In my first week as a new CISO, I asked for a list of every system in the environment that no longer received vendor patches. I was told I’d have it the next day. It took eleven weeks. We found it in pieces: a spreadsheet from the infrastructure team, a half-finished CMDB export, a procurement folder of old maintenance contracts, and a facilities manager who mentioned that the badge system “runs on an old Windows box in the closet.” That last one wasn’t on any list, and it taught me more about end of support software risks than any audit report ever had.
I’ve told that story to a lot of peers over the years, and nobody is ever surprised. Every IT leader knows this gap, the distance between what leadership thinks is running and what’s actually running. That gap is where end of support software risks live, and in 2026 it’s getting more expensive to ignore than at any point I can remember.
This piece isn’t a vendor pitch or a checklist copied from a framework. It’s how I explain the problem to executives, how I rank it against everything else on my plate, and what has actually worked when I’ve had to fix it under real budget limits.
What End of Support Means for Software Risks
When a product reaches end of support, the vendor stops shipping security fixes, bug fixes, and technical help. The software doesn’t stop working. That’s what makes it dangerous. Nothing breaks on the day support ends. Servers still boot, applications still load, and the dashboards stay green.
Microsoft’s guidance on Windows 10 says it plainly: after October 14, 2025, the machines still work, but they no longer get technical support, software updates, or security fixes. The Australian Cyber Security Centre defines end of support the same way in its guidance for small and medium businesses, and its advice starts with a basic step most organizations still skip: find out when your devices and software will reach that date.
People sometimes split hairs between “end of life” and “end of support.” Vendors use the terms differently, and some keep taking bug reports after they’ve stopped building new features. For security purposes, I treat them as the same thing. If security patches aren’t coming, you’re exposed. The label on the vendor’s lifecycle page doesn’t change the math.
Why End of Support Software Risks Have Sharpened in 2026
I’ve managed unsupported systems my whole career, so I don’t scare easily on this topic. But three things changed over the last eighteen months, and together they shifted the risk from “chronic” to “acute.”
More Unsupported Software Means More End of Support Risks
The first is volume. Windows 10, the most widely deployed desktop operating system in corporate history, left support in October 2025. Lansweeper data reported by Help Net Security in July 2026 found Windows 10 still on 16.9% of Windows devices, and on 21.4% of devices at small and midsize businesses. The same analysis found that 18.7% of all Windows devices were already on an operating system past end of support, and another 22.2% would get there within six months. That’s roughly four in ten Windows machines either unsupported or about to be.
Attackers Are Moving Faster
The second is speed. Attackers aren’t waiting weeks to weaponize disclosed flaws anymore. A July 2026 cyber snapshot from the Michigan State Police cyber command center noted that vulnerabilities that once took days or weeks to exploit are now often targeted within hours of public disclosure. Acronis researchers pointed to a Gartner survey from August 2026 in which senior risk managers named AI assisted discovery of weaknesses as the top emerging risk. When finding a flaw gets cheaper, end of support software risks grow, because every system that can never be patched becomes a more attractive target.
Regulators Have Stepped In
The third is regulation. CISA’s Binding Operational Directive on end of support edge devices, issued February 5, 2026, moved this issue out of “IT hygiene” and into “mandated program” territory. It applies directly only to federal civilian agencies, but CISA, the FBI, and the UK’s National Cyber Security Centre released a joint fact sheet urging every organization to follow the same guidance. When three national agencies sign the same warning, your auditors and insurers notice.
The Seven End of Support Software Risks I Brief Every Leadership Team On
When I walk a board or executive committee through end of support software risks, I don’t lead with CVE counts. I lead with consequences. Here’s how I break it down.
1. The Vulnerability Count Only Goes One Direction
On a supported system, vulnerabilities come and go. A flaw gets disclosed, a patch ships, you deploy it, and your exposure drops. On an unsupported system, the count only climbs. Every new disclosure becomes a permanent resident, which makes this the most basic of all end of support software risks.
The Lansweeper figures make this concrete. Windows 10 devices averaged 1,903 known vulnerabilities each, almost three times the 652 found on Windows 11 machines. Two thirds of those Windows 10 flaws were rated high or critical. That gap will keep widening every Patch Tuesday for as long as those machines stay in service.
I tell executives to picture a building where the locks get weaker every month and the locksmith has stopped taking calls. You can add guards, but you’re paying more each month to protect something that’s still declining.
2. Edge Devices Carry the Sharpest End of Support Software Risks
If I could fix only one category of unsupported technology, it would be anything that faces the internet: firewalls, VPN concentrators, routers, load balancers, remote access gateways. These devices sit at the boundary, they’re reachable by anyone, and attackers scan them constantly.
The data backs this up. VulnCheck’s 2026 State of Exploitation research found that 42.5% of vulnerabilities exploited in 2025 affected devices that were end of life or likely end of life. Botnets lean on these devices even harder: 65% of vulnerabilities exploited by botnets hit end of life or likely end of life products.
What worries me more is a finding that got less attention. VulnCheck reported that only about a quarter of the exploited edge device flaws it identified showed up in CISA’s Known Exploited Vulnerabilities catalog. If your patching program prioritizes based on KEV alone, and many do, you’re missing most of the active exploitation against your oldest perimeter gear.
That’s why the federal directive is structured the way it is. As Help Net Security summarized it, agencies had three months to inventory affected edge devices, twelve months to remove devices already past end of support, eighteen months to remove every identified unsupported edge device, and two years to build continuous discovery so new ones get caught before they become a problem. Private companies aren’t bound by those deadlines, but I think the sequence is the right one for anybody.
3. Compliance Gaps You Can’t Close With Paperwork
There’s a quiet trap in most security frameworks. They require timely patching. If the vendor no longer makes patches, you can’t meet the requirement as written, no matter how good your process is.
PCI DSS is the clearest example. Requirement 12.3.4, mandatory since March 31, 2025, requires a review of hardware and software technologies at least once every twelve months. That review has to confirm vendors are still shipping security fixes, track announced end of life plans, and produce a remediation plan approved by senior management. Meanwhile, Requirement 6.3.3 expects critical patches within 30 days of release. As HeroDevs points out in its analysis of PCI DSS, HIPAA, and SOC 2, an end of life component turns that patching gap into something documented and auditable.
In plain terms: your assessor will find it, and “we’re working on it” without a signed plan won’t hold up. SOC 2 auditors take a similar view. They treat running unsupported software as a risk decision that someone accountable has to own on paper.
4. Cyber Insurance Is Getting Specific
Five years ago, insurance questionnaires asked whether you had antivirus and backups. Today the questions are sharper, and unsupported operating systems are a common line item. The Lansweeper analysis noted that insurers are scrutinizing unsupported systems more closely, and organizations can face higher premiums, coverage exclusions, or denied claims when an unsupported device contributes to an incident.
That last part matters most. A higher premium is a budget conversation. A denied claim after a ransomware event is an existential one. I now make sure the person who fills out our insurance application talks to my team first, because an inaccurate answer about end of support software risks is worse than an honest one with a remediation timeline.
5. Hidden Components Inside “Supported” Systems
Not every end of support problem shows up as an old operating system. Some of the worst ones hide inside systems that look current.
Look at Microsoft’s list of products leaving support in 2027. Alongside Windows Server 2016, it includes items like .NET Framework 4.6.2 and IIS 10 on Windows Server 2016. Those are runtime and platform components that business applications quietly depend on. An application team may honestly believe their system is fine because the application vendor still supports it, while the framework underneath is about to go dark. Moving off that framework often forces a bigger decision about whether to rewrite or refactor legacy code.
Open source is the same story at a bigger scale. Qualys research cited by SC World found that 20% of federal agency assets contained high risk end of support software. When I run software composition analysis on internal applications as part of an application security assessment, I almost always find unsupported libraries that nobody chose on purpose. They came along with something else.
6. You Lose Your Safety Net During an Incident
People forget that end of support also means end of help. When something goes wrong on an unsupported platform, you can’t open a priority ticket with the vendor. You can’t get a hotfix. You’re often stuck with forum posts and whatever your most senior engineer remembers.
I’ve been in that position during an active incident. A legacy application server started behaving erratically, we suspected compromise, and the vendor’s answer was polite but final: the product was out of support and they couldn’t assist. We spent two days rebuilding something that, on a supported platform, would have been a call and a patch. Recovery time is a business metric, and unsupported systems make it worse.
There’s also the problem of fragile dependencies. Unsupported systems tend to break when the things around them change, like a new TLS requirement, an updated identity provider, or a cloud service retiring an old API. Each of those becomes an outage risk you didn’t budget for.
7. Extended Support Can Mask End of Support Software Risks
I want to be fair here, because extended security update programs are useful. Microsoft offered a paid Extended Security Updates program for Windows 10 commercial customers for up to three years, and consumers could get one more year through October 13, 2026. Microsoft also said it would keep shipping security updates for Microsoft 365 apps on Windows 10 until October 10, 2028. For Windows Server and SQL Server, ESU is available on premises through Azure Arc or in Azure itself.
Those programs buy time. They don’t buy safety forever, and they don’t fix the root cause. I’ve watched organizations treat ESU as the plan instead of the bridge, and then find themselves three years later with the same systems, a larger bill, and no money left for migration. Lansweeper’s Laura Libeer described consumer ESU as a temporary solution, and I’d apply that same thinking to every extended support contract.
Compensating controls deserve the same skepticism. Network segmentation, application allowlisting, and strict access restrictions are the core of securing legacy applications, and they really do reduce exposure. But they also need upkeep, testing, and people who understand why they exist. When those people leave, the controls drift. A segmented legacy network with an undocumented firewall exception is still an open door.
Upcoming Software Deadlines That Raise End of Support Risks
As I write this in late November 2026, a few dates deserve attention right now:
- Windows Server 2016 loses extended support on January 12, 2027. Microsoft has said the January 2027 security update will be the last one, and the cutoff covers Datacenter, Essentials, MultiPoint Premium, and Standard editions. That’s about seven weeks away.
- SQL Server 2016 left extended support on July 14, 2026, and moved into the paid ESU period.
- Windows Server 2012 ESU coverage ended on October 13, 2026, so anything still on that platform is now fully unpatched.
- Windows 10 consumer ESU ended October 13, 2026, while commercial ESU can run longer for organizations that paid for it.
Each of these dates changes the end of support software risks for every system still running that product. Microsoft’s lifecycle pages, the Center for Internet Security’s End of Support Software Report List, and community trackers like endoflife.date are the sources I rely on. I’d suggest building those dates into your planning calendar at least eighteen months ahead. Twelve months isn’t enough time to budget, test, and migrate legacy Windows applications that have real business dependencies.
A Practical Program for Managing End of Support Software Risks
Every organization I’ve worked with has had some unsupported technology. Zero is not a realistic goal. The goal is that every unsupported system is known, owned, justified, protected, and on a dated path out. Here’s the program I’ve used, refined over several companies.
Build an Inventory That Exposes End of Support Software Risks
Start with discovery, not with spreadsheets people already have. Use authenticated scanning, endpoint agents, network discovery, and cloud asset APIs, and then reconcile the results against your CMDB. Expect the reconciliation to be painful. The disagreements between sources are where the unknown systems live.
Cover firmware and appliances, not just servers and laptops. Operational technology belongs on the list if you have any. So do the “closet systems” that facilities, marketing, or a regional office set up years ago. And match every asset to its vendor lifecycle date, not just its version number.
Assign a Named Owner to Every Unsupported System
This sounds bureaucratic, but it’s the single change that moves the needle most. Every unsupported system needs a business owner, a person with a name and a title, who knows the risk is theirs. When there’s no owner, there’s no decision, and the system just sits there.
Rank by Exposure, Not by Age
Not all end of support software risks are equal. As one expert quoted by Infosecurity Magazine put it, an unsupported router at the network edge is much more likely to be exploited than a patient monitor on a segmented clinical network. I rank based on three questions. Is it reachable from the internet? Does it touch sensitive data or critical operations? Could an attacker use it to move toward something more valuable?
Anything that answers yes to the first question goes to the top of the list, regardless of how small or boring it seems.
Pick One of Five Paths for Each System
Every unsupported system should land in one of five categories, with a target date:
- Upgrade to a supported version of the same product.
- Replace with a different product, a managed service, or, for a small firm with a truly unique workflow, custom software built for your small business.
- Retire it completely, which happens more often than people expect once you ask whether anyone still uses it.
- Isolate it behind strong compensating controls while a longer migration runs.
- Extend support through a paid ESU or third party support contract, as a bridge with an end date.
The fourth and fifth options must have expiration dates. An exception without an end date isn’t an exception. It’s a policy. For in-house business applications, the upgrade and replace paths usually turn into a legacy application modernization project, so plan them that way.
Document Risk Acceptance Properly
When a business leader decides to keep running something unsupported, that decision needs to be in writing. It should say what the system is, why it can’t move yet, what controls protect it, what the residual risk is, and when the decision will be reviewed. It should be signed by someone with the authority to accept that risk. This is exactly what PCI assessors and SOC 2 auditors want to see, and it’s also what protects your security team when something goes wrong.
Fix the Funding Model
Most end of support problems are budget problems in disguise. Systems get bought with capital funds and nobody plans the refresh. A recruiter quoted in coverage of the federal directive made the same observation about agencies: when the manufacturer stops supporting hardware, there’s no money set aside for the upgrade.
I push for lifecycle funding to be part of the original purchase decision, whether you build or buy the software. If a firewall has a seven year support window, the replacement cost belongs in year six of the budget, not in an emergency request in year eight.
Put Lifecycle Terms Into Procurement
Before signing any new contract, I want to know the vendor’s published support window, how much notice they give before ending support, and whether they’ll commit to security fixes for a minimum period. Those questions cost nothing to ask during procurement and are almost impossible to negotiate later. The same applies when choosing a software development partner for custom work: get it in writing who will patch the code after launch, and for how long.
Make Discovery Continuous
The final step in the federal directive is the one that matters most long term: continuous discovery that catches systems approaching end of support before they get there. Quarterly reviews aren’t enough. Lifecycle dates should be a live field in your asset data, with alerts at eighteen months, twelve months, and six months out.
How I Present End of Support Software Risks to the Board
Boards don’t need a list of CVEs. They need to understand three things: how much of the environment is exposed, what it would cost if that exposure were exploited, and what it would cost to fix.
I present end of support software risks as a single chart showing unsupported assets over time, split by exposure level, with planned remediation dates. Next to it, I show the regulatory and insurance consequences in plain language. Then I show the funding ask, backed by a realistic software development cost estimate for each major migration. Microsoft’s own business guidance makes the case in executive terms: dealing with end of support risks protects business continuity, cuts downtime, and protects revenue and reputation.
The conversation changes when directors see that this isn’t a technology preference. It’s a known, measurable, and increasingly regulated liability with a clear fix.
Final Thoughts on End of Support Software Risks
I don’t believe in fear as a management tool, and I don’t think unsupported technology is a sign that an IT team has failed. Every environment carries some of it. What separates mature organizations from vulnerable ones is whether they can see it, whether someone owns it, and whether there’s a real date when it goes away.
The stakes are higher in 2026 than they’ve ever been. Attackers are faster, the biggest operating system migration in a decade is still unfinished in many companies, Windows Server 2016 is weeks from losing support, and regulators and insurers are asking pointed questions. If you start your inventory this quarter, you’ll be ahead of most of your peers in managing end of support software risks. If you wait for the incident, you’ll be explaining to your board why the risk was known and nobody acted.
Frequently Asked Questions About End of Support Software Risks
What are end of support software risks?
They’re the security, compliance, operational, and financial exposures that come from running software or firmware after the vendor stops delivering security patches and technical help. The software keeps working, but every newly discovered flaw stays open permanently. The Australian Cyber Security Centre’s end of support guidance gives a clear overview for businesses of every size.
Is it ever acceptable to keep running unsupported software?
Sometimes, yes, as a temporary and documented decision. Specialized equipment, regulated systems, and complex business applications can take time to replace. The key is formal risk acceptance, strong compensating controls, and a dated exit plan. HeroDevs explains how PCI DSS, HIPAA, and SOC 2 view this choice.
Does Microsoft’s Extended Security Updates program make Windows 10 safe?
ESU delivers critical and important security fixes for a limited time, which lowers risk while you migrate. It doesn’t include new features or regular technical support, and it ends on a fixed date. Microsoft’s Windows 10 end of support page outlines the options.
Why are unsupported edge devices considered so dangerous?
They face the internet, attackers scan them constantly, and they often contain well documented flaws that will never be fixed. Research from VulnCheck on network edge exploitation found that 42.5% of vulnerabilities exploited in 2025 hit end of life or likely end of life devices.
When does Windows Server 2016 reach end of support?
Extended support ends on January 12, 2027, and the January 2027 update will be the last security release for every edition. BleepingComputer covered Microsoft’s announcement, and Microsoft’s 2027 end of support list shows related products leaving support at the same time.
Do private companies need to follow CISA’s directive on end of support edge devices?
The directive only binds federal civilian agencies, but CISA, the FBI, and the UK NCSC have urged every organization to adopt the same practices. Their joint fact sheet on reducing the attack surface for end of support edge devices is a practical place to start.
How often should we review end of support software risks?
At minimum once a year, which matches PCI DSS Requirement 12.3.4. In practice, lifecycle dates should be tracked continuously in your asset inventory, with alerts well ahead of each vendor deadline. The Center for Internet Security’s End of Support Software Report List is useful for tracking upcoming dates.
References
- Windows 10 Support Has Ended on October 14, 2025, Microsoft Support
- Ending Support in 2027, Microsoft Learn
- Extended Security Updates for Windows Server and SQL Server, Microsoft
- Is End of Support Leaving Your System Vulnerable?, Microsoft Windows for Business
- Microsoft: Windows Server 2016 Reaches End of Support in January, BleepingComputer
- The Windows 10 Hangover Is Becoming a Security Problem, Help Net Security
- CISA Orders US Federal Agencies to Replace Unsupported Edge Devices, Help Net Security
- Reducing the Attack Surface for End of Support Edge Devices, CISA, FBI, and UK NCSC
- 2026 State of Exploitation: The Risk of End of Life Infrastructure, VulnCheck
- Network Edge Devices Still Widely Used After Reaching End of Life Status, Cybersecurity Dive
- Every PC Has a Junk Drawer, and Attackers Know It, Acronis Threat Research Unit
- 5 Ways Federal Agencies Can Reduce Cyber Risks From Tech Debt, SC World
- Does Running EOL Software Violate PCI DSS, HIPAA, or SOC 2?, HeroDevs
- Why Addressing Legacy IT Is an Urgent Strategic Priority for CISOs, Infosecurity Magazine
- End of Support Guidance, Australian Cyber Security Centre
- End of Support Software Report List, Center for Internet Security
- Cyber Snapshot: End of Life, Michigan State Police Michigan Cyber Command Center