Running a PC with Microsoft’s Copilot+ fundamentally changes your PC

I had the opportunity to attend the recent launch of Microsoft’s new Surface products, which featured the first Copilot+ PCs, a new category of Windows PCs specifically designed for artificial intelligence. At this launch, Microsoft showed its new vision for the PC, which includes AI at the operating system level while encouraging developers to create applications for this new platform.

While on the surface the platform doesn’t look like a major departure from Windows versions of the past, it actually implements a lot of changes under the hood to both hardware and software. When I dig into the details, I see a lot of things from Microsoft that give us a glimpse into the future of the PC—and some of those tangible experiences and tools are already available today.

Copilot+ computer

The Copilot+ PC is fundamentally different from the “AI PC” of the last six months, mainly because almost every PC company has its own definition of what constitutes an AI PC; but Microsoft’s definition is very specific. It’s also important to note that Copilot+ doesn’t just mean a computer that’s been optimized to run on the entry-level cloud-based Microsoft Copilot. Copilot+ computers are specifically designed to run device-side artificial intelligence at the operating system level and above. This set a very specific set of hardware and architecture requirements as well as optimizations for specific models.

For starters, Microsoft has set minimum hardware specs including 40+ TOPS per NPU and a minimum RAM configuration of 16GB. In order to launch Copilot+, Microsoft had to deepen its already close partnership with Qualcomm and solve some of the existing problems with using Windows on Arm. Currently, the Copilot+ computer is exclusively powered by Qualcomm Snapdragon X Plus and Snapdragon X Elite processors, both of which feature NPUs with 45+ TOPS. The TOPS minimum NPU requirement is critical because it sets the standard for the performance and power consumption that developers can expect from a Copilot+ computer, letting them know what AI computation they have to play with in their applications. If you’d like to learn more about NPUs and their origins, check out this more detailed analysis I wrote a few weeks ago.

These Copilot+ PCs come with a new version of Windows 11 that is naturally AI-accelerated and optimized for Arm processors. In fact, this is the first time ever that Microsoft has led with Arm processors as opposed to Intel’s x86 architecture, which is also used by AMD. During the launch, Microsoft made some very bold performance claims against Apple’s 15-inch MacBook Air, including 58% better sustained multi-threaded performance, all while claiming 20% ​​longer video playback with 22 hours of battery life. Microsoft sought to allay concerns about Arm64 app compatibility, claiming that 87% of the total app minutes people spend on their computers will be in apps that already have native Arm64 versions. Microsoft also has a new x86 emulator called Prism, which it claims works at least as well as Apple’s Rosetta 2.

As part of the launch, Microsoft introduced a pair of new Surface PCs, the Surface Pro (Gen 11) and the Surface Laptop (Gen 7), both of which meet all the minimum specifications for a Copilot+ PC. Microsoft also hosted a number of PC partners on site to show off their own Copilot+ PCs, which I’ll cover in more detail in an upcoming article comparing prices, specs and options. The important thing to know is that the Copilot+ PCs start at $999 and start shipping on June 18 – the first day the Snapdragon X Elite is available to the public.

Microsoft’s first-party AI applications and OS-level AI

At the launch of Copilot+, Microsoft showcased several first-party AI applications that it believes will transform the Windows user experience. One of the most controversial—and potentially useful—of these apps is Recall, which periodically takes screenshots of your device usage and then lets you go back and find things you saw while you were working or browsing. Microsoft has already moved away from implementing opt-out towards opt-in to satisfy critics.

Another app is Cocreator, which allows a user to create something in Paint and use generative artificial intelligence to help iterate that artwork into a more polished product. Cocreator can also help the user transform an existing photo or image into a certain style – one of the first GenAI applications. Microsoft has also enhanced its already AI-accelerated Studio Effects application with the new Enhanced Windows Studio Effects, further improving lighting and adding new features such as a teleprompter for eye contact.

Another new feature for the entire operating system is the implementation of live subtitles from more than 40 languages ​​into English; this can be done in any application and is a good example of the power of AI acceleration at the operating system level. For gaming, Microsoft has also implemented a feature called Automatic Super Resolution (Auto SR for short), which improves gaming performance by rendering games at a lower resolution and then using the NPU to upscale the resolution back to the original level.

Third-party applications and frameworks

While Microsoft has built a respectable list of developers already shipping apps with Copilot+ support, the company needed to implement the new Windows Copilot Runtime to enable both third-party and first-party apps. This new Copilot Runtime is the AI ​​core that enables Copilot+ PCs by extending the Microsoft Copilot suite on Windows. Included in the Copilot Runtime is the Copilot library, which is a set of APIs that power the 40+ on-device AI models that ship with this new version of Windows.

Microsoft has also implemented a new semantic index that helps redefine search in Windows and is the backbone of features including Recall. Microsoft says it will show this capability to developers later this year with the Vector Embeddings API so that developers can implement their own vector store and RAG in their own applications. Microsoft also created Phi Silica, a small local language model based on the Phi family of models that are specifically designed to run on devices and optimized for NPUs. Microsoft also announced support for PyTorch, with its DirectML framework designed to abstract hardware and make it easier to support different hardware. Microsoft is also envisioning the use of artificial intelligence for the web with the launch of WebNN, which also uses DirectML. WebNN is a Windows API that is expected to allow the open internet to access AI hardware including CPU, GPU or NPU on a device to accelerate AI on the web.

Adobe came out of the gate with support for Photoshop, Lightroom, and Express in Windows Copilot+, and said it will add support for Illustrator, Premiere Pro, and other applications later this summer. Adobe’s suite of applications has strongly embraced AI creation and editing; in the past, most of these capabilities relied on the GPU, which can be powerful but also power-hungry. Adobe’s biggest video editing competitor, DaVinci Resolve, has also announced support for features like Magic Mask in Resolve Studio. DaVinci Resolve has proven time and time again to be an ambitious adopter of new technologies and features, and I’m not surprised that it was involved in the launch of Copilot+.

Microsoft showcased other applications including CapCut, LiquidText, djay Pro and Cephable to demonstrate the power of the NPU. I was particularly impressed with how easily the NeuralMix feature in djay Pro isolated different tracks within a track to make them easier to queue for the DJ. Cephable, a facilitation application, was the only application shown that I saw using it, demonstrating the benefits of using NPU acceleration with its facilitation tool. I was able to control the slide just by tilting my head. It was great to see how much NPU it was using in real time while tracking my face and head movements. Cephable, even compared to an M3 MacBook and an x86 plus GPU configuration, showed improvements in performance, battery life, and latency by running on the NPU.

These apps only scratch the surface of what will be possible with Windows on Copilot+ PCs, and we can expect developers to take some time to understand the full potential of the platform and optimize their software for it. Fortunately, developers now have more tools than ever, including the newly announced Snapdragon Dev Kit for Windows, which is supported by Qualcomm’s new AI Hub, which the company recently expanded to support PC-to-device models.

Wrap up

Microsoft undoubtedly changed the trajectory of the PC industry with the announcement of the Copilot+ PC and the suite of first-party apps that support it, plus the new third-party frameworks and apps that came along the way. Microsoft has been working with Qualcomm on this for years, and this launch ushered in a new era of Arm-based computing. It will be interesting to see how Apple responds to its own AI push at WWDC in a few weeks, but I don’t believe Apple’s approach will be that advanced. Yes, Apple benefits from full vertical integration, but the collaboration between Microsoft and Qualcomm seems to be so closely in sync that things aren’t far off with these Surface Copilot+ PCs.

After Copilot+, I believe Apple is now playing catch-up; after all, it launched the M4 iPad Pro somewhat prematurely with heavy AI PC counter-messages. I believe that the AI ​​race for PC and Mac has only just begun and that we are in the early innings of the AI ​​game, so it is still too early to declare any winners. We will likely continue to see companies try to compete with each other on AI TOPS performance, whether on NPU, CPU or GPU. Copilot+ will hopefully help solve some of the expected fragmentation issues on x86 and Arm platforms, and hopefully iron out the typical friction we see with new Windows platforms. By setting strict expectations for NPU performance and minimum memory specifications, Microsoft helps developers understand what kind of hardware they can expect from Copilot+ computers. I’m excited to use these new Copilot+ PCs in the real world and compare them to the M4 iPad Pro as well as older PC platforms. Look for more of my coverage in the coming weeks and months.