Arm details its mobile-first AI reconstruction approach, contrasting with the "black box" style of DLSS 5

Arm, the UK‑based semiconductor and software design firm, recently staged its Arm Everywhere China showcase in Shanghai to announce a suite of new AI‑focused technologies aimed at reshaping mobile graphics. The centerpiece of the announcement is the AI‑native compute platform CSS for Mobile 2 and a brand‑new GPU, the Mali G2‑Ultra NX, which is built around neural processing units rather than traditional raster pipelines.

Arm argues that its on‑device AI‑driven graphics pipeline offers a more transparent and developer‑friendly alternative to Nvidia’s controversial DLSS 5, aligning more closely with the original intent of game creators. While Arm did not disclose specific handset models that will ship with the Mali G2‑Ultra NX, its predecessor, the Mali G1‑Ultra, already powers devices from brands such as Xiaomi, Vivo and OnePlus. Deyan Lazarov, a member of Arm’s GPU product team within the Edge AI division, explained that the team has spent roughly two and a half years crafting the new chip.

Their goal was to give mobile developers access to heavyweight features—most notably real‑time ray tracing—without sacrificing the power and thermal envelopes that smartphones must respect. "We’re seeing a genuine uptick in ray‑tracing adoption on mobile titles," Lazarov said.

"It moved slowly at first, but now more developers are actively looking to integrate it into their games." The Mali G2‑Ultra NX is designed to support Unreal Engine capabilities such as MegaLights, and Arm partnered with Sumo Digital to produce a demo title, Neural Dawn, that showcases the chip’s capabilities. Lazarov emphasized that the GPU targets "mobile‑first" games that still aspire to the visual fidelity of desktop experiences. He also highlighted cross‑platform titles that aim to deliver a consistent experience whether the player is on a console at home or a phone on a train.

To achieve this, the chip must operate within the strict thermal and power limits of a battery‑driven device, which differ dramatically from the generous budgets of consoles and PCs. Games slated to benefit from the Mali G2‑Ultra NX include Tencent’s Arena Breakout Infinite, NetEase’s Where Winds Meet, and Infinity Nikki from Infold Games. "Mobile devices have unique thermal constraints and power budgets," Lazarov noted.

"They can’t dissipate heat the way a console or a desktop can, and they run constantly off a battery." Arm’s answer is to fuse neural processing with the conventional rendering pipeline. By offloading portions of the workload to dedicated neuro‑accelerators, the GPU can render fewer pixels directly and let AI reconstruct the remainder.

This reduces both GPU compute time and overall power draw, which in turn improves system‑level efficiency. "The result is a new level of efficiency that eases many of the compromises developers normally face," Lazarov explained. "They can allocate saved resources to higher visual fidelity, better performance, longer battery life, or any combination thereof." To foster adoption, Arm is taking an open‑source stance.

The company provides a toolbox of software utilities and has made its neural networks publicly available on GitHub and Hugging Face. "Our networks are convolutional, not generative, so they don’t hallucinate content," Lazarov clarified. "We trained them on real‑world data and then released the models so developers can fine‑tune them with their own art styles. This keeps intellectual‑property control firmly in the hands of the creator." By contrast, Nvidia’s DLSS 5 relies on proprietary, closed‑box models that cannot be retrained by developers.

"DLSS is a black box—you don’t know what it’s been trained on or how it will modify your art," Lazarov said. "Our approach is transparent: you can download the model, retrain it on your own dataset, and be confident that the upscaled frames respect your artistic intent." Another distinction is the mobile‑first design philosophy. Arm’s neural networks are built to operate within the tight power, bandwidth, and memory constraints of smartphones.

Attempting to run DLSS on a phone would instantly exceed those limits. Arm claims the Mali G2‑Ultra NX delivers up to four times the performance and efficiency of its predecessor, while slashing DRAM traffic by roughly 70 percent—one of the biggest sources of power consumption on mobile chips. In practical terms, the new GPU can run Neural Dawn at a steady 60 fps, whereas the previous generation struggled to reach 40 fps under ideal conditions and fell to 15 fps in typical use.

Development tools are already available as Unreal Engine plug‑ins, with Unity support on the roadmap and a broader SDK for custom engines. Arm recently announced a partnership with Unity China, underscoring its commitment to reach developers wherever they work. Lazarov stressed that integrating Arm’s AI‑native graphics into an existing title can be done in as little as three months, allowing studios to quickly ship enhanced versions of games already in the market. The Neural Dawn demo itself was built by a team of fewer than 20 people in under 18 months, illustrating how the technology can accelerate development cycles while delivering high‑quality visuals.

In summary, Arm’s Mali G2‑Ultra NX and its associated AI reconstruction suite—Neural Super Sampling (NSS), Neural Super Sampling and Denoising (NSSD), and Neural Frame Rate Upscaling (NFRU)—offer a transparent, open, and mobile‑centric alternative to Nvidia’s DLSS 5. By embedding neural accelerators directly into the GPU pipeline, Arm aims to give developers the flexibility to enhance resolution, frame rate, and visual effects without sacrificing battery life or thermal headroom, thereby bringing desktop‑grade graphics to the palm of a hand.