Blog
The Copilot+ question: planning a hardware refresh when prices are rising
Memory prices are climbing, lead times are stretching, and the Windows 10 exit wave is still working through most estates. Into that market walks the Copilot+ PC, carrying a premium and a promise. Here's how to decide what to buy, when, and how to make the answer a fleet decision rather than a gamble.
Every hardware refresh involves a bet on the future. This one involves two: whether on-device AI will matter to your users within the life of the fleet, and whether the components inside every quote will cost more next quarter than they do today. Both bets have better and worse ways of being placed.
What Copilot+ actually specifies
Strip away the branding and Copilot+ is a hardware certification: a neural processing unit rated at 40 TOPS or more, at least 16GB of RAM and 256GB of storage. Devices that meet it unlock a set of on-device AI features in Windows 11, things like live translation, improved semantic search and studio camera effects, that run locally on the NPU rather than in the cloud. What it is not is a prerequisite for Copilot itself. The Microsoft 365 Copilot your users see in Word, Outlook and Teams is a cloud service, and it runs identically on any capable Windows 11 device.
The price problem is real
The component market has turned sharply against buyers. AI datacentre demand is absorbing memory and flash production, and DRAM contract prices roughly doubled in the first quarter of 2026, with further rises following through the year. PC manufacturers have warned that finished device prices could rise by 15 to 20 per cent, and industry analysts expect the supply imbalance to persist into 2028. Lead times on popular enterprise models have stretched accordingly. The practical consequence: a quote from last autumn is fiction, and a quote today has a shorter shelf life than the approval cycle that follows it.
The Windows 10 wave hasn't finished breaking
Windows 10 left mainstream support in October 2025, but plenty of estates are still riding Extended Security Updates while their refresh programmes catch up. That bridge is finite and priced to encourage crossing it. So refresh volume that would normally spread over several years is compressed into a market with rising prices and constrained supply. Waiting for prices to fall is a strategy that analysts' forecasts do not currently support.
So is the premium worth paying?
An honest answer has two halves. The memory half of the Copilot+ specification is worth meeting regardless of your view on AI: 16GB should now be the floor for any enterprise device, because in this market, buying thin and upgrading later is the most expensive path available. The NPU half is more genuinely a judgement call. If your users' AI experience is Microsoft 365 Copilot, the cloud does the work and standard hardware serves them well. If you keep devices for four or five years, a modest NPU premium is cheap insurance against the features that will arrive during that lifecycle. Buy for the life of the device, not for the launch keynote.
Buy for the life of the device, not for the launch keynote.
One window, not a drip
In a stable market, drip-feeding replacements out of the support queue is a defensible habit. In this one it means repricing your refresh every quarter, in the wrong direction, while your estate stays inconsistent for years. A consolidated refresh window lets you lock pricing and allocation with your supplier while today's quote still stands, land a single hardware standard, and give the programme an end date. The trade is a bigger procurement decision up front, which is exactly why that decision needs data behind it.
Eligibility data before purchase orders
The costliest refresh mistake is buying to assumptions: replacing devices that were eligible for Windows 11 all along, or sizing the order from an asset register that stopped being true two years ago. Before anyone signs a purchase order, you want a per-device picture of what you actually have: which machines meet requirements, which fail and why, and which are close enough that the answer is a firmware setting rather than a pallet of new laptops. That evidence-based picture of the estate is precisely what our Meridian platform is built to produce, and it routinely changes the size of the order.
Make the landing zero-touch
However the procurement question resolves, the delivery question has a settled answer. New hardware should ship straight from the supplier to the user and configure itself out of the box through Windows Autopilot: no imaging infrastructure, no engineer touching each device, no depot stage adding weeks to constrained lead times. If your refresh plan still routes every box through a build bench, fixing that before the hardware arrives will save more than any negotiation on unit price. A rising market punishes waste, and hand-built provisioning is the most avoidable waste a refresh programme carries.
The Copilot+ question, then, is really three smaller ones: what does the data say you need, when should you buy it, and how does it land? Answer those in order and the premium stops being a gamble.
Planning a refresh against this market?
Bring us the estate. We'll put evidence behind the eligibility picture, the order size and the provisioning plan before you commit to hardware.