Arm, the British semiconductor and software design powerhouse, recently staged its Arm Everywhere China showcase in Shanghai, unveiling a suite of fresh AI‑computing innovations aimed at reshaping the mobile graphics landscape. Central to the announcement is a holistic, AI‑native ecosystem that includes the new CSS for Mobile 2 compute platform and the Mali G2‑Ultra NX GPU – a processor engineered around neural‑accelerated rendering. Arm argues that its on‑device AI workflow can boost visual fidelity while keeping thermal and power footprints far lower than traditional solutions, positioning the approach as a more open alternative to Nvidia’s contentious DLSS 5 and more aligned with developers’ original creative intent. While Arm did not disclose specific product partners, the predecessor Mali G1‑Ultra has already found its way into smartphones from Xiaomi, Vivo and OnePlus, suggesting that the G2‑Ultra NX could appear in a similar range of flagship devices.
Deyan Lazarov, a senior member of Arm’s GPU product team within the Edge AI division, explained that the chip has been in development for roughly two and a half years. Its primary goal is to give mobile developers access to compute‑heavy features such as real‑time ray tracing, which has historically been limited on handheld platforms.
"We’ve seen a steady rise in ray‑tracing adoption in mobile titles," Lazarov noted. "The technology was initially slow to take off, but now more games are actively exploring it." Recent reports from GamesIndustry.biz highlighted that the Mali G2‑Ultra NX enables developers to leverage Unreal Engine capabilities like MegaLights, and Arm collaborated with Sumo Digital to produce the demo title Neural Dawn, a showcase of the GPU’s capabilities.
Lazarov emphasized that the chip targets three distinct categories of games: those that are "mobile‑first" yet aspire to desktop‑level visual quality; cross‑platform titles that need a consistent experience across consoles, PCs and phones; and games that demand a seamless transition for players moving between devices without a noticeable drop in performance. Titles slated to benefit from the new hardware include Tencent’s Arena Breakout Infinite, NetEase’s Where Winds Meet, and Infold Games’ Infinity Nikki.
A key challenge for mobile graphics, Lazarov explained, is the stark difference in thermal envelopes and power budgets compared to consoles or PCs. "Phones can’t dissipate heat the way a console can, and they run on batteries constantly," he said. Arm’s answer is to fuse neural processing directly into the traditional rendering pipeline.
By offloading portions of the workload to dedicated neural accelerators, the GPU’s power draw is reduced, which in turn improves overall system efficiency and frees up resources for developers. "In practice this means you get a new level of efficiency that lifts many of the constraints developers normally face," Lazarov added. "It also means they don’t have to make as many trade‑offs between visual fidelity, frame rate and battery life." To make this promise tangible, Arm is pursuing an open‑software strategy, releasing a toolbox of SDKs, plugins and documentation.
"We’ve spent the last two years working closely with the ecosystem to create a developer‑friendly AI workflow," he said. "The idea is that developers can reap huge efficiency gains and then reallocate that budget toward whatever they value most—whether that’s sharper graphics, smoother performance, longer battery life, or a blend of all three." Arm’s portfolio of AI‑enhanced techniques includes Neural Super Sampling (NSS), Neural Super Sampling and Denoising (NSSD) and Neural Frame‑Rate Upscaling (NFRU).
These technologies collectively upscale lower‑resolution images, synthesize additional rays, and generate extra frames using AI‑driven reconstruction. Lazarov illustrated the process with a two‑frame example: in the first, the GPU renders only a quarter of the image while the neural accelerator reconstructs the remaining three‑quarters via NSS. In the second frame, the GPU does no traditional rendering at all; every pixel is produced by the neuro‑accelerator, resulting in a fully AI‑reconstructed frame. Nvidia’s DLSS 5 also employs AI to enhance image detail, but it has attracted criticism for its opaque, generative models that can diverge from a developer’s artistic vision.
Arm draws a clear line between its approach and Nvidia’s. "Our networks are not generative; they’re convolutional, meaning they don’t invent new content, they simply know how the final image should look based on real data," Lazarov explained. "We train them on authentic content and then release the models publicly on GitHub and Hugging Face." This openness enables developers to fine‑tune the models for their own artistic styles.
"If you have a game with a unique visual language, you can capture data from your own assets and retrain our network so it upscales and inserts frames exactly the way you intend," Lazarov said. "Because you control the retraining, there’s no risk of losing IP or having the AI behave unexpectedly." By contrast, DLSS is described as a closed, proprietary black box. "You can’t see what it’s been trained on, nor can you retrain it with your own data," Lazarov asserted. "It’s a take‑it‑or‑leave‑it proposition, whereas our solution lets you take the model and adapt it to your needs." Another differentiator is the mobile‑first design philosophy.
Arm’s networks are built with strict power, bandwidth and thermal constraints in mind, making them viable on smartphones where DLSS would simply exceed any realistic budget. "DLSS would blow past mobile limits instantly," Lazarov warned. Power efficiency is a cornerstone of the Mali G2‑Ultra NX.
Arm reports up to four‑fold gains in both performance and efficiency, along with a 70 % reduction in DRAM traffic—a major source of power consumption on mobile devices. "The less you need to fetch from external memory, the less power you waste," Lazarov noted. To illustrate real‑world impact, Lazarov compared the performance of last year’s GPU running Neural Dawn at a modest 40 fps under ideal conditions, dropping to 15 fps in typical use.
The new Mali G2‑Ultra NX can comfortably sustain 60 fps on the same title, delivering a noticeably smoother experience. On the tooling side, Arm currently offers Unreal Engine plug‑ins and plans to expand to Unity and other engines. While custom engines won’t receive dedicated plug‑ins, an SDK provides a streamlined integration path. "We’ll meet developers where they are," Lazarov said.
"If you have an existing shipped title, you can integrate our AI‑native graphics in as little as three months and release an upgraded version." The rapid development cycle is exemplified by Neural Dawn itself, built by a team of fewer than 20 engineers in under 18 months. Arm’s technology, Lazarov concluded, empowers developers to quickly enhance existing games or bring groundbreaking new experiences to market without the usual overhead of extensive re‑engineering. In summary, Arm’s mobile‑first AI reconstruction strategy offers an open, efficient and developer‑centric alternative to the black‑box approach of DLSS 5, promising higher visual quality on handheld devices while respecting the strict power and thermal limits inherent to mobile hardware.