NVIDIA, OpenAI, and SB Energy Partner on 10-Gigawatt AI Data Center Campus in OhioNVIDIA and OpenAI have expanded their infrastructure partnership by joining forces with SB Energy to construct a massive AI data center hub at the PORTS-Pike Technology Campus in Ohio, United States. Under the tripartite agreement, the facility will be fully equipped with NVIDIA’s DSX AI architecture, with OpenAI serving as the primary anchor tenant under a 20-year long-term lease, while SB Energy handles construction and facility operations.
To finance the project, NVIDIA will extend a line of credit to SB Energy covering land acquisition and structural development. The campus is designed for an initial compute capacity of 4.25 IT Gigawatts (GW) measuring dedicated power for processing hardware with provisions to expand by an additional 3.75 IT GW, bringing total compute capacity to 8 IT GW. Including cooling and operational overhead, total campus power demand will reach approximately 10 Gigawatts. Additionally, NVIDIA will make a $1.5 billion equity investment in SB Energy, joining existing investors OpenAI and SoftBank.
Addressing local environmental and utility concerns surrounding large-scale data center developments in the U.S., the companies emphasized several sustainability measures:
Grid Independence: Power supply lines and electrical infrastructure will be built entirely from scratch, ensuring zero impact on local municipal power grids.
Closed-Loop Cooling: The facility will utilize a closed-loop, air-cooled water recirculation system, maintaining water consumption levels comparable to standard commercial office buildings.
Infrastructure Investment: All three firms will co-invest in upgrading local municipal water utility infrastructure.
Initial operations are scheduled to begin in 2028 with an early phase of 800 Megawatts (MW) running on existing power infrastructure before transitioning to the custom-built grid upon completion.
The difference between IT power capacity and total data center power, measured in "IT GW" units, separates the electricity used only by GPU clusters, high-speed networks, and storage systems. The remaining up to 10 GW is used for power loss in distribution and environmental controls, demonstrating a highly optimized Power Use (PUE) ratio for data centers of this size.
Data center development across the United States is facing increasing public resistance regarding power grid strain and local water shortages. NVIDIA and OpenAI are creating a new enterprise model to circumvent municipal zoning hurdles and regulatory delays by explicitly committing to private power lines and the use of closed-loop dry-cooling technologies.
NVIDIA's $1.5 billion direct investment, coupled with a loan guarantee, to SB Energy demonstrates a shift from hardware manufacturers to primary infrastructure financiers, supported by OpenAI's 20-year long-term commitment. This model allows developers like SB Energy to secure massive private financing while guaranteeing NVIDIA a multi-year roadmap for its enterprise hardware architecture.
Source: NVIDIA
NVIDIA, OpenAI, and SB Energy Partner on 10-Gigawatt AI Data Center Campus in OhioNVIDIA and OpenAI have expanded their infrastructure partnership by joining forces with SB Energy to construct a massive AI data center hub at the PORTS-Pike Technology Campus in Ohio, United States. Under the tripartite agreement, the facility will be fully equipped with NVIDIA’s DSX AI architecture, with OpenAI serving as the primary anchor tenant under a 20-year long-term lease, while SB Energy handles construction and facility operations.
To finance the project, NVIDIA will extend a line of credit to SB Energy covering land acquisition and structural development. The campus is designed for an initial compute capacity of 4.25 IT Gigawatts (GW) measuring dedicated power for processing hardware with provisions to expand by an additional 3.75 IT GW, bringing total compute capacity to 8 IT GW. Including cooling and operational overhead, total campus power demand will reach approximately 10 Gigawatts. Additionally, NVIDIA will make a $1.5 billion equity investment in SB Energy, joining existing investors OpenAI and SoftBank.
Addressing local environmental and utility concerns surrounding large-scale data center developments in the U.S., the companies emphasized several sustainability measures:
Grid Independence: Power supply lines and electrical infrastructure will be built entirely from scratch, ensuring zero impact on local municipal power grids.
Closed-Loop Cooling: The facility will utilize a closed-loop, air-cooled water recirculation system, maintaining water consumption levels comparable to standard commercial office buildings.
Infrastructure Investment: All three firms will co-invest in upgrading local municipal water utility infrastructure.
Initial operations are scheduled to begin in 2028 with an early phase of 800 Megawatts (MW) running on existing power infrastructure before transitioning to the custom-built grid upon completion.
The difference between IT power capacity and total data center power, measured in "IT GW" units, separates the electricity used only by GPU clusters, high-speed networks, and storage systems. The remaining up to 10 GW is used for power loss in distribution and environmental controls, demonstrating a highly optimized Power Use (PUE) ratio for data centers of this size.
Data center development across the United States is facing increasing public resistance regarding power grid strain and local water shortages. NVIDIA and OpenAI are creating a new enterprise model to circumvent municipal zoning hurdles and regulatory delays by explicitly committing to private power lines and the use of closed-loop dry-cooling technologies.
NVIDIA's $1.5 billion direct investment, coupled with a loan guarantee, to SB Energy demonstrates a shift from hardware manufacturers to primary infrastructure financiers, supported by OpenAI's 20-year long-term commitment. This model allows developers like SB Energy to secure massive private financing while guaranteeing NVIDIA a multi-year roadmap for its enterprise hardware architecture.
Source: NVIDIA
Comments
Post a Comment