Arm, the British chip and software design powerhouse, staged its Arm Everywhere China showcase in Shanghai to announce a suite of fresh AI‑driven computing solutions aimed squarely at mobile devices. The centerpiece of the announcement is a new AI‑native compute platform, CSS for Mobile 2, paired with the Arm Mali G2‑Ultra NX graphics processor.
This GPU is built around neural‑accelerator technology, promising higher visual fidelity while keeping heat and power draw far lower than traditional mobile graphics pipelines. Arm’s messaging positions the on‑device AI stack as a more transparent and developer‑friendly alternative to Nvidia’s controversial DLSS 5. Whereas DLSS relies on a proprietary, generative deep‑learning model that many creators feel can stray from their artistic intent, Arm’s solution uses convolution‑based networks that simply “know” how a scene should look based on real‑world training data. Those networks are openly published on GitHub and Hugging Face, allowing developers to fine‑tune them with their own art assets and retain full control over intellectual property.
The Mali G2‑Ultra NX is the successor to the Mali G1‑Ultra, which already ships in smartphones from brands such as Xiaomi, Vivo and OnePlus. Deyan Lazarov, a senior engineer on Arm’s GPU team within the Edge‑AI business unit, explained that the new silicon has been in development for roughly two and a half years. Its primary goal is to give mobile game creators access to compute‑heavy techniques—most notably real‑time ray tracing—without the thermal and power penalties that have historically limited such features on phones and tablets. "We’ve seen a clear uptick in interest from mobile developers for ray‑tracing," Lazarov said.
"The technology was slow to take off at first, but now more titles are looking to incorporate it, and the G2‑Ultra NX is designed to make that feasible." One concrete demonstration of the chip’s capabilities is the demo title *Neural Dawn*, built in collaboration with Sumo Digital. The game showcases how Arm’s AI‑enhanced rendering pipeline can deliver desktop‑class visuals on a handheld form factor.
In addition to *Neural Dawn*, several upcoming mobile releases—including Tencent’s *Arena Breakout Infinite*, NetEase’s *Where Winds Meet*, and Infold Games’ *Infinity Nikki*—will leverage the new GPU’s features. Lazarov highlighted the unique constraints of mobile hardware: unlike consoles or PCs, smartphones cannot dissipate large amounts of heat and must operate continuously on battery power. To address these limits, Arm has tightly integrated neural processing units (NPUs) with the traditional rasterisation pipeline. The NPUs take over portions of the workload, performing tasks such as up‑scaling, denoising and frame‑rate interpolation, which reduces the amount of work the GPU must perform and slashes overall power consumption.
"When the GPU only has to render a quarter of the pixels and the NPU reconstructs the rest, you get a massive efficiency boost," Lazarov explained. "In extreme cases the entire frame can be generated by the neural accelerator, meaning the GPU does virtually no work for that frame." Arm’s AI suite includes three core technologies: * **Neural Super Sampling (NSS)** – an up‑scaling algorithm that reconstructs higher‑resolution images from lower‑resolution inputs.
* **Neural Super Sampling and Denoising (NSSD)** – combines up‑scaling with AI‑driven noise reduction for cleaner visuals. * **Neural Frame Rate Upscaling (NFRU)** – generates intermediate frames to increase perceived frame rates without extra GPU rendering. These techniques allow developers to allocate saved GPU cycles to other priorities, whether that be richer textures, higher frame rates, longer battery life, or a blend of all three. Arm’s open‑source stance means developers can download the pre‑trained convolutional networks, retrain them on bespoke art styles, and embed the customised models directly into their games.
This flexibility eliminates the risk of losing creative control—a criticism often levied at Nvidia’s closed‑source DLSS implementation. "DLSS is a black box you can’t peek into or re‑train," Lazarov said. "Our networks are public, modifiable, and built to run within the tight power, bandwidth and memory budgets of a phone. Trying to run DLSS on a mobile device would instantly blow those limits." Power efficiency is a key selling point.
Arm claims the G2‑Ultra NX can deliver up to four times the performance and four times the efficiency of its predecessor, while also cutting DRAM traffic by roughly 70 %. Since external memory accesses are a major source of power drain on mobile SoCs, this reduction translates directly into longer battery life and cooler operation. In real‑world testing, the previous‑generation Mali GPU could only sustain *Neural Dawn* at about 40 fps when pushed to its limits, and dropped to roughly 15 fps under normal conditions.
The G2‑Ultra NX, by contrast, comfortably maintains 60 fps, demonstrating the tangible impact of AI‑assisted rendering. From a developer tooling perspective, Arm currently offers plug‑ins for Unreal Engine and is working on Unity integration, having recently announced a partnership with Unity China.
An SDK provides a streamlined path for studios using custom engines, enabling them to embed Arm’s AI graphics without needing a dedicated plug‑in. Lazarov noted that retrofitting an existing shipped title with Arm’s technology can take as little as three months, allowing publishers to quickly roll out visual upgrades or performance improvements. The broader vision is a unified gaming experience across devices. Whether a player is on a high‑end console at home or a phone on a commuter train, Arm’s AI‑native graphics aim to deliver consistent visual quality without the steep compromises that traditionally separate mobile and desktop gaming.
In summary, Arm’s mobile‑first AI reconstruction strategy diverges sharply from Nvidia’s DLSS 5 by embracing openness, developer control, and strict adherence to mobile power and thermal envelopes. By embedding neural accelerators directly into the graphics pipeline, Arm promises to unlock ray tracing, high‑resolution up‑scaling and high‑frame‑rate generation on smartphones, potentially reshaping the expectations for mobile gaming graphics in the years ahead.