How finance can cut a hardware refresh budget and prioritize risk

The squeeze in two numbers

The Sage Sustainable Electronics benchmark released in January 2026 found 43 percent of organizations expecting flat hardware spend in 2026 and 22 percent expecting less. In the same quarter, Gartner forecast average PC prices 17 percent higher than in 2025, and IDC forecast the same 17 percent growth in average selling price with no meaningful memory relief before the end of 2027. At prices 17 percent higher, the same budget buys about 14.5 percent fewer devices. A calendar that says half the fleet is due does not care.

The IDC release in March 2026 put it in a sentence: the era of bargain-priced PCs is behind us for now. Gartner's third-quarter 2026 release recorded shipments down 22.4 percent year over year and noted that DRAM costs had surged five times year over year in the second quarter. Dell's chief operating officer said in late 2025 that about 500 million PCs were too old to run Windows 11 and roughly another 500 million could but had not upgraded. Demand that was pulled forward has been spent, and what remains is being bought at higher prices.

Where the budget cut comes from

Reviewing each device can cut refresh spending by replacing fewer machines, and it shows the team which replacements carry the risk. An age-based calendar cannot tell you. It marks every device that crossed a date as due, and finance has no basis to challenge any line, so the whole list gets funded or the whole list gets deferred. Both outcomes are wrong, and the second one is how the service desk ends up with the risk the budget avoided.

A verdict per device changes the shape of the request. Each device due by the calendar gets scored on age for its class, workload fit, measured performance, reliability and Windows 11 readiness, and the result is one of four outcomes with a reason. Keep defers the purchase, and that deferral is the saving. Right-size buys a cheaper class, and the difference is the saving. Redeploy moves a healthy over-specified device to a due seat, and that is upside once the move happens. Replace spends the money, and the report says why. The budget request becomes the replace and right-size cohorts by quarter, with the keep cohort as the line finance can see it did not fund.

Counting the saving so it survives audit

Three rules make the number defensible. First, the headline must equal the sum of its lines on the page: deferred purchases, avoided wrong-size purchases and the provisioning hours that no longer happen. Second, the redeploy pool is shown separately as upside and never added to the floor, because a redeploy only saves money when the device has moved and the purchase order is confirmed as not placed. Third, every external assumption carries its source and can be replaced with your own price or labor rate, and the ledger records the change.

The verdict model applies those rules, and the sample report shows the result on a fictional 7,400-device fleet: a calendar budget of $7,432,000 for 4,030 due devices, a floor of $2,574,284 made of three lines that add up, 34.6 percent of that budget, a redeploy pool of up to $1,996,700 beside it, and the Windows 10 exposure priced at Microsoft's cumulative Extended Security Update rate of $183 per device. Savings stay labeled identified until your team executes the plan and finance records the avoided purchase. The ledger moves each line from identified to executed to banked, and the quarterly readout shows all three columns.

Auditors ask two questions about a savings claim: what would have been spent, and how do you know the money stayed unspent. The first answer is the class price, from your own purchase orders or the class default with its source. The second is the purchase order that was never raised, recorded against the device and the verdict. A ledger that holds both for every line is a document an audit committee can read; a slide with a single headline number cannot be.

Where the risk actually sits

The risk in a refresh is in the devices that should go and do not. Replace verdicts carry it: devices past class life with crashes or unexpected restarts in the last 90 days, or unable to take Windows 11 with no redeploy target. They claim no savings in the model, and the report prices what it costs to leave them in service, from your own service desk ticket data for support and from Microsoft's published ESU price for Windows 10 exposure. Funding that cohort first is the risk control. Keep verdicts still need monitoring and review as device condition changes.

The second risk is the one finance creates by deferring the whole list. Microsoft's commercial ESU runs to 10 October 2028 at most, and a device that cannot take Windows 11 has a hard date whatever the budget says. A verdict per device turns that date into a quarter-by-quarter plan rather than a cliff.

The honesty rule, and what to ask your vendor

A fair test of any refresh advisor is whether they are paid the same when you keep a device as when you replace it. We sell no hardware, take no referral fee from anyone who does, and if the identified savings floor does not justify a subscription, the assessment readout says so and recommends you do not subscribe. Ask every vendor in the room the same question, and ask to see a sum check on their savings slide.

The first step is an export. The initial report takes an Intune or Configuration Manager export and returns the verdict split for your fleet, the savings floor with its sum check, the redeploy pool as upside, the calendar against the evidence, the cost of doing nothing and your Windows 11 readiness count, with a 30-minute readout for your endpoint lead and your budget owner. It is written for fleets of 1,000 devices and up, and it is the document to bring to the next budget review instead of the age report.

Common questions