Overview

Arm CSS for Mobile 2:
Enabling AI-native mobile experiences at the edge

AI Summary

AI is moving from reactive assistants to agentic systems that understand intent, coordinate tasks, create content, and act on behalf of users. These continuous workloads place new demands on mobile devices, requiring responsive CPU performance, AI-native graphics, and efficient execution across the entire compute system.

 

Arm CSS for Mobile 2 provides the foundation for the next generation of AI-native mobile experiences, from agentic AI to neural graphics. It brings together the Arm C2 CPU cluster, the Arm Mali G2-Ultra NX GPU, SI L2 system interconnect, software, and developer tools in a configurable platform that helps partners and OEMs tailor performance, power, and product requirements and build differentiated custom SoCs.

Benefits

Built for agentic AI and immersive graphics

  • Responsive agentic AI:

    Up to 1.7x faster execution for AI models supporting voice, personal memory, and reasoning, with the C2 CPU cluster and SME2 helping orchestrate responsive on-device agents.1

  • AI-native graphics:

    Achieve up to 4x FPS at a sustained power budget with neural graphics versus native rendering, enabling smoother, more immersive mobile experiences without increasing power demands.1

  • Accelerated deployment:

    Helps silicon partners and OEMs bring AI-native mobile devices to market faster.

Arm CSS for Mobile 2 benefit image
Use cases

AI-native experiences on mobile

Agentic assistants

AI assistants can understand what users need, remember context, and complete tasks across apps with less friction. CSS for Mobile 2 supports these experiences with responsive on-device compute for faster, more personal AI interactions.

CSS for Mobile 2 Use cases - Agentic assitstants | Desktop view

Live translation

Users can have more natural conversations across languages with responsive speech and translation experiences. The C2 CPU cluster with SME2 enables more accurate, low-latency language processing directly on device.

CSS for Mobile 2 Use cases - Live translation | Desktop view

AI native graphics

Mobile games can look sharper, run smoother, and feel more immersive without exceeding mobile power limits. Mali G2-Ultra NX uses Arm Neural Technology to reconstruct detail and improve visual quality efficiently.

CSS for Mobile 2 Use cases - AI Native graphics | Desktop view

Multimodal AI

Users can interact through voice, images, text, and context in a more natural way. CSS for Mobile 2 coordinates CPU, GPU, and system resources to support richer AI experiences across the device efficiently.

CSS for Mobile 2 Use cases - Multimodal AI| Desktop view
CSS for Mobile 2 Use cases - Agentic assitstants | Desktop view
Inside CSS for Mobile 2

The building blocks for the CSS for Mobile 2 platform

Built agentic AI icon

Built for agentic AI orchestration with C2-Ultra, C2-Pro, and SME2

AI native Mali GPU icon

AI-native Mali GPU with neural technology for AI-accelerated graphics

low latency high bandwidth icon

Lower latency and higher bandwidth for system-level AI performance

Arm CSS for Mobile 2

Visit Arm Product Support for product specifications, technical documentation, downloads, training, security information, and support services.

Partners

Built with the ecosystem, ready for real devices

CSS for Mobile 2 is designed with the Arm ecosystem at the center. Arm works with silicon partners, OEMs, game studios, game engines, and AI developers to bring agentic AI and neural graphics from platform capability into real products and experiences.

Bringing neural graphics into gaming pipelines


See how partners are integrating Arm Neural Technology (NT) and Mali G2-Ultra NX to bring AI-accelerated graphics into mobile gaming, enabling richer visuals, smoother frame rates, and more efficient graphics within mobile power budgets.

Enabling responsive agentic AI on device


See how Arm and Google are using heterogenous computing to run latency-sensitive AI workloads locally and accurately, supporting voice assistants, live translation, speech recognition, multimodal AI, and personalized agentic experiences.

Abstract futuristic digital architecture with glowing purple and blue geometric blocks arranged around a circular portal.
Developer

Build AI for Arm hardware

Discover pre-optimized models, agent-ready code and workflows, and optimization tools that help you take advantage of Arm hardware capabilities and build AI applications from cloud to edge.

Explore AI Portal

Talk with an expert

Learn how Arm CSS for Mobile 2 can help you build responsive agentic AI, neural graphics, and efficient mobile compute into future devices.

Contact us
FAQs

Frequently asked questions

What is Arm CSS for Mobile 2?

Arm CSS for Mobile 2 provides the foundation for the next generation of AI-native mobile experiences, from neural graphics to agentic AI. It brings together our latest IP in a configurable platform, enabling partners to tailor the system to their own performance, power and product requirements and build differentiated custom SoCs.

How does CSS for Mobile 2 support agentic AI?

CSS for Mobile 2 supports agentic AI through the Arm C2 CPU cluster and Arm SME2. The platform helps coordinate repeated AI workflows, manage local execution, and maintain responsiveness across heterogeneous compute resources.

What is Arm Mali G2-Ultra NX?

Arm Mali G2-Ultra NX is an AI-native Mali GPU designed for neural graphics. It integrates Neural Accelerator (NX), a new execution engine, and improvements to ray tracing to support AI-accelerated mobile graphics.

What role does Arm SME2 play?

Arm SME2 enables programmable AI execution on the CPU. This helps run latency-sensitive AI workloads locally, supporting use cases such as speech, translation, lightweight models, and agentic AI orchestration.

Who is CSS for Mobile 2 designed for?

CSS for Mobile 2 is designed for silicon partners, OEMs, developers, game studios, and ecosystem partners building AI-native mobile and edge AI experiences.


  1. Compared to previous generation. Testing done internally, March to August 2026; actual results may vary.