Chip designer Arm Holdings plc has taken a major step toward shaping the future of AI-driven data centers by joining the Open Compute Project (OCP). The move comes as global energy demands skyrocket due to the rapid expansion of artificial intelligence workloads.
Arm’s goal is to help the industry develop the next generation of purpose-built silicon — chips designed specifically for AI data centers that prioritize power efficiency and integration across all system components. Rather than relying on the traditional model of separate CPUs, GPUs, and networking hardware, the next wave of AI infrastructure will feature converged architectures. These systems will integrate multiple chiplets within the same package using 2.5D and 3D stacking technologies, dramatically improving performance and density.
To support this shift, Arm is contributing its new Foundation Chiplet System Architecture (FCSA) specification to the OCP. FCSA builds upon Arm’s existing Chiplet System Architecture (CSA) work, providing a vendor-neutral framework that enables companies to co-design processors, memory, and networking components more efficiently.
According to Mohamed Awad, Senior Vice President and General Manager of Arm’s Infrastructure Business, the biggest challenge in building AI data centers today isn’t just cost — it’s energy consumption.
“For anybody building a data center, the specific challenge that they’re running into is not really about the dollars associated with building, it’s about keeping up with the power demand,” Awad said.
As AI workloads continue to grow, performance per watt has become the key metric. Data center operators are now moving away from off-the-shelf solutions and taking a hands-on role in custom silicon design to create highly optimized systems. The goal is to bring acceleration closer to the compute, resulting in higher efficiency and lower energy use.
Arm’s participation in the OCP aligns with the organization’s broader mission to openly share data center technologies — from server racks and networking to silicon-level designs. The announcement coincides with the 2025 OCP Global Summit in San Jose, California.
In addition to joining OCP, Arm is also part of the new Ethernet for Scale-Up Networking (ESUN) initiative, launched alongside major industry players such as AMD, Arista, Broadcom, Cisco, HPE, Meta, Microsoft, Nvidia, and others. The ESUN collaboration aims to advance Ethernet networking technology to better handle the massive data traffic generated by AI-scale infrastructure.
As the AI revolution continues to push the limits of data center performance and energy efficiency, Arm’s open collaboration with the OCP and its partners could play a pivotal role in defining the silicon foundations of the next generation of computing.
