The new buzz in the world of PCs is all about the “AI PC”. This term is the latest messaging being banded around from PC/Laptop manufacturers with a view to ensure that people, families and organisations are taking this next generation of PC into account when planning their legacy hardware upgrades over the next 12-18 months.
AI analytics and modeling require vast amounts of data, which are best suited for cloud, but performing some workloads at the device level delivers more efficient processing, or inferencing, by being offloaded to the device. These efficiencies could be applied to:
- Visual inferencing: AI applied directly to the camera feed
- Audio inferencing: AI applied to audio inputs
- Live transcription: AI applied to language processing
New AI-PCs are being designed and built with specialised processors (NPUs) to support executing those models, and more, locally, in real time.
You don’t need to look very far to see the phrase “AI PC” everywhere both from Intel and leading device manufacturers including Microsoft, but what is the tech behind the latest tech phrase?
You may be aware that Microsoft recently launched the Surface Pro 10 and Surface Laptop 6 – both of these fit into the new “AI-PC” catagory!
Note: AI is the capability of a computer program or a machine to "think, learn, and take actions without being explicitly encoded with commands".
So what is an AI PC?
In short an AI-PC needs to have these three key components:
- An Chipset with a new NPU, CPU and GPU built-in
- Windows 11 and Copilot in Windows
- A Copilot Key (technically).
Of course CPUs and GPUs have been around for a long time, but it is the NPU (Neural Processing Unit) that is new in PCs and it is this that is the key part of an AI-PC. The NPU (or AI-Boost as Intel calls it) is designed to process AI workloads locally rather than in the cloud which brings many improvements for the user, including processor efficiency, improved battery life and increased security.
The NPU is a dedicated microprocessor designed to efficiently handle AI-specific tasks, such as executing predictive models, processing machine learning algorithms, or powering generative AI applications.
The Intel Core Ultra processors featured in AI-PCs like Surface Pro 10 and Laptop 6 provide a holistic approach to running AI workloads on PC. These different processors offer programming versatility to developers that maximise the three type of microprocessors – CPU, GPU, and their new, integrated NPU. With an AI-PC, each microprocessor (compute engine) has its own specialisation, but in the context of AI on Cloud PCs mean they work together when using AI workloads with:
- The CPU being ideal for general and traditional compute workload as well as light-weight, single inference AI tasks. Think “accelerating cloud-delivered AI” while minimising impact on overall system performance.
- The GPU being ideal for AI infused into media and 3D rendering apps and tasks. Users can perform data visualisation tasks, create and manipulate 3D images, or generate elaborate reports, enabling your team to efficiently tackle complex data workloads.
- The new AI-Boosting NPU which is designed specifically for sustained and efficient offloading of AI models such as Windows Studio Effects.
What are the benefits of an AI-PC?
In short – performance, battery life and enabling new AI-Powered experiences.
Windows and many other apps have AI features within their applications sets, but put simply with AI-PCs these AI features and tools will work much quicker and will unlock additional NPU-Specific features.
The NPU is a dedicated microprocessor designed to efficiently handle AI-specific tasks, such as executing predictive models, processing machine learning algorithms, or powering generative AI applications. The image below shows the difference between a video app’s back-ground blur using a CPU vs an NPU. The device on the right is a Surafce Pro 9 with dedicated NPU and it is easy to see how much better the AI effect is when it is processing the AI workload locally.
The key thing about the AI-PC is the impact on the CPU processing AI workloads when there is an NPU present. With AI features like voice and video effects in Windows and the wealth of AI features in apps like Adobe, CPU workload is dramatically reduced as the NPU does the AI grunt work (and is far more efficient at running them) This leads to a more efficient device, increased battery life and a more responsive OS and applications for users. The image below illustrates the impact savings running smilar tasks on a PC vs an “AI-PC”.
AI-PCs will empwer and improve many experiences – as compute for AI workloads moves to more local-mode and hybrid-mode processing. This goes beyond just video effects and speech and will include things like object detection “seeing AI”, gesture tracing, security and many more.
Intel claim that “AI PCs with an Intel Core Ultra processor are able to edit videos more than 230% faster and perform AI image editing tasks 52% faster.
AI PCs are also set to be defined by how they align with Microsoft’s Windows operating system as well. One of the most consumer-facing AI products right now is Microsoft Copilot, which is slowly rolling out to Windows 11 devices and we’re set to see a dedicated Copilot key on some laptops in the future too. Over time this will also take advantage of local AI-processing with AI-PCs
Note: Today, Copilot operates in the cloud and thus does not utilize a computer’s onboard NPU. However, this is expected to change, as Microsoft has announced plans for new system architectures that will enhance Windows AI experiences by integrating the GPU, CPU, NPU, and the cloud.
Will the birth of the AI-PC reboot the PC market?
I think the AI PC will be the PC of the next decade.
The PC market has been in decline for several years, but IDC predict that 2024 is estimated to see a 3.7 percent year-over-year increase in shipments, ending the decline the market has seen since the highs of the COVID-19 pandemic.
We are seeing Microsoft, Dell, HP and Lenovo, jumping on AI PCs and with the next two years likely to be the time that more organisations really start to experiment and test AI within their organisations, the rise of the AI-PC is set (according to IDC) to kickstart the PC business again.
