Arm Advances AI Infrastructure with New Neoverse CSS N4 and AGI CPU for Agentic Computing

Arm announced significant advances in its central processing units (CPUs) and graphics processing units (GPUs) across its Edge AI, Physical AI, and Cloud AI business units, with executives detailing the shift toward agentic AI computing and physical AI deployment. The announcements centered on performance per watt advantages and Arm's software ecosystem strength ahead of the Arm Everywhere China event.

According to Ami Badani, Arm chief marketing officer, agentic workflows represent a fundamental shift in compute requirements. "When you look at the agentic workflow, there's a lot happening behind just running the model," Badani said. "Agents are calling tools, they're accessing databases, retrieving information, orchestrating other agents, and managing increasingly complex workloads. All of this is happening on the CPU." Badani highlighted that agents are increasingly moving beyond data centers to edge devices due to token costs, privacy concerns, and latency considerations.

Arm unveiled the Neoverse Compute Subsystem (CSS) N4, its most configurable CSS to date, delivering up to 2x socket-level performance and up to 1.25x performance per watt compared to the CSS N3. The N4 supports up to 128 cores per die, LPDDR6 memory, and PCIe Gen 7 connectivity, with up to 1.75x memory bandwidth of its predecessor. Mohamed Awad, EVP of Cloud AI at Arm, noted that Arm has shipped 1.5 billion NeoVerse cores into data centers since launching Neoverse in late 2018, with 500 million cores shipped in the nine months since the beginning of the year.

Awad confirmed that Arm has overtaken x86 as the primary CPU of accelerated AI data center solutions, driven largely by Nvidia's adoption of Arm CPU cores. In the market for accelerated server solutions, Arm has reached $53.0 billion compared to $34.6 billion for x86, according to IDC market research, though x86 solutions from Intel and AMD remain dominant in the broader $497 billion server CPU market.

The company emphasized the importance of performance per watt as compute requirements for AI grow rapidly. Badani noted Arm's advantage in leveraging a software ecosystem of 22 million developers building on Arm today, creating natural adoption pull across cloud, edge, and physical AI deployments. Drew Henry, executive vice president of physical AI at Arm, discussed the $25 billion physical AI market with a projected $200 billion total available market by 2030.

Chris Bergey, EVP for edge AI at Arm, introduced Arm CSS for Mobile 2, focusing on AI-native CPUs and GPU enhancements for mobile graphics. The company also introduced the Arm Total Design for Physical AI program to support deployment across robotics and other physical systems.