Copilot+ PCs Are 40% of New Business Laptops: What It Means for Private Writing
At an October 2026 event, Microsoft’s Windows leadership said that more than 40 percent of business laptops now being built are Copilot+ PCs, machines with a dedicated neural processing unit, and that Windows is pairing them with “hybrid” routing that splits work between on-device models and the cloud. Microsoft also claimed more than two trillion local inferences a month across Windows, a figure reported without a stated method, so treat it as directional.
Strip away the branding and a concrete fact remains: a lot of new work laptops can now run small AI models on the machine itself. For anyone who writes confidential text, that is the part worth understanding.
What The NPU Actually Does
A neural processing unit is a chip designed to run AI models efficiently, at low power, without leaning on the main processor or a graphics card. On a Copilot+ PC it mostly handles small, always-on tasks: live captions, image effects, and small preinstalled models. The current hardware bar is around 40-plus TOPS, a throughput number. We explain the categories in NPU versus CPU versus GPU for local AI writing, and why the TOPS number stopped being the useful one in NPU TOPS scores stopped mattering.
The honest caveat, which analysts have made repeatedly, is that memory bandwidth, not the TOPS rating, usually decides how fast a larger local model runs. An NPU helps; it is not magic.
You Do Not Need A Copilot+ PC To Write Locally
Here is the part the marketing obscures: a small writing model does not need an NPU at all. A 1.7B or 4B model that rewrites text runs on an ordinary CPU, on hardware years older than any Copilot+ branding. The NPU makes some things faster and more power-efficient; it is not the thing that makes local writing possible. We spell this out in do you need a Copilot+ PC for local AI and does local AI need a graphics card.
So if you are tempted to buy a new machine to write privately, you probably already have the hardware. The model matters more than the chip.
Hybrid Routing Is Convenient, Not Private By Default
“Hybrid” routing, where the system decides whether a request runs locally or in the cloud, is the detail to watch. Convenience-wise it is sensible: small tasks stay on the device, hard ones go to a server. Privacy-wise it means you are not always in control of where a given piece of text goes, unless the tool makes the choice explicit. For a confidential draft, “the system decided to send this to the cloud” is exactly the outcome you were trying to avoid.
The guarantee that matters is not “usually local” but “local when I choose, and provably so.” A tool where local mode makes zero network calls during a rewrite is a different promise from a router that leans local. We cover the distinction in local AI versus cloud AI for confidential writing. Microsoft’s own Windows blog, at blogs.windows.com, is the primary place to read what Copilot+ routing does.
Before:
We should get everyone Copilot+ PCs so our confidential writing stays private on the device.
After:
Copilot+ PCs add an NPU, but our confidential writing is kept private by the model running locally, not by the hardware. Our current laptops can already run a small local model. The question to settle is whether a tool keeps a draft on the machine by our choice, not whether a router usually keeps it local.
The second version separates the hardware from the privacy guarantee, which are not the same thing.
A Wrivio Context for an IT hardware note could say:
Rewrite this as a neutral note for colleagues. Keep every figure, percentage, and product name exactly as written, and mark vendor statistics as vendor-reported. Distinguish clearly between what the hardware does and what keeps text private. Do not add a purchase recommendation unless the draft already contains one.
Press Ctrl+Shift+Space, paste your notes, and check the diff. A rewrite that keeps the “vendor-reported” hedge on the two-trillion figure is doing its job.
Common Questions
What is a Copilot+ PC?
A Copilot+ PC is a Windows laptop with a dedicated neural processing unit meeting Microsoft’s threshold (around 40-plus TOPS), paired with on-device AI features and hybrid local-and-cloud routing.
Do I need a Copilot+ PC to use local AI for writing?
No. A small writing model in the 1.7B to 4B range runs on an ordinary CPU, including on machines older than any Copilot+ branding. The NPU makes some tasks faster and more efficient, not possible.
Does hybrid routing keep my text private?
Not by default. Hybrid routing may send a request to the cloud when it decides to, so for confidential text you want a tool that keeps the draft local by your explicit choice rather than a router that usually leans local.
Is the two-trillion-inferences figure reliable?
Treat it as directional. Microsoft reported it without a stated method or date range, so it indicates scale rather than a verified count.
Download Wrivio for Windows to keep confidential drafts on your own machine with a local model, whatever chip your laptop has.
Read Next
Why Your Local AI Writing Feels Slow
A local model that feels sluggish is usually fixable. The real causes of slow local AI writing, from cold loads to RAM pressure, and what to change first.
Windows 10 Extended Updates End 13 October 2026: Moving Your AI Writing Setup
Consumer ESU for Windows 10 ends on 13 October 2026. A checklist for moving your writing tools, local AI models, templates, and settings to a Windows 11 PC.
Is the Snapdragon X2 Elite Worth It for Local AI Writing?
The 2026 Snapdragon refresh pushes the NPU to 80 TOPS. For running a small local writing model, here is what that buys you and what it does not.
How To Write A Message To Contest A Charge
Disputing a bill or a card charge is won on specifics, not outrage. How to write the message that gets the charge reversed: the facts, the ask, and the paper trail.
This article is filed underLocal & Private AI, which has 141 articles.