Arm Holdings Limited (NASDAQ: ARM, hereinafter referred to as “Arm”) today announced at SIGGRAPH the launch of Arm Neural Technology, a groundbreaking technology that introduces dedicated neural accelerators into the Arm GPU set to be released in 2026. This technology elevates GPU performance used for graphics rendering to a higher level, reducing GPU workloads by up to 50% for increasingly complex mobile content that originated in the mobile gaming sector. This is just the beginning, as the introduction of this new technology lays the foundation for more edge-side artificial intelligence (AI) innovations in the industry in the future.
At the same time, Arm also launched the world’s first fully open neural graphics development kit, aimed at integrating AI rendering into existing workflows, allowing developers to start development a year before the hardware is released. All Arm Neural Technology within the kit will be fully open, including the model architectures, weights, and tools needed for game studios to retrain models. Currently, this development kit has the support of partners such as Enduring Games, Epic Games (Unreal Engine), NetEase Games, Sumo Digital, Tencent Games, and Traverse Research.
Geraint North, Arm’s AI and Developer Platform Technology Fellow, stated: “This release marks a significant milestone as desktop-level neural graphics technology can now be realized on mobile devices, which is undoubtedly important for cutting-edge game developers transitioning to edge-side AI. Arm Neural Technology is not only applicable to the gaming field but will also have a tangible impact on applications such as neural camera workloads, providing developers with powerful tools for large-scale implementation of graphics technology in edge use cases, from quality optimization to path tracing.”
Developer-Centric: An Open Development Kit for Neural Technology
Developers can immediately start using the neural graphics development kit to integrate AI graphics technology before the hardware is released. This development kit is designed for mobile games and includes all the resources needed for integrating and customizing AI visual effects, including:
Unreal Engine plugin
PC-based Vulkan simulation
Updated analysis tools
Fully open models provided through GitHub and Hugging Face
Arm Machine Learning (ML) extensions for Vulkan
The open Arm ML extensions for Vulkan allow developers to bring AI directly into familiar rendering pipelines. Traditional Vulkan supports graphics pipelines and compute pipelines, while Arm ML extensions introduce a third pipeline specifically designed for neural network inference, known as the graph pipeline. This enables developers to more easily integrate AI as a native part of the graphics pipeline in mobile rendering.
Practical Applications of Neural Graphics: Neural Super Sampling
Arm Neural Super Sampling (Arm NSS) is an AI-driven graphics optimization engine that fully utilizes the resources within the development kit. It is built on Arm’s Accuracy Super Resolution (Arm ASR) technology, which has been adopted by game studios such as Fortnite and Arknights.
Arm NSS can optimize the resolution from 540p to 1080p with a processing time of four milliseconds per frame, achieving image quality nearly indistinguishable from native quality. Compared to traditional full-frame rendering methods, developers can reduce GPU workloads by up to 50% with Arm NSS, allowing for flexible resource allocation, whether to lower overall game power consumption, increase frame rates, or enhance visual effects. This technology also enables developers to leverage AI to retain surface details, lighting, and motion clarity, allowing for a flexible balance between visual fidelity and power consumption based on game requirements.
Beyond Image Optimization: More Neural Graphics Applications Coming Soon
Arm expects to further expand its neural technology application roadmap in 2026, introducing Neural Frame Rate Upscaling, which uses AI to double frame rates without doubling rendering loads, and Neural Super Sampling and Denoising, which achieves real-time path tracing on mobile devices with fewer rays per pixel. Both applications will be brought to market before the hardware is released.
Through the development of these new technologies, Arm is realizing neural graphics technology that is open, accessible, and optimized for actual performance. Arm provides developers with a unified open platform, making it easier to deploy AI across various edge-side experiences on Arm computing platforms.