OpenAI is in advanced discussions to anchor one of the most ambitious artificial intelligence infrastructure projects ever conceived, according to reporting from The Information. The ChatGPT maker is looking to lease a proposed 10-gigawatt data center campus on Department of Energy land in southern Ohio. The facility would be developed by SoftBank's SB Energy, with OpenAI taking control of the equipment under a 20-year lease structure that kicks in once operations begin, with the first phase expected to come online in 2028.

This deal represents a fundamental shift in how AI companies are approaching infrastructure. Rather than building data centers themselves, OpenAI is locking in long-term capacity through a lease arrangement that transfers operational risk to SoftBank while securing guaranteed access to computing power for two decades. The scale is staggering: 10 gigawatts of power capacity is enough to run a small city, and the 20-year commitment reflects OpenAI's confidence that demand for AI computing will only grow.

Why Is This Deal So Significant for AI Development?

The Ohio data center project reveals the true cost of building frontier AI models like GPT-5 and future versions of the o1 and o3 reasoning models. Training and running these systems requires unprecedented amounts of electricity and computing hardware. By securing a dedicated 10-gigawatt facility, OpenAI is essentially betting that the next generation of AI models will demand even more computational resources than today's systems.

Nvidia, the dominant supplier of AI chips, is playing a critical role in this arrangement. The chip giant is expected to supply the hardware for the data center and provide a financial guarantee covering both OpenAI's lease obligations and SB Energy's financing. This arrangement adds another layer to Nvidia's fingerprints across the AI infrastructure buildout that is reshaping the U.S. energy and real estate landscape.

The timing matters too. With the first phase expected to come online in 2028, OpenAI is planning for a world where AI models are even more capable and computationally intensive than they are today. The company's existing partnership with Oracle, which includes a reported $300 billion, five-year deal, already demonstrates the scale of investment flowing into AI infrastructure.

How Are AI Companies Securing Computing Power for the Future?

• Long-Term Leasing Agreements: Rather than owning data centers outright, companies like OpenAI are negotiating multi-decade leases that lock in capacity and transfer operational complexity to specialized developers like SoftBank's SB Energy.
• Chip Supplier Guarantees: Nvidia is providing financial backing for these infrastructure deals, effectively betting on the continued growth of AI demand and securing its position as the essential hardware provider for the industry.
• Government Land Partnerships: The use of Department of Energy land in Ohio signals a shift toward public-private collaboration, where government property and resources support private AI infrastructure development.
• Phased Deployment Strategies: Rather than building entire data centers at once, companies are planning phased rollouts, with OpenAI's Ohio facility's first phase coming online in 2028, allowing for flexibility as technology and demand evolve.

The broader context is crucial here. AI companies are not just building infrastructure for today's models; they are building for a future where models like GPT-5 and beyond will require exponentially more computing power. The 10-gigawatt Ohio facility is not an outlier; it reflects an industry-wide recognition that AI infrastructure is becoming the primary constraint on model development.

This infrastructure race also has implications for energy markets and regional development. A 10-gigawatt data center requires reliable, abundant electricity, which is why OpenAI and other AI companies are targeting regions with access to hydroelectric power, natural gas, or nuclear energy. The Ohio location, with its proximity to energy resources and existing industrial infrastructure, makes strategic sense for a facility of this scale.

For enterprises and developers using OpenAI's services, this infrastructure investment translates to more reliable access to ChatGPT, Sora, DALL-E, and future models. The long-term lease structure also suggests that OpenAI is confident in its ability to monetize AI services at a scale that justifies a $300 billion commitment to computing infrastructure over the next five years.