Neural Processing Units (NPUs) are specialized processors designed to handle AI and machine learning tasks more efficiently than traditional CPUs or GPUs. Big tech companies like Google, Microsoft, and Apple are increasingly integrating NPUs into their devices to support on-device AI capabilities, promising faster and more efficient AI applications. However, the current software demand for these capabilities has not yet caught up with the hardware advancements, leading to a mix of hype and genuine innovation in the tech industry.
Microsoft has introduced the AI-powered Windows Recall feature for Windows 11, designed to run on Copilot+ PCs with powerful NPUs. However, a user has managed to get it working on unsupported Arm64 hardware, demonstrating that the feature can function without an NPU. Despite its potential, the feature has raised significant privacy and security concerns, as it captures and stores snapshots of user activity. Microsoft assures that the feature is privacy-focused and runs locally, but users remain skeptical.
Intel and Microsoft are working on developing "AI PCs" with built-in neural processing units (NPUs) and Microsoft's Copilot key for keyboards, aiming to run more elements of Copilot locally on the client to reduce lag and improve user privacy. Microsoft will require NPUs in "next-gen AI PCs" to hit speeds of 40 trillion operations per second (TOPS), with Intel, AMD, and Qualcomm competing in this space. Intel is offering an "AI PC development kit" centered on an Asus NUC Pro 14, a mini PC built around Intel's Meteor Lake silicon, as part of its developer program.
Intel has revealed that a minimum of 40 AI TOPs is required to run Microsoft's Copilot locally on Windows-based AI PCs, marking the era of AI PCs for consumers. Current CPUs in the market fall short of this requirement, with AMD's Hawk Point APUs and Intel's Meteor Lake SKUs being the closest. However, Qualcomm's Snapdragon X Elite SoC promises to meet the threshold with its "Hexagon" NPU. Future lineups from AMD and Intel are expected to surpass this requirement, with Intel's Lunar Lake CPUs offering 3x the NPU AI performance uplift versus Meteor Lake CPUs, and AMD planning to offer a 3x AI NPU uplift with its upcoming Strix Point APUs.
AMD has launched the Ryzen 7 8700G processor featuring Zen 4 cores, AMD Radeon 780M graphics, and a new AMD XDNA NPU for AI capabilities. The NPU aims to support AI applications such as image editing and background blurring. The processor also boasts a 65W TDP, high-end integrated graphics, and PCIe Gen4 support, making it suitable for gaming and AI inference tasks.
AMD debuts the Ryzen 8000G series chips with a dedicated neural processing unit (NPU), making it the first desktop processor with such a feature. The NPU is designed to handle AI-based tasks on desktop computers without relying on cloud processing. This move marks the beginning of the AI PC revolution in the desktop market, with AMD aiming to capitalize on the growing interest in AI applications. The new chips also boast the world's fastest integrated desktop graphics processor, allowing for impressive gaming performance. This launch is a direct challenge to rival Intel as AMD continues to gain market share in the tech industry.
