Advisers say memory costs and Windows deadlines are breaking PC refresh into role-based cycles

The advisers ITPro consulted could not agree on whether the conventional cycle is dead. All of them said it is shortening and lengthening at the same time, depending on the technology and the user ^[3]. Michael Rosholt, an SVP at the IT investment bank martinwolf, said: "Rather than one universal refresh cycle, we're seeing the market increasingly segmented by technology and use case." ^[1] The three named voices are an investment banker, the head of a managed IT services firm and a business development director at Egiss ^[1][2][11].

The one large fleet in the record moved to a longer calendar. Ole Bülow of Egiss supported a 120,000-strong global enterprise in its switch from a three-year device refresh cycle to a four-year one ^[11]. The client kept a fixed schedule. Assuming one device per employee and even replacement, its annual purchases fall from about 40,000 machines to about 30,000, roughly 10,000 fewer a year ^[15].

Bülow's own advice goes further than his client did. "A hybrid lifecycle, supported by signals such as DEX, is recommended," he said ^[12]. He described how it works: "A supported device used for predictable tasks may remain productive longer. Employees with increasing demands may require earlier replacement. The original device can then be redeployed to another user after secure recovery and preparation." ^[13] The trade-off is that money saved on light users is partly spent on monitoring device experience, wiping machines and reissuing them. Whether that comes out cheaper than a flat calendar is not yet known; none of the sources put a cost on it.

The timing pressure comes from outside the IT budget. Adam Bowles of Act360 said Microsoft's Windows 11 hardware requirements were "forcing replacements on OS support grounds, not performance" ^[4]. The end of Windows 10 support then set off a rush of purchases ^[5]. Memory prices have surged with the AI infrastructure buildout, with what ITPro described as seemingly permanent effects on device costs ^[6]. IDC analysis cited by ITPro projects a 20% year-on-year fall in PC sales, with low-budget devices the main casualty ^[7].

That collides with the cheapest part of a role-based plan. Bowles said employees with lighter performance needs are keeping hardware longer to save budget, while power users may get a new device every couple of years ^[2]. Extending the light-user tier defers spending this quarter. The machines those users would eventually receive sit in the low-budget category IDC says is shrinking, and PC makers are struggling to keep full product portfolios ^[7][8]. If that persists, next year's deferred order lands in a smaller catalogue at higher memory prices ^[6][8].

AI PCs sharpen the split at the top of the fleet. The stronger case is for newer machines that let power users run tools such as AI agents without a constant cloud connection ^[14]. Bülow framed the test as a question: "Which capabilities will employees actually use during the device's lifecycle?" ^[9] He said the answer "requires understanding applications and processing locations, rather than using an AI label as a universal refresh trigger" ^[10]. In my view the role split is sound for that power-user tier. For the rest of the fleet, the evidence so far supports a longer calendar with documented exceptions ^[11][12].