Arm, the British semiconductor and software design powerhouse, hosted its Arm Everywhere China showcase in Shanghai this week, unveiling a suite of fresh AI‑computing innovations aimed at reshaping the mobile graphics landscape. Central to the announcement is a new AI‑native compute platform, CSS for Mobile 2, and the Mali G2‑Ultra NX GPU – a processor built around neural‑accelerated rendering that promises higher visual fidelity while keeping heat and power draw to a minimum. Arm positions its on‑device AI solution as a more transparent alternative to Nvidia’s controversial DLSS 5, arguing that its method aligns more closely with developers’ original creative intent.
While the company did not name specific products that will ship with the Mali G2‑Ultra NX, the predecessor Mali G1‑Ultra already powers smartphones from brands such as Xiaomi, Vivo and OnePlus, suggesting a similar market trajectory for the new chip. Deyan Lazarov, a senior member of Arm’s GPU product team within the Edge AI division, explained that the Mali G2‑Ultra NX has been in development for roughly two and a half years. Its design goal is to give mobile developers access to compute‑heavy features—most notably ray tracing—without sacrificing the battery life or thermal envelope required for handheld devices.
"There’s a growing appetite for ray tracing in mobile games," Lazarov said. "It was slow to catch on at first, but now we see more titles actively exploring it." The new GPU integrates tightly with Unreal Engine, enabling developers to leverage technologies like MegaLights. Arm collaborated with Sumo Digital to produce a demo title, *Neural Dawn*, that showcases the chip’s capabilities. Other upcoming games slated to benefit from the Mali G2‑Ultra NX include Tencent’s *Arena Breakout Infinite*, NetEase’s *Where Winds Meet*, and Infold Games’ *Infinity Nikki*.
Lazarov highlighted the distinct challenges of mobile hardware: unlike consoles or PCs, phones operate with limited thermal headroom and rely on batteries for power. "A phone can’t dissipate heat the way a console can, and it’s always running on a limited energy budget," he explained.
To address this, Arm blends 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, dramatically cutting power consumption and freeing up resources for other tasks.
The company’s open‑software philosophy underpins this strategy. Arm has released a toolbox that includes Neural Super Sampling (NSS), Neural Super Sampling and Denoising (NSSD), and Neural Frame‑Rate Upscaling (NFRU). These techniques allow images to be upscaled, additional rays to be simulated, and new frames to be generated via AI‑driven reconstruction. Lazarov illustrated the process with a two‑frame example: in the first frame the GPU renders only a quarter of the scene while the neuro‑accelerators fill in the rest using NSS; in the second frame the entire image is produced by the accelerators, with the GPU doing virtually no traditional rendering work.
Arm’s approach starkly contrasts with Nvidia’s DLSS 5, which has drawn criticism for its “black‑box” nature and for sometimes diverging from the creator’s visual intent. "Our networks are not generative; they’re convolutional models that simply know how things should look based on real‑world data," Lazarov said.
"We train them on authentic content and then make the trained models publicly available on GitHub and Hugging Face." Because the models are open, developers can fine‑tune them with data from their own games. If a studio has a distinctive art style, they can capture screenshots, feed them into the training pipeline, and produce a customized network that respects that aesthetic when upscaling or inserting frames. This openness eliminates concerns about losing intellectual‑property control, a risk that arises when using proprietary, closed‑source solutions like DLSS.
Lazarov emphasized that Arm’s technology was conceived with mobile constraints in mind—power, memory bandwidth, and thermal limits are baked into the network design. "DLSS would simply blow past those budgets on a phone," he warned.
"Our solution operates within every mobile budget, delivering up to four‑times the performance and efficiency of previous generations while slashing DRAM traffic by roughly 70 percent, which is a major source of power draw." In practical terms, the performance uplift is tangible. Running *Neural Dawn* on the prior‑generation GPU capped at about 40 fps under ideal conditions and dropped to 15 fps in typical use.
The Mali G2‑Ultra NX, by contrast, can sustain a smooth 60 fps experience with the same visual settings. To accelerate adoption, Arm has already released Unreal Engine plug‑ins and plans to extend support to other engines, with Unity integration on the roadmap. A recent partnership with Unity China signals that broader engine compatibility is forthcoming. For studios using custom engines, Arm offers an SDK that provides a streamlined path to integrate its AI‑native graphics stack without needing a dedicated plug‑in.
Lazarov noted that retrofitting an existing, shipped title with Arm’s technology can be accomplished in as little as three months, enabling developers to quickly deliver visual upgrades, performance gains, or extended battery life to their user base. The *Neural Dawn* demo itself was built by a team of fewer than 20 people in under 18 months, illustrating how Arm’s tools can accelerate development cycles and lower the barrier to creating cutting‑edge mobile experiences. Overall, Arm’s mobile‑first AI reconstruction strategy aims to give developers a transparent, flexible, and power‑efficient alternative to closed‑source solutions, promising richer graphics on handheld devices without compromising the practical limits of smartphones and tablets.