Windows Server 2016 Support Ends January 2027. Your Options.

Microsoft stops supporting Windows Server 2016 on 12 January 2027. If you are running one, that is roughly four months away, and it lands in the middle of the budget year most businesses are planning right now.

The deadline is easy to ignore because nothing visible happens on the day. That is exactly why it gets ignored, and why the businesses that handle it well start the conversation in autumn rather than in January.

What Actually Changes on January 12

Your server does not stop working. It does not shut down, lock you out or display a warning. Monday morning looks exactly like the Friday before.

What stops is security updates. From that date Microsoft no longer issues patches for newly discovered vulnerabilities in that operating system. Every flaw found after January is permanent on your machine, publicly documented, and never fixed.

That is the entire change, and it is worse than it sounds. An unsupported server does not get gradually less safe. It gets less safe every single month, in one direction, for as long as you keep it.

Why "It Still Works Fine" Is the Wrong Test

The most common reason businesses keep an old server is that it is running perfectly well. That is usually true, and it is not the question.

Three things tend to force the issue before the hardware does.

Your cyber insurance application. Insurers ask whether you are running supported, patched software. That question is an attestation on a legal document, and answering it inaccurately is a coverage problem rather than a paperwork problem. This is the one that catches people, because the renewal date and the end of support date rarely line up.

Your clients' security questionnaires. If you work in a supply chain, or serve regulated industries, somebody is going to ask. Increasingly they ask in writing and want a date.

Your software vendors. Line of business applications drop support for old server operating systems on their own schedules. The version you are on today may be the last one that runs where it is running.

Your Four Options

There are fewer real choices here than it feels like from the inside.

1. Replace the server. New hardware, current operating system, same shape as what you have. This is the simplest option to understand and the easiest to budget, because it is one capital line. It makes sense when you have applications that genuinely need to be on a machine in your building, or when your internet connection is not good enough to put everything in the cloud. The trade is that you are buying another five to seven year cycle and you will be having this conversation again.

2. Move the workload to the cloud. Depending on what the server actually does, much of it may not need to be a server at all. File shares, identity, email and application hosting all have cloud equivalents that remove the box entirely. This usually costs less up front and more per month, which changes it from a capital decision to an operating one. That is a genuine benefit for some businesses and a genuine drawback for others, and it depends on how your accountant prefers to see it.

3. Retire the workload. Sometimes the server exists because of one application, and that application now has a hosted version the vendor would rather you used. Worth asking before you price hardware for it. The cheapest server is the one you no longer need.

4. Pay to extend, if it is offered. Microsoft has historically offered paid extended security updates for server products past end of support. If that is available for 2016, treat it as a bridge you rent annually while you do one of the first three, not as a plan. Confirm what is actually on offer before you put a number in a budget, because the terms and the pricing change.

Doing nothing is technically a fifth option. It is a decision to run an unsupported, permanently unpatched system on your network, and it should be made deliberately and written down rather than arrived at by silence. Unsupported systems are one of the threats in our guide to the top cybersecurity threats facing businesses.

How Long This Really Takes

This is the part that surprises people, and it is the reason to start now rather than in December.

Across the server and cloud migration projects we have delivered, the elapsed time runs in months rather than weeks, and very little of it is engineering work. The engineering is a matter of days. What fills the calendar is waiting.

Waiting on a decision about which option you want. Waiting for leadership to approve the spend at the next meeting that happens to be scheduled. Waiting on a software vendor to confirm what their application supports and to book an engineer. Waiting on hardware to arrive. Waiting on a weekend everybody can agree to.

Almost every one of those waits belongs to somebody other than your IT provider. That is why a project started in October finishes comfortably and the same project started in December does not.

What Waiting Actually Costs

The cost of delaying an upgrade like this is rarely the thing people expect. It is not usually a breach.

It is that you lose your options. A business that starts in September chooses between replacing, migrating and retiring, on its own timeline, with time to get the application vendors on the phone. A business that starts in February is buying whatever is available, on whatever weekend it can get, and paying a premium to compress a schedule that did not need compressing.

The second cost is that everything on that server is now sitting on an unpatched platform, which matters most for the thing you would least like to lose. Worth checking that your backups of that server are being tested rather than simply running, because an old server is exactly where an untested backup tends to be discovered.

This Is Not the Only Date on the Calendar

Windows 10 already reached end of support in October 2025, and the paid extended updates programme for it runs out in October 2027. If you have machines still on it, that is a second deadline inside the same budget year.

Handled separately, those become two disruptive projects and two separate arguments about money. Handled together, they become one technology plan with dates and numbers that you can take to a board or a partner meeting and defend. That is a meaningfully easier conversation, and it is the reason to look at both now while there is still room to sequence them.

What to Do This Month

Three things, and the first two cost nothing.

• Find out what you actually have. How many servers, what operating system each one runs, and what each one is for. A surprising number of businesses cannot answer the third question, and that answer is what decides between replacing and retiring.
• Ask your software vendors what they support. Do this before you price anything. Their answer can eliminate two of your four options in a single phone call.
• Get it into next year's budget with a real number. Not a placeholder. The people who struggle in January are the ones who knew about it in September and never put a figure against it.

If you are not sure what you are running or which option fits, that is a short conversation rather than a project. Book a free 15 minute call and we will tell you what we would look at first and roughly what the decision hinges on, whether or not you end up doing the work with us.

The Short Version

Frequently Asked Questions