customer’s privacy (no telemetry), it does not force customers to Azure cloud.” A systems administrator was more resigned: “Not entirely unexpected, but that definitely closes a chapter on some of my early-career knowledge.”
Reddit reaction cited by TechRadar landed on the same three points — free, no telemetry, no forced Azure adoption.
It helps to remember what MDT actually did. It supported zero-touch (fully automated), user-driven (prompted), and lite-touch (bootable media) installations. That last mode is why it lasted. It imaged bare metal from a USB stick, and it did not care whether the machine had ever seen a cloud tenant.
Microsoft’s recommended destinations are Windows Autopilot for cloud deployments or Configuration Manager’s on-premises OS deployment feature. Many administrators quoted in the coverage said both look unnecessarily complex relative to what they were doing with MDT.
The gap nobody at Microsoft is filling
Marc Roth, a fractional CRO who works with Swimage Endpoint Resilience, laid out the practical consequences in a LinkedIn post aimed at IT teams facing the migration: no more updates, bug fixes or security patches; no compatibility guarantees with future Windows releases; download packages pulled from Microsoft Learn and Intune distribution channels; no in-place upgrade path; and Microsoft’s guidance to move to Autopilot or SCCM OS Deployment.
His framing of the risk is that nothing breaks on day one. “While existing MDT deployments will continue to run,” he wrote, “each day they operate is a step further from support and security, leaving organizations vulnerable if issues arise.”
SCCM remains relevant, Roth acknowledged, but Microsoft is pushing toward cloud-native, zero-touch deployment through Autopilot and Intune. That shift, in his words, “creates challenges for organizations that require traditional bare-metal imaging, disconnected environments, hybrid-join devices, or customized task sequences—areas where Autopilot may not perform well.” Many IT teams, he noted, are left managing numerous custom task sequences with no clear long-term solution.
That population is not small: manufacturing floors and clinical sites with unreliable connectivity, hybrid-join fleets that never fit a cloud-only model cleanly, and shops with years of tuning invested in sequences that would have to be rebuilt from scratch to move to Autopilot.
The lateral-move argument
Roth’s case for Swimage is that it is a replacement rather than a re-architecture. Unlike an MDT-to-Autopilot migration that requires rebuilding deployment logic, moving to Swimage “retains the familiar imaging paradigm while enhancing automation and security,” and can be operational in days rather than the months a full Autopilot re-architecture would take.
He highlighted three things in particular: a workflow engine that monitors each step of a deployment, so a failing step can be restarted, repaired or skipped without restarting the entire deployment; features aimed at the current threat landscape, including automated ransomware and malware recovery, encryption-aware imaging, snapshot and rollback, and compliance enforcement; and the ability to operate fully offline, for remote workforces, disconnected sites and disaster recovery.
Swimage’s own published material supports that shape. The company says it performs automated OS repair and rebuilds in under 15 to 30 minutes, provisions devices zero-touch with applications and security settings already configured, and runs offline with no network connectivity required. It eliminates and recovers from ransomware and malware, converts between encryption types, and can rebuild and recover a PC without removing encryption — including non-BitLocker encryption. Full drive snapshots allow rollback if a deployment goes wrong, a Splashlock feature locks the system during deployment, real-time dashboards report on deployment activity, and user data, profiles and personalized settings are preserved through imaging.
On Windows 11 migration specifically, Swimage says it restores all functionality — applications, settings and data — in about an hour, “even for the hardest to reach remote user,” using an offline deployment method and a local recovery cache that permits updates and repairs with no network requirement. Against Autopilot, its stated differentiators are faster setup and maintenance, systems fully built and operational at first login, the ability to install any type of application from any installation source, and support for traditional on-premise domains, hybrid environments or Azure-only setups. The company cites more than 25 years behind the platform.
What to do this quarter
Three things are worth doing regardless of which direction an organization eventually takes.
Inventory the MDT dependency honestly — not just the deployment share, but the task sequences, drivers, scripts and runbooks built around it, and the sites that image offline. That inventory is the input to every subsequent decision.
If you run Configuration Manager, remove the MDT task sequence steps and then the MDT integration, in that order, to avoid task sequence corruption. This is the one piece of concrete remediation Microsoft’s guidance calls for.
Then decide the sequencing, not just the destination. Autopilot and Intune may well be the right three-year answer. But MDT is unsupported today, and the interim needs a supported answer that does not require rebuilding every task sequence before the first machine can be imaged.
Standing still is the only option that is definitively wrong. An unsupported, unpatched deployment toolchain is a compliance conversation waiting to happen, and it gets harder to defend with every Windows release Microsoft ships without checking whether MDT still works.
Sources: The Register, “Microsoft euthanizes ancient deployment toolkit” (January 2026); TechRadar Pro, “Microsoft shutters fan-favorite deployment platform to the dismay of many”; Marc Roth on LinkedIn; swimage.com.