Agents AI

Funding
ai

Nvidia Backs OpenAI's Ohio Data Center With Up to $105 Billion in Financing

Nvidia will guarantee up to $105 billion in credit for a new SB Energy-built data center in Pike County, Ohio, that OpenAI will lease exclusively, deepening the circular financing ties between the chipmaker and its largest AI customer.

AgentsAI NewsroomAugust 17, 20263 min read

Nvidia said August 17 it will guarantee up to $105 billion in credit to support a new AI data center campus in Pike County, Ohio, that OpenAI has agreed to lease exclusively for 20 years — one of the largest single infrastructure financing commitments of the AI buildout so far.

The deal structure

The campus, called PORTS-Pike Technology Campus, will be built and operated by SoftBank-backed SB Energy rather than Nvidia or OpenAI directly. Nvidia's role is to guarantee financing covering land, power and shell construction for an initial 4.25 gigawatts of capacity, with an option to extend the guarantee to a further 3.75 gigawatts, bringing the site's planned total to 8 gigawatts. Nvidia is separately investing $1.5 billion directly in SB Energy. In exchange, Nvidia becomes the campus's exclusive AI compute infrastructure provider, meaning the chips running inside it will be Nvidia hardware. OpenAI will lease the finished capacity from SB Energy over a 20-year term and begin paying once each phase comes online, with the first phase targeted to begin operating in 2028.

SB Energy and SoftBank also committed to build at least 10 gigawatts of new power generation and $4.2 billion of regional grid upgrades in partnership with AEP Ohio, the local utility subsidiary of American Electric Power, to supply the campus. OpenAI said the project is expected to support roughly 35,000 construction jobs through 2032 and around 2,500 permanent positions once operational.

Deepening a circular financing pattern

The Ohio deal is Nvidia's latest move to underwrite the infrastructure being built around its own chips — a strategy that guarantees demand for Nvidia hardware but has drawn scrutiny for creating financing loops in which the chipmaker effectively backs the buildout of its own biggest customers, OpenAI chief among them. Nvidia has now committed tens of billions of dollars in direct investment and credit support across multiple OpenAI-linked data center projects this year, on top of its existing compute-supply agreements with the company.

Why it matters

At up to $105 billion, the guarantee is among the largest financing figures disclosed for a single AI data center site to date, underscoring how much capital frontier AI labs now need simply to secure power and compute at scale — commitments increasingly structured through complex financing guarantees and long-term leases rather than straightforward capital purchases. For OpenAI, locking in 8 gigawatts of dedicated capacity gives it a multi-year runway of compute as it races Google, Anthropic and others to train and serve ever-larger models, while for Nvidia the deal reinforces its position as the indispensable — and increasingly financially entangled — infrastructure layer beneath the entire industry.

AI-assisted reporting, overseen by the AgentsAI team. Spotted an error? Let us know.

More ai news

Pricing
ai·

DeepSeek Finalizes Steep API Price Hikes, Ending Its Flat-Rate Era

DeepSeek confirmed a new peak/off-peak pricing structure for its V4 Pro and V4 Flash models effective August 16, raising some output-token rates by more than 1,100% as the low-cost Chinese lab moves away from the flat pricing that fueled its rise.

Update
ai·

China's Z.ai Ships GLM-5.3, Claiming Coding and Cyber Gains Without a Bigger Base Model

Beijing-based Z.ai released GLM-5.3 on August 14, reusing its ~700B-parameter GLM-5.2 base model but claiming a 50% jump on internal coding benchmarks and a leading CyberGym score, positioning it against Anthropic and OpenAI on coding without training a larger model.

Research
ai·

Researchers Show Encrypted Reasoning Traces Can Be Stolen Across OpenAI, Anthropic and Google APIs

A paper published August 10 by researchers from the ELLIS Institute Tübingen and the Max Planck Institute found that encrypted chain-of-thought blocks returned by OpenAI, Anthropic and Google reasoning APIs are interchangeable across sessions and models, letting a weaker model decode and leak a stronger model's hidden reasoning in plaintext.

Update
ai·

DeepSeek Ships V4-Pro-0813 as Its Flagship Model Leaves Preview, Doubling Down on Agent Tasks

DeepSeek officially released DeepSeek-V4-Pro-0813 on August 13, moving its flagship model out of preview with sharply improved agent and coding benchmarks, a 1-million-token context window, and new Responses API and Codex-style tool support.