Arm Unveils Neural Network GPU for Developers, Accelerating the On-Device AI Era

Large-Scale On-Device Graphics Implementation
Immediate Development Enabled with Graphics Development Kit
Neural Network Accelerators to Be Introduced by 2026

Global computing platform Arm has ignited the on-device AI competition by unveiling next-generation neural network technology aimed at the graphics and mobile industries.

On August 13 (local time) in Los Angeles, USA, at the computer graphics conference SIGGRAPH, Arm introduced the industry's first "Arm Neural Network GPU (Graphics Processing Unit)" technology and announced that it will integrate dedicated neural network accelerators into Arm GPUs starting in 2026.

This technology significantly enhances GPU performance for graphic rendering. In particular, it can reduce GPU workloads by up to 50% in compute-intensive mobile content, including mobile games.

Image related to Arm neural network GPU (Graphics Processing Unit) technology. Arm.

On the same day, Arm also unveiled its "Neural Network Graphics Development Kit." Designed to integrate AI-based rendering into existing workflows, this kit enables developers to begin development a year before the hardware is released. All of Arm's neural network technologies are provided as fully open platforms, including model architectures, weights, and tools required for model retraining.

Supported partners include Enduring Games, Epic Games (Unreal Engine), NetEase Games, Sumo Digital, Tencent Games, and Traverse Research.

By 2026, Arm plans to expand its neural network technology application roadmap by introducing "Neural Network Frame Rate Upscaling," which doubles frame rates without increasing rendering loads through AI, and "Neural Network Super Sampling and Denoising," which enables real-time path tracing on mobile by reducing the number of rays per pixel. Both technologies will be available before the hardware is released.

Through these advancements, Arm will be able to deliver neural network graphics that are open, highly accessible, and optimized for real-world performance. The company plans to provide an integrated open platform to make it easier to deploy AI across a wide range of on-device experiences based on Arm.

Geraint North, Fellow of Arm's AI and Developer Platforms division, stated, "This announcement marks the full-scale introduction of desktop-level neural network graphics on mobile, serving as a major milestone for game developers at the forefront of the on-device AI transition. It will not only benefit games but also a variety of applications, such as neural network-based camera workloads, providing tools to enable large-scale graphics implementation directly on devices."

This content was produced with the assistance of AI translation services.