Jakarta, INTI - Arm, a British semiconductor and software design company, is starting to produce its own chips. For more than 30 years, Arm has designed chip architecture and licensed it to third-party producers and manufacturers.
CEO Arm, Rene Haas, said that the company aims to enter the competition in the growing demand for computing and artificial intelligence (AI).
“AI has fundamentally redefined how computing is built and deployed. Agentic computing is accelerating that change,” said Haas. “With the expansion into delivering production silicon with our Arm AGI CPU, we are giving partners more choices, all built on Arm’s foundation.”
Arm’s in-house ship is named Arm AGI CPU, especially designed for the needs of generative AI and agentic AI. The main function will be aimed at the use of high-performance data centers, not mobile phones or PCs.
Arm says this CPU will be the computing foundation for the "agentic AI cloud" era, where AI systems run continuously and coordinate with each other automatically without human intervention.
In this context, Arm states that the role of the CPU is becoming increasingly crucial. This is because the CPU must now manage thousands of AI tasks in parallel in a short time. These tasks include memory and storage management, workload management, and coordinating interactions between AI agents on a large scale.
Specification of Arm AGI CPU
Secara teknis, Arm AGI CPU dibangun menggunakan proses fabrikasi 3 nanometer (nm) oleh Taiwan Semiconductor Manufacturing Company (TSMC).
Arm AGI CPU is currently built entirely in Taiwan, at the Taiwan Semiconductor Manufacturing Company (TSMC). The manufacturing process uses a 3-nanometer (nm) fabrication process. Arm AGI CPU has up to 136 cores based on the Neoverse V3 architecture and is designed for 1U rack servers.
For a full data center rack configuration, the core count can reach over 8,000 cores. In large-scale liquid-cooled systems, the core count can be significantly increased to 45,000 cores per rack. In terms of performance, Arm claims the Arm AGI CPU is capable of delivering up to twice the performance of x86-based CPUs in agentic AI workloads.
However, this CPU still boasts superior power efficiency (performance per watt), which is expected to significantly reduce data center electricity costs. The Arm AGI CPU is also designed to work alongside existing AI accelerators such as GPUs or NPUs.
This chip will function as an "orchestrator" that manages data flow, networking, and coordination of large-scale AI systems. Arm targets mass availability of the Arm AGI CPU in the second half of 2026.
Meta and OpenAI Want to Use Arm’s Chips
Although AGI CPU is Arm's first in-house chip, several major technology companies have expressed interest in the product. Meta is said to be a major customer and an early development partner. According to Meta, the chip can provide a more efficient computing platform, increase data center performance density, and support the development of the company’s future AI systems.
In addition, OpenAI, SAP, Cerebras, Cloudflare, and Korean companies such as SK Telecom and Rebellions have also agreed to use this chip in their infrastructure.
OpenAI says the Arm AGI CPU will play a critical role in improving the efficiency, bandwidth, and coordination of large-scale AI systems used by hundreds of millions of users.
With Arm starting to produce its own chips, there is speculation that it will create fiercer competition in the industry. Arm will transform from a partner into a competitor for companies that previously used its designs.
Arm's move out of its comfort zone demonstrates the vast business opportunity in the AI chip market, to the point that they are tempted to develop their own chips. Going forward, the market for AI chips for data centers is expected to continue to grow as computing demand in global data centers increases, and new competitors
Conclusion
Arm's move to produce its own chips with the Arm AGI CPU marks a major shift from a licensing business model to a direct player in the AI hardware market, driven by the surge in computing demand in data centers. These chips are designed to efficiently handle generative and agentic AI workloads at scale, while strengthening Arm's position in the evolving "AI cloud" era. While this move opens up new growth opportunities, it also has the potential to spark competition with Arm’s past partners, demonstrating the increasingly competitive and strategic future of the AI chip market.
Read more: SLB and Nvidia Partners in Developing AI Infrastructure for Energy Sector