Windows Server 2016 support ends January 12, 2027

What actually stops on that date

Three things end together: security updates for newly discovered vulnerabilities, bug fixes for reliability problems, and technical support from Microsoft when something breaks. Nothing switches off. Your domain controllers, file servers and line-of-business apps carry on. The difference is that from January 13 onward, every new CVE affecting the platform is permanent on that machine unless you have paid for Extended Security Updates.

Why six months is the right time to act, not twelve weeks

Server migrations are rarely limited by the OS. They are limited by the application sitting on top of it, the vendor who has to certify the new platform, and the change window you can actually get approved. Six months is enough to inventory, test and schedule. Six weeks is enough to panic. The organisations that handle these deadlines calmly are the ones that started while the countdown still looked comfortable, which is the entire argument for tracking lifecycle dates as a routine rather than as a fire drill.

Your four options, honestly compared

In its February 2026 announcement Microsoft set out four realistic paths. They are not equally good for every workload.

• Upgrade in place or rebuild on Windows Server 2025. The cleanest long-term answer. A fresh build is usually safer than an in-place upgrade on a server that has been running since 2016, because it leaves years of configuration drift behind.
• Migrate the workload to Azure. Worth considering when the hardware underneath is also aging, since it turns two refresh problems into one project. Check the licensing implications before assuming it is cheaper.
• Buy Extended Security Updates. Available for up to three years after the deadline. Treat this as a bridge with a meter running, not a destination, because the cost typically climbs each year and you still have to migrate eventually.
• Retire the workload. The option teams forget. Some 2016 servers exist only because nobody has questioned them since 2018. Ask what breaks if you switch it off, and occasionally the honest answer is nothing.

A six-month plan that actually fits

• July to August: inventory every Windows Server 2016 instance, including the forgotten ones. Record what runs on each and who owns it.
• September: confirm application and vendor support for Windows Server 2025. This is where surprises live, so do it before you commit to a path.
• October: pick a path per workload and get budget approved. Buy ESU now if any workload clearly cannot move in time.
• November: migrate the straightforward servers and keep the difficult ones for a dedicated window.
• December to early January: finish the hard cases and leave the final fortnight clear. Nothing good gets scheduled in deadline week.

Do not stop at the operating system

A server that hits its OS deadline is usually old enough that the hardware underneath is close to its own End of Service Life, and vendor support dates rarely line up neatly with Microsoft's calendar. While the inventory is open, capture both dates for each box. It costs nothing extra now and stops you from migrating an OS onto hardware that goes unsupported eight months later. On typical lifespans, see how long servers really last.

Related deadlines worth noting

Windows Server 2016 is not arriving alone. SQL Server 2016 left extended support on July 14, 2026, so a 2016-era server running a 2016-era database is now carrying two expired platforms. The full picture, updated as dates move, sits on the IT end-of-support calendar.