Arm is building for an AI market that refuses to sit still. The company has introduced new compute platforms for data centers and mobile devices, pairing higher performance claims with more options for customers. That matters because AI workloads are spreading across infrastructure, phones, and research pipelines at the same time.
For cloud and data center systems, Arm is positioning Neoverse CSS N4 as its most configurable CSS to date. The company also calls it the fastest route from a compute subsystem to finished silicon, a claim aimed at customers that want to adapt their hardware without starting from an empty design sheet.
Against Neoverse CSS N3, Arm claims the N4 design delivers up to twice the performance, up to 1.25 times the performance per watt, and up to 1.75 times the memory bandwidth. Those figures target the three pressures that shape AI infrastructure: getting more work from a chip, controlling power use, and moving data without leaving expensive compute resources waiting.
“There is no one-size-fits-all approach to AI infrastructure, so Arm gives customers flexibility to choose what best fits their workloads,” said Mohamed Awad, Arm’s cloud AI executive vice-president. The message is straightforward: AI infrastructure will not be solved by one fixed design, no matter how many launch presentations insist otherwise.
Mobile devices get a combined AI platform
Arm is also directing its platform strategy at mobile hardware. Chris Bergey, Arm’s edge AI executive vice-president, said, “Mobile devices need a compute platform built for this new era, bringing CPUs, GPUs and system IP together in a way that developers can easily utilise and apply to next-generation AI workloads.”
That platform includes the Arm Mali G2-Ultra NX, which Arm claims delivers up to four times higher performance per watt for neural graphics. The focus is not only on raw processing speed; the performance-per-watt figure points to the power limits that define mobile devices, where every useful operation competes with battery life.
Arm also introduced C2-Ultra as a successor to C1-Ultra. Against C1-Ultra, C2-Ultra offers up to 1.7 times higher AI performance and 15% higher single-thread performance, while using up to 38% less power at the same performance.
These figures give Arm a broader argument than a faster processor alone. The company is presenting CPUs, GPUs, system IP, memory bandwidth, and configurable compute subsystems as parts of one AI platform strategy, covering data centers and mobile devices without pretending they have identical needs.
Rene Haas links hardware demand to medical research
Arm CEO Rene Haas framed the demand behind these platforms in larger terms. He said, “AI could help humanity cure cancer within our lifetimes.” That is a vast promise, and no chip specification can prove it, but the connection he drew is specific: AI may reduce the time needed to develop and test drugs.
“AI is going to not only shorten the amount of time that those drugs can be invented, it’s going to shorten the amount of time that you test them,” Haas said. The statement places medical research beside data centers and mobile devices in the same story — AI needs computing capacity before it can deliver results in any field.
Haas also said, “AI’s demand won’t slow.” That prediction explains the push across Arm’s new products: more performance, better performance per watt, wider memory bandwidth, and designs that customers can configure for their own workloads.
The numbers are claims, not guarantees. Still, Arm’s announcement shows where the company wants to compete: in the infrastructure that powers AI, the mobile platforms that bring AI to devices, and the hardware needed for applications that may eventually change how drugs are invented and tested.
Based on
