Open Data Center for AI
Open Data Centers for AI (also "Open DC for AI") is an OCP Strategic Initiative launched in response to the late-2025 open letter in which most hyperscalers called for more open collaboration on fungible data center facilities, grid interconnect, open AI hardware, and systems management. With rack power densities rising toward 1 MW racks and GW-scale facilities and multi-GW campuses on the horizon, the initiative unites the community to deliver standardized, efficient, and flexible solutions, coordinated within the Open Systems for AI umbrella. Its five core priorities are open data center reference designs, energy and grid solutions, open and standard SLAs, telemetry and management solutions, and power estimation methodologies.
Get involved
Project Leads
About This Workstream
The rapid evolution of data center infrastructure driven by AI is causing unprecedented disruption, marked by an order-of-magnitude increase in rack power densities, with roadmaps pointing to 1 MW racks in the next few years. This shift necessitates the development of GW-scale computing facilities and multi-GW campuses, demanding new technologies and collaborative approaches.
In late 2025, most hyperscalers signed an open letter published by the OCP Community to encourage more open collaboration for more fungible data center physical facilities and grid interconnect, more open IT hardware for AI, and systems management specifications. In response, OCP has launched the Open Data Center for AI (also “Open DC for AI”) initiative to address the myriad challenges associated with delivering these massive AI data centers at scale, uniting the community and project teams to develop standardized, efficient, and flexible solutions. This initiative is coordinated within the overarching Open Systems for AI umbrella initiative to drive continued OCP Community-driven open contributions that have served the data center industry for the last 15 years.
Scope
The Open Data Center for AI initiative is focused on solving the most pressing challenges facing the deployment of at-scale AI data center facility infrastructure and operational technology (OT). The leadership team has identified five core priorities that define the scope of work:
- Open Data Center Reference Designs : develop fungible, standard facility-level designs that incorporate advanced power (e.g., higher voltage levels of LVDC) and cooling (e.g., liquid cooling CDUs), to ensure flexibility, improve supply chain integrity, minimize deployment delays (“time to first token”), and unlock broader investment
- Energy and Grid Solutions : address the primary bottleneck of securing power: flexibility to utilize latent grid capacity, standard technical specifications (e.g., Low Voltage Ride Through), and best practices for on-site generation and energy storage (e.g., BESS, microgrids)
- Open and Standard Service Level Agreements (Open SLA) : standardize and improve SLAs for third-party operators (colo, neocloud) given the deeper interaction between IT equipment and advanced facility cooling and power systems, optimizing energy efficiency, cost, water use, and capacity outcomes
- Telemetry and Management Solutions : enhance standards and common approaches for monitoring IT and facility state to enable AI-driven management: harmonizing IT and facility (OT) operations for power and water consumption, predictive failure analysis, and reliability, including cooling and power equipment and distribution systems
- Power Estimation Methodologies : drive continuous improvement and best practices for estimating equipped power needs for AI data centers, minimizing stranded power and addressing the power variability introduced by new architectures, workloads, cooling, and power delivery systems