Tuesday, May 13, 2025
HomeTechnologyCopilot+ PC Fail: No AI on Desktop? Blame Microsoft!

Copilot+ PC Fail: No AI on Desktop? Blame Microsoft!

Copilot+, AI PC, NPU, GPU, Microsoft, Windows, desktop PC, laptop, AI features, gaming PC, performance, hardware, Copilot Runtime, AI acceleration, DaVinci Resolve, PC industry, Qualcomm Snapdragon X Elite, Intel Lunar Lake, AMD Ryzen AI 300, Surface Laptop 7, Build 2025, Chris Hoffman

The Copilot+ PC Paradox: Why My $3,000 Gaming Rig Can’t Do AI

I just dropped three grand on a new gaming PC. It’s got a beastly AMD CPU and a screaming-fast Nvidia GeForce 5080 GPU. Naturally, I figured I’d be all set to dive headfirst into the latest AI features Windows has to offer. AI-powered file search? Yes, please! Enhanced photo editing? Bring it on!

But here’s the kicker: my shiny new PC is locked out. In fact, no desktop PC on the market can fully access Microsoft’s Copilot+ AI features right now. The reason? You need a specific type of processor called a Neural Processing Unit, or NPU, that meets Microsoft’s stringent qualifications. And, as of mid-2025, there simply aren’t any desktop NPUs that cut the mustard.

It’s been this way since Copilot+ PCs launched back in 2024. While Intel has dabbled with NPUs in some of their desktop CPUs, they haven’t been powerful enough to meet Microsoft’s requirements. AMD, on the other hand, hasn’t even entered the desktop NPU game yet.

So, the reality is stark: if you prefer a traditional, powerful desktop PC, you’re currently excluded from the Copilot+ AI party. This, in my opinion, is a major flaw in Microsoft’s Copilot+ strategy, and it’s a problem without a clear justification.

The term "AI PC" itself has become incredibly muddled. Over the past few months, I’ve reviewed numerous laptops marketed as "AI PCs" or "AI laptops," but the meaning behind those labels has varied wildly. Sometimes it signifies a system with a dedicated NPU for accelerating AI tasks. Other times, it simply means a laptop with a modern CPU and GPU capable of running some AI workloads.

Think about that for a second: these terms, "AI PC" meaning "has an NPU" and "AI PC" meaning "can run AI on the CPU/GPU" are practically opposites! Yet manufacturers are freely using both labels, creating a confusing landscape for consumers.

The PC industry declared 2024 "the year of the AI PC," but the first wave of these machines is already being eclipsed by the Copilot+ PCs. What’s worse, many of the new PCs being sold today in 2025 are also being left in the dust. They may have the raw processing power to handle AI tasks, but they lack the necessary NPU to unlock the full suite of Copilot+ features.

The solution, in my view, is simple: Microsoft should allow PCs to leverage their GPUs for Copilot+ AI features. This would immediately open up these capabilities to a vast range of existing and new machines. Suddenly, all those so-called "AI PCs" could actually run AI features, rather than just the select few equipped with cutting-edge NPUs.

If you’re considering buying a desktop PC right now, you face a difficult choice. Do you wait an indefinite amount of time for desktop NPUs to be announced and released? Or do you buy a desktop PC now and resign yourself to missing out on Copilot+ AI features?

Currently, Copilot+ AI features are limited to laptops powered by Qualcomm Snapdragon X Elite/Plus, Intel Lunar Lake, and AMD Ryzen AI 300 series processors – all mobile platforms with qualified NPUs. There are a few niche exceptions, like lightweight mini PCs with mobile CPUs, such as the Asus ROG NUC series. But, as I’ve said before, no desktop PCs are included.

For me, the decision was clear: Copilot+ features weren’t compelling enough to justify waiting. I already have a Surface Laptop 7 with Insider builds of Windows 11, giving me access to these features. But honestly? I rarely use them. I have Recall activated, but I haven’t touched it in months. Privacy concerns aside, the biggest problem is that it’s just not that useful. Other AI features, like generating images directly in the Photos app, feel more like tech demos than essential tools.

Despite my own lukewarm experience, Microsoft is clearly committed to Copilot+. Recent feature announcements suggest they’re doubling down on AI integration. And some of these newer features actually pique my interest. The idea of AI streamlining PC settings adjustments sounds genuinely helpful. And, of course, my GeForce 5080 GPU could easily handle tasks like that… if Microsoft would just let it. My powerful AMD CPU could probably manage them too.

NPUs occupy a middle ground between CPUs and GPUs when it comes to AI processing. They’re faster than CPUs, but slower than GPUs. However, NPUs are significantly more power-efficient than GPUs, which is crucial for laptops. But on a desktop PC that’s always plugged in? Battery life is irrelevant.

While AI features might put a temporary strain on GPU resources, I’m not particularly concerned if my GPU takes a brief performance hit when searching the Settings app, for example. There’s so much untapped potential in a GPU that isn’t constantly running games or demanding applications. Why not utilize it for AI tasks when it’s idle?

The issue extends beyond the core Copilot+ AI features built into Windows. Microsoft is actively encouraging third-party developers to integrate the "Copilot Runtime" into their apps, enabling AI-powered functionalities. These features will rely on the PC’s NPU to accelerate AI processing, but again, only on Copilot+ PCs.

For instance, the popular DaVinci Resolve video editor can already use the Copilot Runtime to offload certain tasks to the NPU. Let’s assume most developers follow suit and integrate the Copilot Runtime for demanding AI tasks. And let’s further assume that every desktop PC eventually gets a capable NPU. Even then, limiting the Copilot Runtime to NPUs would be a mistake. These apps wouldn’t be able to fully exploit the powerful GPUs found in desktop gaming PCs and laptop workstations. They would be significantly slower at processing AI tasks than they could potentially be. Remember, GPUs may consume more power, but they deliver superior raw AI performance compared to NPUs.

Worse still, developers might be forced to implement multiple AI processing "backends" in their applications: one for NPUs on Copilot+ PCs and another to take advantage of powerful GPUs. That’s extra work for developers, making Windows a less appealing platform. Currently, developers can offload AI processing to cloud servers, ensuring compatibility with their apps across various platforms, including Chromebooks, iPads, Android phones, and Windows PCs without NPUs.

If Copilot+ features were compatible with GPUs, they would "just work" on a much broader range of machines. They would also run significantly faster on systems with capable GPUs. This would give Microsoft a crucial advantage in the nascent field of local AI.

I understand why Microsoft initially launched Copilot+ AI features exclusively for NPUs. It was a strategic move to promote Arm-based laptops like the Surface Laptop 7 and their Qualcomm Snapdragon X Elite/Plus CPUs.

It took Microsoft several months to optimize these features for Intel and AMD systems equipped with the required NPUs. Supporting more hardware required time and effort. They likely anticipated that the PC industry would have a wider array of PCs with NPUs ready by now.

However, almost a year later, we still lack Copilot+ features on desktop PCs, and many laptops remain excluded. Laptop manufacturers are even pushing "AI PCs" and "AI laptops" that can’t run the built-in Windows AI features. This is incredibly frustrating because many of these machines could run the features if Microsoft would simply allow it.

That’s why I believe Microsoft should embrace GPUs and allow Copilot+ PCs to tap into their power. It’s not just beneficial for users; it’s crucial for the future of Windows as an AI application platform.

Unfortunately, Microsoft may never extend Copilot+ features to GPUs. After all, the Copilot+ PC branding is designed to drive hardware sales and encourage upgrades from older Windows versions like Windows 10. Microsoft may believe that freely granting these AI features to existing PCs would undermine that strategy.

While that perspective is understandable from a business standpoint, it’s a shame that PC hardware manufacturers haven’t kept pace. Spending $3,000 on a cutting-edge desktop rig only to feel like it’s already obsolete is a deeply unsatisfying experience. Is the future of Windows exclusively Arm-based laptops? Perhaps. But I’d much prefer a future that doesn’t leave powerful desktop PCs behind.

It doesn’t have to be this way, Microsoft! There’s still time to course-correct. I sincerely hope Microsoft announces GPU support for Copilot+ features at Build 2025 in late May. It would be a win-win for everyone involved.

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular