Arm, the British chip and software design powerhouse, staged its Arm Everywhere China showcase in Shanghai to announce a suite of fresh AI‑computing advances aimed at the mobile market. The centerpiece of the announcement is a full‑stack AI‑native ecosystem that includes the second generation of its AI‑centric compute platform, CSS for Mobile 2, and a brand‑new graphics processor, the Arm Mali G2‑Ultra NX GPU.
This GPU is built around neural‑processing capabilities, promising higher visual fidelity while keeping power draw and heat generation far lower than traditional designs. While Arm did not name specific products that will ship with the Mali G2‑Ultra NX, the predecessor Mali G1‑Ultra has already found its way into smartphones from manufacturers such as Xiaomi, Vivo and OnePlus, suggesting a similar trajectory for the newer silicon. Deyan Lazarov, a senior member of Arm’s GPU product team within the Edge AI division, explained that the G2‑Ultra NX has been under development for roughly two and a half years. Its core ambition is to give mobile developers the tools they need to exploit compute‑heavy features—most notably real‑time ray tracing—without the thermal and battery constraints that have historically limited such techniques on handheld devices.
"We’re seeing a clear uptick in interest from mobile studios for ray tracing," Lazarov said. "It was a slow burn at first, but now more titles are actively looking to integrate it." The new GPU is tightly coupled with Arm’s AI‑enhanced graphics pipeline. By blending conventional rasterisation with neural‑accelerated reconstruction, the chip can offload a large portion of the rendering workload to dedicated neuro‑accelerators.
In practice, this means the GPU may render only a fraction of the final image—sometimes as little as a quarter—while the remaining pixels are synthesized by the AI engine using techniques Arm calls Neural Super Sampling (NSS) and Neural Super Sampling and Denoising (NSSD). In an even more extreme mode, the entire frame can be generated by the neuro‑accelerators, eliminating traditional GPU rendering for that moment entirely. This approach yields several concrete benefits.
First, it slashes the amount of data that must be shuttled to and from DRAM, cutting memory traffic by up to 70 %. Since external memory access is a major source of power consumption on mobile SoCs, this reduction translates directly into longer battery life and cooler operation. Second, the AI‑assisted pipeline can boost effective performance by up to four times, delivering smoother frame rates and higher resolutions without demanding more silicon power.
Arm’s strategy is deliberately open, standing in stark contrast to Nvidia’s DLSS 5, which has been criticized for its closed‑source, black‑box nature. Lazarov emphasized that Arm’s neural networks are convolution‑based rather than generative, meaning they do not hallucinate content but instead learn how existing imagery should appear.
All of Arm’s resolution‑upscaling and frame‑generation models are publicly available on platforms such as GitHub and Hugging Face. This openness lets developers fine‑tune the models with data from their own games, preserving artistic intent and intellectual‑property control. "If you have a game with a very distinctive visual style, you can feed that style into our network and retrain it so the upscaling respects your aesthetic," Lazarov explained.
"You retain full ownership of the process, unlike a proprietary system where you can’t see or modify the underlying training data." The Mali G2‑Ultra NX is already being tested with several upcoming titles, including Tencent’s Arena Breakout Infinite, NetEase’s Where Winds Meet, and Infold Games’ Infinity Nikki. A collaborative demo called Neural Dawn, built with Sumo Digital, showcases the chip’s ability to run at 60 fps on a mobile device—a marked improvement over the previous generation, which struggled to reach 40 fps under ideal conditions and fell to 15 fps in real‑world scenarios. Arm is also investing heavily in developer tooling.
Plug‑ins for Unreal Engine are currently available, and a Unity integration is on the roadmap, complemented by a broader SDK that can be used with custom engines. Lazarov noted that integrating Arm’s AI‑native graphics into an existing shipped title can be accomplished in as little as three months, allowing studios to quickly deliver visual upgrades or performance enhancements to their player base. Beyond gaming, the mobile‑first design philosophy means the AI pipelines are engineered to respect the strict power, bandwidth, and thermal envelopes of handheld devices. Lazarov warned that a direct port of DLSS 5 to a phone would instantly exceed those limits, whereas Arm’s solution was conceived from the ground up to operate within them.
In summary, Arm’s Mali G2‑Ultra NX GPU and accompanying AI‑driven graphics stack aim to democratise high‑quality visual effects for mobile platforms. By offering open, retrainable neural networks, substantial reductions in memory traffic, and a clear path for developers to adopt the technology across a range of engines, Arm positions itself as a compelling alternative to closed‑source solutions like Nvidia’s DLSS 5. The company’s emphasis on efficiency, flexibility, and developer empowerment could reshape how next‑generation mobile games look and feel, bringing console‑level fidelity to devices that run on a battery and a tiny heat sink.