Alright, so this Copilot thing, the new PC Insights feature, it’s a mess, a real mess. Microsoft is rolling it out, slowly, in the US right now. It’s supposed to help you figure out what’s slowing down your Windows 11 machine, right? Like, you ask it “what’s my current CPU usage” or “do I have enough space for a 100 GB game” and it’s supposed to tell you. It pulls data from Windows APIs, monitors CPU, RAM, GPU load, storage space, battery health, USB devices, network status, BIOS version, all that stuff. It can even tell you how much space your Downloads folder takes up, but it won’t read individual files unless you give it explicit permission, which is something, I guess. This granular access is meant to provide comprehensive system diagnostics, yet its own operational footprint raises immediate questions about its utility for performance optimization.

But here’s the kicker, the absolute irony of it all. This Copilot, this AI assistant that’s supposed to fix your slow PC, it’s a RAM hog itself. Windows Latest, they did some tests, and they found Copilot can use up to 1GB of RAM while just sitting there, doing nothing. One gigabyte. That’s a lot of memory, especially for a tool that’s meant to help with performance. PC Gamer also reported it sitting at 560 MB for them while writing an article, making it a very heavy application. It runs as a web app, basically an embedded, private build of Microsoft Edge, and Edge, well, we know Edge likes its RAM. So you’re trying to figure out why your PC is sluggish, and Copilot itself might be a big part of the problem, eating up precious memory. It’s like asking a really hungry person to help you find your missing sandwich, only to discover they’ve eaten half of it before you even started looking. This inherent contradiction undermines the very purpose of the feature, turning a potential solution into another source of system strain.

Windows 11 already uses a fair bit of RAM, you know. A clean install, just sitting idle, can take 2.2 to 3GB of RAM. If you have 32GB of RAM, Windows 11 might use 5-7GB at idle. Microsoft’s minimum requirement for Windows 11 is 4GB of RAM, but that’s just barely getting by, it’s not really optimal for anything beyond light browsing. For gaming or multitasking, you really need 16GB or 32GB. And now Copilot wants another gigabyte? It’s just adding to the bloat. Microsoft has been trying to reduce the OS’s baseline memory usage, but then they do something like this. It’s a contradiction. This pattern of introducing resource-intensive features, despite ongoing efforts to streamline the operating system, creates a frustrating cycle for users and developers alike. The promise of a lean, efficient OS often clashes with the ambition to integrate cutting-edge, albeit demanding, technologies.

This “PC Insights” feature is experimental, it’s opt-in, so you have to give it permission to access your system data. Microsoft says it won’t store your personal files or use them to train AI models, only conversation activity. But still, the resource consumption is a concern. Many PCs are still shipping with only 8GB of RAM, and this Copilot usage could really impact those systems. For users with entry-level or older machines, dedicating a significant portion of their already limited RAM to an AI assistant could severely degrade overall system responsiveness and multitasking capabilities. It’s a question of whether the convenience of asking Copilot is worth the performance hit. Users have been using Task Manager and File Explorer for years, right? Why do we suddenly need an AI assistant that consumes so much memory to tell us basic system info? The traditional tools, while perhaps less intuitive, offer the same data without the overhead, making the value proposition of Copilot’s PC Insights questionable for many.

Microsoft is pushing this whole AI PC thing, with Copilot+ PCs requiring 16GB of DDR5/LPDDR5 RAM and an NPU with 40+ TOPS. These are specialized machines for AI workloads, designed from the ground up to handle such demands efficiently. But the regular Copilot on existing Windows 11 machines, it’s a different story. It’s not fixing the underlying problem of bloat, it’s contributing to it. The distinction between these dedicated AI PCs and standard Windows 11 installations is crucial; what works seamlessly on high-spec, purpose-built hardware becomes a burden on conventional systems. This creates a two-tiered experience, potentially alienating a large segment of the user base who cannot afford or do not need a new AI-centric machine.

I bought some MSFT stock back on July 3, 2023, at $341.00 a share. (The closing price for Microsoft (MSFT) on July 3, 2023, was $341.00.) I’m holding it until it hits $450, or if they somehow manage to make Copilot use less than 100MB of RAM at idle, then maybe I’ll reconsider my sell conditions. That’s a long shot, though, given the current trajectory of resource consumption.

This whole situation, it just highlights the ongoing tension between adding new features and maintaining system performance. Microsoft wants AI everywhere, and that’s clear, but they need to figure out the resource management. It’s a balancing act, and right now, it feels like the scales are tipping towards bloat, potentially at the expense of the everyday user’s computing experience. The long-term success of AI integration into mainstream operating systems will depend heavily on finding a sustainable balance between innovation and efficiency.