Microsoft's Copilot AI service is set to run locally on next-gen AI PCs, which will require built-in neural processing units (NPUs) with over 40 TOPS of power. Currently, Copilot runs everything in the cloud, creating lag for smaller jobs. Intel's upcoming Lunar Lake chips, set to arrive in 2025, will ship with triple its current NPU speeds. The company also introduced 300 new AI features optimized for its OpenVino platform and announced an AI PC development kit based on its current Meteor Lake silicon.
Microsoft and Intel are collaborating to define the "AI PC," with Intel confirming that Windows PCs will soon utilize NPUs to run Microsoft Copilot locally, reducing its reliance on the cloud. Microsoft is expected to integrate NPUs into Copilot with the 24H2 update, possibly arriving this summer, and is introducing a dedicated Copilot key to the specifications for Windows keyboards. Intel has listed 40 TOPs of NPU performance as a requirement for next-generation AI PCs and is providing resources for developers to create new software to leverage AI. NPUs have begun to appear in various chipmakers' platforms, and while they can improve battery life by alleviating workload on CPUs and GPUs, the impact on system requirements remains unclear.
HP has introduced the industry's most extensive range of AI PCs, featuring NPUs for enhanced AI capabilities, advanced security features, and long battery life. The portfolio includes business convertibles, workstations, conferencing room solutions, and business laptops, all equipped with preinstalled encryption, authentication, and malware protection. The PCs also offer quantum-resistant cryptographic schemes to safeguard UEFI BIOS firmware integrity. Additionally, HP's AI PCs are powered by Intel and AMD processors and are optimized for Windows 11, offering features like Copilot and automatic framing.
Shipments of AI PCs, equipped with neural processing units (NPUs) for running generative AI tasks locally, are projected to reach nearly 50 million units in 2022 and could exceed three times that figure by 2027. These AI PCs are categorized by their NPU performance, with hardware-enabled AI PCs currently shipping and next-generation AI PCs expected to arrive later this year. Advanced AI PCs offering over 60 tera operations per second (TOPS) are anticipated in the coming years. IDC forecasts a sharp increase in hardware-enabled AI PC shipments over the next two years, followed by a dominance of next-gen AI PCs by 2027, with commercial and consumer applications driving the market growth.
Neural Processing Units (NPUs) are AI chips designed to perform AI tasks faster than GPUs and CPUs, reducing the load on these processors and improving overall computer efficiency. NPUs handle short and repetitive AI tasks, such as background blurring in video calls or object detection in photo editing, allowing GPUs and CPUs to focus on their larger assigned tasks. The inclusion of NPUs in the latest generation of devices enables the industry to leverage the latest AI technologies, providing users with more AI-related conveniences and efficient processes. NPUs are expected to bring advancements in video editing, AI filters, and generative AI imagery, opening up new possibilities for gaming handhelds as well.
