Why clinging to ageing PCs is such an expensive mistake

For decades, a periodic refresh of PC hardware fleets has been a routine part of IT managers’ responsibilities. Recently, however, tech leaders have been forced to grapple with the question of whether they can afford to continue these update cycles on their previous cadence.

IT budgets are always closely scrutinised, but over the past two years, demand from AI data centres has driven a global memory shortage, which has pushed business laptop prices sharply upwards. Simultaneously, the same trend is fuelling greater pressure to divert IT budgets towards AI enablement projects, leaving less cash for hardware refreshes.

“We are currently seeing weak PC demand across European businesses,” said Marie-Christine Pygott, Senior Analyst at research firm CONTEXT, “especially in smaller companies where budgets are tight. This results from the refresh wave we saw last year and which has been declining, and also significantly from increased PC prices.” 

Pygott says the European market saw the average selling price of business notebooks rise by more than 25% year-on-year in Q3, with a 4% rise just compared to the previous quarter. As such, an increasing number of organisations are choosing to wait and squeeze another 12 months out of their existing machines.

According to analyst firm Omdia, UK commercial PC shipments fell by 22% year on year in Q3, and a Gartner forecast from February 2026 says that the average lifespan of a business PC is set to lengthen by around 15% by the end of the year as buyers delay upgrades.

“Increased prices mean smaller organisations revert to upgrading products only when they break, rather than implementing full fleet refreshes,” Pygott said. 

Sweating assets can look like the prudent choice in the short term. For many organisations, though, delaying a refresh without a plan turns out to be a false economy – and there are arguably better ways to make a hardware budget go further.

The price of patience

In most cases, these postponements are based on the assumption – or the hope – that prices will settle, making a refresh more affordable a year later. Analysts, however, are not optimistic. 

“With prices still going up steadily, this situation is risky,” said CONTEXT’s Pygott. “While the rate of increase will slow in the next few quarters, prices will still rise, so delaying a purchase now means paying more later.” 

Omdia Principal Analyst Kieren Jessop says that component prices are a good indicator that these trends will continue: “Memory and storage now make up nearly 40% of a PC’s component cost, up from around 15%. Prices will keep rising until 2028, and even then they will not return to where they were.”

There’s a scheduling risk, too. Devices bought in the same refresh tend to age at the same rate, so a fleet that has been stretched for a year or two can need replacing all at once. That leaves IT teams facing one large, unplanned purchase at the very prices they were trying to avoid, along with the job of deploying hundreds of devices in a hurry.

The risks aren’t just about misjudging savings. According to a 2019 Techaisle study of Canadian SMBs commissioned by Microsoft and Intel, PCs more than four years old are 4.6 times more likely to need repairs, resulting in an average of 196 hours of lost productivity per device.  

In addition to repair bills, older devices often carry a hidden drag on productivity. The battery is often one of the first components to degrade on an ageing laptop, which means employees have reduced mobility, are forced to carry a charger, and may end up unexpectedly losing power mid-flow. More time spent submitting helpdesk tickets and waiting on repairs means more time away from tasks that add business value. 

It’s worth noting too that a company is often judged on its employee devices. The modern equivalent of the company car, the quality of a worker’s laptop says a lot about how much a business values its staff, both to the workforce itself and to prospective clients and partners.

Mind the gap

Jessop suggests that stretching a refresh cycle is “a sound way to ride out a price spike” for a three- to five-year-old fleet that’s still patched. While repair bills and lost productivity may be pricey, they are at least visible. However, the security risks that older hardware is more prone to tend to stay out of sight until something goes wrong. 

“The real exposure is the unsupported tail,” he said, “where Windows 10 security cover costs more each year.”

Many PCs built before 2018 can’t be updated to Windows 11, which requires a TPM 2.0 chip and a relatively recent processor. Windows 10 reached end of support in October 2025, so businesses that can’t upgrade face a choice between leaving unpatched devices on their network or paying Microsoft for Extended Security Updates (ESU). 

For commercial customers, the first year of ESU (which ended in October 2026) cost $61 per device, which doubles annually up to the three-year cap, topping out at $244 for the third year. In addition, organisations that sign up late have to pay for the years they missed, meaning the full three-year programme costs $427 per device.

Ironically, while many IT budgets are squeezed in order to support AI projects, delaying hardware upgrades can hinder an organisation’s ability to effectively utilise AI tools. 

Over the past two years, PC manufacturers have developed a variety of “AI PCs”, which use dedicated chips to power AI tasks. These neural processing units (NPUs) are designed to run AI workloads efficiently on the device itself without draining the battery or monopolising the processor.

These devices are becoming increasingly common, with Gartner estimating that NPU-equipped models made up around 36% of laptop shipments in 2025 (although it has since warned that rising prices will slow adoption through 2027). The firm expects 40% of software vendors to prioritise on-device AI features by the end of 2026, up from just 2% in 2024.

Older devices without these chips may struggle to effectively support on-device AI tasks, driving staff to unapproved cloud-based services instead. Data processed on the device, by contrast, stays under the organisation’s control, and there are no per-use fees of the kind external AI providers typically charge.

“Cloud costs rise with adoption,” said Jessop. “Each new user adds roughly the price of a chat subscription every month, so a successful rollout means a bigger bill. On a machine with enough memory, that same user costs nothing extra.”

As he points out, though, the NPU isn’t the only factor governing how capable a laptop is for AI tasks. Memory is just as important, and on thin-and-light laptops it’s usually impossible to upgrade once it leaves the factory. Current memory prices may make it tempting to cut back on RAM to save cash, but this may be a false economy: if new devices aren’t specced with enough memory from the outset, the business may be forced to make up for it with higher token bills until its next refresh, Jessop says.

Stretching the pennies

For many organisations, budgetary strictures are still a real challenge, but there are several ways to keep a device fleet up to date without embarking on a wholesale refresh. One option is what Jessop calls “refurbish and redeploy”, where an existing fleet is given a new lease of life by replacing components such as batteries and storage. These updates can be staggered, starting with staff whose roles demand the most from their devices, or whose machines are closest to falling out of support.

Alternatively, leasing and device-as-a-service (DaaS) models can flip the cost of a refresh from CapEx to OpEx, spreading it over the lifespan of the hardware. These contracts can also allow customers to build the next refresh into the agreement, which avoids an entire fleet reaching end of life at the same time. 

Even when a device has reached the end of its useful life, it’s not without value. IT asset disposition (ITAD) providers collect end-of-life business hardware, securely wiping it before refurbishing or reselling it. This helps recover some of its value, which can be put towards replacements, while giving hardware a second life avoids the emissions that come from making new devices. 

“Extending life, refurbishing and redeploying all cut emissions as well as cost,” Jessop said. “A machine that can no longer be patched should go back into the reuse chain through a certified refurbisher, not into landfill.”

Price volatility may be the new standard, and the routine PC refresh may become a thing of the past. However, even if hardware updates never return to the cadence IT managers were previously used to, postponing fleet upgrades indefinitely with no plan in place leaves device capabilities to slowly degrade.

Whatever the pace of upgrades, auditing the age, condition and support status of every device gives businesses the information they need to decide what to keep, what to refurbish and what to replace – and to budget for it before the decision gets made for them.