Arm Holdings has announced the launch of its first in-house chip in its nearly 35-year history, signalling a major transformation in its long-standing business model. The development, revealed in March 2026, marks a shift from designing and licensing chip architectures to directly producing its own semiconductor products.
The newly introduced processor, called the Arm AGI CPU, is designed specifically for artificial intelligence workloads in data centres. Unlike graphics processing units that dominate AI training, this CPU focuses on managing large-scale computing tasks such as memory handling, workload scheduling, and data movement, which are critical for efficient AI system operations.
Arm developed the chip using its Neoverse CPU architecture and in collaboration with Meta, which will also serve as the first customer. The chip is intended to work alongside Meta’s AI accelerators, helping optimize both training and inference processes within large-scale AI infrastructure.
The move represents a historic departure for Arm, which has traditionally licensed its chip designs to companies such as Apple, Nvidia, and other semiconductor firms rather than manufacturing its own processors. By entering the chip production space, Arm is now positioning itself as a direct competitor to some of its own partners in the semiconductor ecosystem.
The company began developing this chip around 2023, reflecting a strategic pivot driven by the explosive growth of artificial intelligence and increasing demand for specialized computing hardware. The processors are already production-ready and available for order, indicating Arm’s readiness to scale its presence in the AI hardware market.
This development also highlights a broader shift in the semiconductor industry, where companies are increasingly building custom chips tailored for AI workloads. As demand for efficient and scalable computing rises, CPUs are becoming a critical component in managing distributed AI systems, complementing GPUs and other accelerators.
Overall, Arm’s entry into chip manufacturing marks a turning point for the company and the industry at large. By moving beyond its traditional licensing model, Arm is positioning itself at the center of the AI infrastructure race, aiming to capture greater value in a market increasingly defined by specialized hardware and large-scale computing needs.
