Meta has become the first customer for the new CPUs. According to Arm CEO Rene Haas, the development could generate around $15 billion in annual revenue by 2031. The company’s total revenue could reach $25 billion, with earnings per share projected at $9.

CPUs are regaining strong demand amid the rise of agentic AI, which is reshaping computing requirements. Arm expects demand for such chips to grow fourfold.

“In modern data centers, the CPU manages thousands of tasks: coordinating accelerators, managing memory and storage, distributing workloads, and moving data. With the emergence of agentic AI, CPUs will also coordinate the work of multiple assistants,” the company said.

The launch of the Arm AGI CPU marks a significant shift in the company’s business model. Previously, Arm did not manufacture its own chips and spent 35 years generating revenue primarily through licensing and royalties.

“The Arm AGI CPU delivers high performance per task while running thousands of cores in parallel, staying within power and cooling constraints,” the company noted in a blog post.

Arm is best known for the architecture powering most modern smartphones. In 2018, it began competing with x86 server chips from Intel and AMD by launching the Neoverse platform.

“For the first time in more than 35 years, Arm is releasing its own chips — expanding Neoverse beyond IP and Arm Compute Subsystems to give customers more deployment options, from building their own chips to integrating platform solutions or using processors designed directly by Arm,” the company said.

Alibaba unveils rival chip

Chinese tech giant Alibaba also introduced its own solution — the XuanTie C950 CPU designed for agentic AI capabilities.

The chip is capable of handling multi-step tasks performed by AI assistants and is intended for inference workloads in data centers.

Until recently, the industry focused primarily on GPUs, a market dominated by Nvidia. These chips are essential for training large-scale models because they can perform many calculations in parallel.

In contrast, CPUs typically execute operations sequentially, making them better suited for the emerging ecosystem of AI agents that rely on executing structured task chains.

The XuanTie C950 processors can be customized for specific inference scenarios, allowing customers to adapt processors to their workloads. The chip is based on the open RISC-V standard, which serves as an alternative to Arm’s architecture.

Arm’s move into building its own AI data center chips signals a major shift in the AI infrastructure race. As agentic AI grows, CPUs are becoming critical for coordinating complex workloads, intensifying competition with Nvidia, Intel, AMD, and emerging players like Alibaba.