llms.blog
AI

OpenAI Signs 20-Year Lease for Multi-Gigawatt Ohio AI Campus with Nvidia Backstop

OpenAI has signed a 20-year lease agreement to anchor a multi-gigawatt AI data center campus in Pike County, Ohio, developed by SoftBank subsidiary SB Energy. Under the tripartite agreement, Nvidia will serve as the exclusive compute infrastructure provider and supply financial guarantees covering the site's land, power, and shell development. The facility, designated the PORTS-Pike Technology Campus, is designed to support an initial 4.25 gigawatts of information technology load (IT-GW) dedica

axc
axc
2 min
OpenAI Signs 20-Year Lease for Multi-Gigawatt Ohio AI Campus with Nvidia Backstop

OpenAI has signed a 20-year lease agreement to anchor a multi-gigawatt AI data center campus in Pike County, Ohio, developed by SoftBank subsidiary SB Energy. Under the tripartite agreement, Nvidia will serve as the exclusive compute infrastructure provider and supply financial guarantees covering the site's land, power, and shell development.

The facility, designated the PORTS-Pike Technology Campus, is designed to support an initial 4.25 gigawatts of information technology load (IT-GW) dedicated to OpenAI workloads. Nvidia retains an option to secure an additional 3.75 IT-GW at the campus, bringing total planned compute capacity up to 8 IT-GW across the buildout.

PORTS-Pike campus modular power and compute architecture

Financing Structure and Nvidia Backstop

SB Energy will construct, own, and operate the physical data center infrastructure. To enable long-term debt financing for the multi-gigawatt campus, Nvidia is providing credit support and lease backstop commitments that could reach up to $105 billion over the lifespan of the facility.

Alongside the credit facilities, Nvidia announced a direct $1.5 billion equity investment in SB Energy to accelerate infrastructure deployment and grid interconnection. The project also establishes an initial $80 million community benefit fund in southern Ohio to support regional energy transition initiatives and workforce development.

The data center deployment will be built around Nvidia's DSX AI factory architecture, integrating high-density GPU clusters, custom CPUs, and NVLink optical switching fabrics.

Power Demands and Grid Realities

The 20-year commitment reflects the escalating difficulty frontier AI developers face in securing dedicated electrical grid capacity. Securing gigawatt-scale power allocations has become a primary gating factor for frontier model training, prompting labs to contract directly with independent power producers and infrastructure developers years in advance.

The Ohio project represents one of the largest single-site computing commitments signed to date, surpassing typical hyperscale campus footprints by an order of magnitude.

Sources

AIOpenAINvidiaInfrastructureData Centers
ShareX (Twitter)LinkedIn

Written by

axc

More to read

  • The Curse of Multilinguality in Large Language Models: Capacity Dilution, Tokenizer Fertility, and Representation Interference

    The Curse of Multilinguality in Large Language Models: Capacity Dilution, Tokenizer Fertility, and Representation Interference Training a single transformer foundation model to process dozens or hundreds of languages is one of the central goals of modern natural language processing. In theory, massive multilingual pre-training unlocks positive cross-lingual transfer: low-resource languages gain syntactic, factual, and reasoning capabilities from the rich supervision available in high-resource l

    1 min

  • LLM Load Shedding and Overload Control in Production: Adaptive Admission, Preemption Economics, and Graceful Degradation

    LLM Load Shedding and Overload Control in Production: Adaptive Admission, Preemption Economics, and Graceful Degradation Standard web services rely on well-established overload protection patterns: reverse proxies monitor CPU utilization, memory thresholds, or static queue depths and reject excess HTTP requests with 429 Too Many Requests or 503 Service Unavailable status codes. When applied naively to Large Language Model (LLM) serving infrastructure, these conventional heuristics fail catastro

    1 min

  • Reward Model Overoptimization in Large Language Models: How Goodhart's Law, Proxy Exploitation, and KL Drift Degrade Alignment

    Post-training alignment of large language models relies on optimizing a policy toward objectives defined by human intent and preferences. Because querying human evaluators during every step of continuous reinforcement learning or high-throughput rejection sampling is computationally and logistically infeasible, alignment workflows construct a parameterised proxy reward model. Trained on pairwise preference datasets through formulations such as the Bradley-Terry model, this proxy acts as a surrog

    1 min