Powering Android Smartphones’ On-Device AI With SME2@primaryHeadingTag>
Google and Arm are transforming mobile AI with Scalable Matrix Extension (SME2), a set of advanced matrix compute instructions integrated into Android through KleidiAI. This enables developers to deliver efficient, real-time AI across billions of devices without rewriting code. From real-time translation to intelligent assistants, SME2 accelerates generative AI experiences directly on CPUs, reducing latency and improving energy efficiency.
Google and Arm are transforming mobile AI with Scalable Matrix Extension (SME2), a set of advanced matrix compute instructions integrated into Android through KleidiAI. This enables developers to deliver efficient, real-time AI across billions of devices without rewriting code. From real-time translation to intelligent assistants, SME2 accelerates generative AI experiences directly on CPUs, reducing latency and improving energy efficiency.
Arm SME2: Accelerating Android Mobile AI Workloads
SME2 is the latest CPU extension on Arm Lumex CSS Platform, the advanced subsystem for next-gen devices, designed to accelerate matrix-oriented compute workloads directly on device. It improves performance for AI and ML models, especially those that rely on operations like matrix multiplication, common in transformers, convolutional neural networks (CNNs), and large language models (LLMs).

Android integrates SME2 using Arm KleidiAI within XNNPACK, LiteRT, MediaPipe and other popular frameworks. This allows AI models like Gemma 3 to deliver 6x faster responses and instant summarization directly on CPUs. Developers benefit automatically by using supported libraries and frameworks.