BreachnodeopsEnterprise Underwriting Intelligence
Security Controls

Still Running Unsupported Servers? A 9-Step Plan to Keep Legacy Systems Insurable

Can't retire that Windows Server 2012 box, the plant-floor HMIs or the old ERP build before renewal? This nine-step plan covers the isolation, controls and written evidence that get an underwriter to quote.

By the Breachnodeops editorial team · November 2026 · 10 min read

How Underwriters Treat End-of-Life Software at Renewal

Your renewal application asks a blunt question: do you run any operating systems or applications the vendor no longer supports? For many companies the honest answer is yes. A Windows Server 2012 R2 host runs a line-of-business app nobody has rewritten. A few Windows 7 or Windows 10 workstations drive machines on the plant floor. The ERP sits on a database version that has left mainstream support. You can't retire any of them before the policy renews, so the real question is what you put in front of the underwriter so they quote instead of walking away.

With end-of-life software, cyber insurance carriers usually land on one of four outcomes:

  • Quote as normal: the systems are few, isolated and well documented.
  • Quote with conditions: a higher retention, a sublimit for incidents involving unsupported assets, or coinsurance on ransomware losses.
  • Quote with an exclusion: losses that start from or spread through unsupported software aren't covered. Some forms tie the coverage cut to how long a known vulnerability has gone unpatched.
  • Decline: usually because the systems face the internet, hold sensitive data, or the applicant can't say how many there are.

Underwriters care less that legacy systems exist than how exposed they are. Is the box reachable from the internet? From the general user network? Does it store regulated data, or can it stop production? An unexplained "yes" pushes you toward the bottom two outcomes. Answering "no" when the truth is yes is far worse: a misrepresentation on the application can let the insurer deny a claim or rescind the policy. The nine steps below turn that "yes" into an exposure the underwriter can price.

Advertisement

Steps 1 to 3: Inventory, Rank and Decide the Fate of Each System

Step 1: Build an inventory you can defend

Underwriters lose confidence fastest when the applicant says "a handful" and later finds forty. Pull data from every source you have: your EDR console, your vulnerability scanner (Tenable, Qualys and Rapid7 can all flag end-of-life versions), your CMDB, cloud asset inventories and, for operational technology, a passive discovery tool such as Claroty, Nozomi Networks or Dragos. Then walk the plant floor and the server rooms, because air-gapped and forgotten machines rarely show up in any console.

For each asset, record the hostname, OS and version, the vendor's end-of-support date, the business owner, its function, the data it holds, its network zone, whether it's internet-facing, the controls already in place, and a target retirement date.

Step 2: Rank by exposure, not by age

A 15-year-old HMI on an isolated cell network is a smaller problem than a web server in the DMZ that's three years out of support. Use three tiers:

  • Tier 1 (critical): internet-facing, holds personal or payment data, or has domain admin credentials cached on it.
  • Tier 2 (elevated): internal, but reachable from the user network or used by many staff.
  • Tier 3 (contained): isolated, single-purpose, with no sensitive data and no inbound access from user networks.

Step 3: Assign each system one path

Every asset gets one of four decisions: retire before renewal, buy extended support, isolate and compensate, or formally accept the risk. Fix Tier 1 systems before the renewal date even if the fix is ugly, such as pulling them off the internet and putting them behind a reverse proxy or VPN. A Tier 1 system left in place is among the most common reasons for a decline. Tier 3 systems can usually stay as they are if Steps 4 and 5 are documented.

Steps 4 to 6: Isolate, Compensate and Buy Time with Extended Support

Step 4: Segment so a compromise can't spread

For legacy assets, underwriters weigh network segmentation more heavily than almost any other control, because it decides whether one breached server becomes a company-wide ransomware event. The minimum credible setup is a dedicated VLAN or subnet behind a firewall with deny-by-default rules. Open only the ports and source hosts the application needs, block internet egress, and allow no direct RDP or SMB from the user LAN. Admin access should go through a jump host with MFA. For OT, follow a zone-and-conduit model based on IEC 62443 or the Purdue levels, with an industrial DMZ between IT and the plant. If re-addressing old systems is too risky, host-based microsegmentation tools such as Illumio or Akamai Guardicore can enforce policy without changing the network.

Step 5: Layer compensating controls

You can't patch the OS, so make it harder to exploit and faster to detect. Typical controls include:

  • An EDR agent where one still exists. Some vendors keep legacy agents for older Windows versions, so check before you assume there's none.
  • Application allowlisting, which works well on single-purpose machines.
  • Disabling SMBv1, unused services and local admin rights, and rotating service account passwords.
  • Privileged access management for the accounts that touch these hosts.
  • Logs sent to your SIEM or MDR provider, with alerts for logons from anywhere other than the expected jump host.
  • Immutable or offline backups with a restore that's been tested and documented.

Step 6: Buy extended support where it exists

Paid security updates can move an asset from "unsupported" to "supported" on many questionnaires, which is often the cheapest way to change an underwriting answer.

OptionTypical useIndicative 2026 cost (USD)
Microsoft Extended Security UpdatesWindows 10, Windows Server 2012/2012 R2Per-device pricing for Windows 10 commercial ESU is modest in year one and rises each year. Server ESU is priced per core and costs considerably more, and is usually cheaper through Azure Arc or when hosted in Azure. Check current terms.
Third-party micropatching (e.g. 0patch)Old Windows builds and some applicationsRoughly tens of dollars per agent per year, with volume discounts
Linux extended lifecycle support (e.g. TuxCare)CentOS 7, older Ubuntu and RHEL-derived buildsRoughly $50 to $200 per server per year
Third-party ERP support (e.g. Rimini Street, Spinnaker Support)Older SAP and Oracle releasesCommonly quoted at a sizable discount to the vendor's annual maintenance fee
OEM or integrator OT supportHMIs, SCADA servers, historiansHighly variable, often bundled into service contracts

These are typical 2026 market ranges. They vary by contract, region and volume, so confirm current pricing with the vendor or your reseller. ESU availability and pricing also differ between enterprise, education and government licensing programs. Some questionnaires don't count third-party patching as "vendor supported," so ask the carrier before you rely on it.

Steps 7 to 9: Build the Evidence Pack and Present It

Step 7: Put everything in writing

Controls you can't prove count for almost nothing in underwriting. Put together an evidence pack of roughly six to ten pages:

  1. The legacy asset register from Step 1, with tiers and decisions.
  2. A network diagram showing the isolated zones and permitted flows.
  3. A firewall rule export or summary for those zones.
  4. An EDR or monitoring coverage screenshot listing the legacy hosts.
  5. Extended-support contracts or order confirmations, with term dates.
  6. The latest backup restore test, with date, system and time to recover.
  7. A retirement plan with milestones, a budget line and a named owner.
  8. A risk acceptance memo signed by the CFO or COO for anything left in the accepted category.

Step 8: Submit through your broker with a one-page narrative

Most cyber insurance underwriting requirements arrive as yes/no checkboxes, and checkboxes can't explain context. Ask your broker to attach a one-page narrative covering how many systems there are, why they can't be retired yet, how they're isolated, and when they'll be gone. Send it with the application, not after a decline. Ask the broker what exclusion wording the carrier usually applies, and push for something narrow. A sublimit or higher retention that applies only to incidents starting on named legacy assets is much better than a broad unsupported-software exclusion, which could be read to apply to any breach that passes through an old machine.

Step 9: Commit to a timeline and report against it

Hypothetical example: a 300-person manufacturer has two Server 2012 R2 hosts and 14 legacy HMIs. It commits to ESU on both servers now, ERP migration in Q2, and HMI replacement over 18 months during planned shutdowns. It sends the broker a short progress update at mid-term. At the next renewal the underwriter sees a shrinking exposure instead of a static one, which is what earns better terms.

Line the timeline up with your regulatory obligations too. The UK's Cyber Essentials treats in-scope unsupported software as a failure unless it's removed or segregated from scope. The EU's NIS2 and DORA expect documented vulnerability management. Frameworks such as Australia's Essential Eight, Saudi Arabia's NCA controls and India's CERT-In directions all expect timely patching or a justified exception. Requirements differ by country and sector, so have counsel or a qualified compliance adviser confirm what applies to you.

Cost Drivers, Red Flags and the Renewal Decision Rule

Expect the work to cost more in staff time than in tooling. The biggest drivers are OT discovery licenses (often five figures for a multi-site plant), the firewall and switching work needed for real segmentation, per-core server ESU, and the internal hours to build and maintain the evidence pack. For a mid-size company with a few dozen legacy assets, total spend in the low tens of thousands of USD is plausible. A large multi-site manufacturer can spend far more. Compare that with what an exclusion would cost you in uninsured ransomware losses or a premium uplift. Local labor and licensing costs vary widely by region.

These red flags make an underwriter cautious fast:

  • Any unsupported system reachable from the internet, including through a vendor remote-access tool left permanently on.
  • Legacy hosts on the same flat network as user laptops.
  • Shared or domain admin credentials used to log into old machines.
  • No backup, or a backup that has never been restored.
  • A retirement plan with no date, no budget and no owner.
  • Inventory counts that don't match between your scanner, EDR console and application.

The decision rule for renewal: if a legacy system is internet-facing or holds sensitive data, fix that exposure before you submit, whatever it costs. If it's internal, isolate it, add the Step 5 controls and buy extended support where it's available. If it's contained and single-purpose, document it and accept the risk in writing. This week, export the end-of-life assets from your scanner and EDR, reconcile the two lists, and send the draft register to your broker so they can tell you which carriers will want the full evidence pack.

Quick answers

Will running unsupported operating systems automatically get my cyber insurance application declined?

Usually not. Most carriers will still quote if the systems are few, isolated from the internet and user networks, and backed by documented compensating controls. Declines tend to follow internet-facing legacy systems, sensitive data on unsupported hosts, or an inventory the applicant can't explain.

Should I just answer 'no' to the end-of-life software question if the systems are isolated?

No. If the system is unsupported, disclose it and explain the isolation in a supplemental narrative. An inaccurate answer can give the insurer grounds to deny a claim or rescind the policy, which is far worse than a sublimit or higher retention.

Does paying for Microsoft Extended Security Updates count as 'supported' for insurance purposes?

Many underwriters treat systems receiving vendor ESU as supported, because they still get security patches. Keep the order confirmation and term dates in your evidence pack and make sure the patches are actually installed. Confirm how your carrier's questionnaire defines 'supported', because wording varies.

What's the minimum network segmentation an underwriter will accept for legacy servers?

A credible baseline is a dedicated subnet behind a firewall with deny-by-default rules, only the required ports open, no internet egress, and admin access only through an MFA-protected jump host. A VLAN with no firewall rules between zones usually won't count. Have a diagram and a rule export ready to prove it.

Can I negotiate an unsupported-software exclusion down to something narrower?

Often, yes, especially with a strong evidence pack and a dated retirement plan. Ask your broker to push for a sublimit or higher retention that applies only to incidents starting on named legacy assets, instead of a broad exclusion. Have a licensed broker or coverage counsel review the final wording before you bind.

How do OT workstations and HMIs differ from IT servers in underwriting?

Underwriters generally accept that OT assets can't be patched or replaced quickly, so they focus on separation between IT and OT, controlled remote access for vendors, and recovery capability. An industrial DMZ, passive OT monitoring and a phased replacement plan tied to planned shutdowns are the evidence that tends to matter most.

Written by the Breachnodeops editorial team. Prices, features and rules change often - confirm details with the vendor or a licensed professional before you decide.