Arm's September roundup: Dimensity 9600 Pro phones, neural graphics and agentic AI

Agentic AI across the compute platform

Arm says agentic AI is pushing intelligence everywhere: into the cloud where it is created, to the edge where it becomes personal, and to the physical world where it acts. According to the company, this is turning previously separate systems into one connected computing continuum with more varied workloads, models and software environments.

At Arm Everywhere China 2026, Arm presented a common compute platform spanning Arm CSS for Mobile 2 and AI-native mobile experiences, as well as infrastructure built on Arm Neoverse CSS N4 and the Arm AGI CPU. At Innovate! EMEA, Arm's Dilip Ramachandran and Supermicro's Trent Prestegar discussed how the two companies are working on systems meant to meet the performance, efficiency and scale demands of agentic AI workloads.

On the developer side, the Arm Create AI Optimization Challenge drew more than 330 submissions. Entries covered local LLM inference on wearables, on-device speech processing, CPU-based cloud inference, agent throughput optimization and more efficient mobile vision pipelines. Arm says the winning approaches show developers can get more from Arm-based systems by optimizing the whole workload, not just the model.

Smartphones and mobile gaming

Arm highlighted MediaTek's Dimensity 9600 Pro, which combines two Arm C2-Ultra and six C2-Pro CPU cores with the Mali G2-Ultra NX GPU. Two devices built on the chip were named: vivo's X500 Pro Series and OPPO's Find X10 Series. Arm says these launches show how its CPU and GPU advances move through the ecosystem into finished devices while staying within the power and thermal limits of flagship phones.

In graphics, Arm points to Neural Dawn, a demo that uses Arm Neural Technology and the Mali G2-Ultra NX to deliver what the company describes as fully ray-traced lighting and smoother gameplay with less graphics work. Arm is also extending the approach through Neural Frame Rate Upscaling (NFRU), which uses AI-generated intermediate frames to reduce rendering demands and improve motion.

With the Mali G2-Ultra NX, dedicated neural accelerators sit directly within the GPU. Tencent Games and Unity China are integrating neural graphics into familiar rendering and engine environments, which Arm says gives developers a more practical path to techniques such as Neural Super Sampling and NFRU.

Robotics, wearables and chiplets

On physical AI, Arm argues that the challenge is shifting from adding more AI to making sensing, inference, planning, control and safety work together reliably. Federico Pecora, Senior Principal Robotics Research Lead at Arm, described robotics as an increasingly complex systems problem at the World Robotics Conference. Arm says it is exploring the compute foundation needed for predictable, efficient and safe operation in the real world.

For wearables, Arm pointed to Snap's SPECS smart glasses and Meta's latest lineup of AI devices, personal agents and wearables as examples of intelligence becoming more personal, ambient and always available. Valve also launched Steam Frame, which Arm describes as a new Arm-based wireless VR headset designed for both PC streaming and standalone gaming.

In chip design, Arm highlighted AMBA C2C, which enables chip-to-chip connectivity across AMBA AXI and AMBA CHI. By separating protocol and packetization from the underlying die-to-die transport, it can work alongside standards such as UCIe and support interoperability across chiplets, vendors and generations, according to Arm.

What it means

The roundup is Arm's own account, but it shows how the company is positioning one platform across very different markets. For phone makers and game developers, the news centers on shipping hardware and neural graphics tools. For infrastructure and robotics teams, it points to a push for consistent compute and software foundations as AI workloads spread. The source text is cut off before its final section on research and healthcare.