The scale of the deal

OpenAI has signed a lease for the largest data center project announced to date. The contract with SoftBank subsidiary SB Energy for the "PORTS-Pike" campus in Ohio runs 20 years and gives OpenAI around 8 gigawatts of IT capacity. The Wall Street Journal puts the project's gross total at 10 gigawatts once cooling and infrastructure are factored in.

Nvidia is backing the project with up to $105 billion and becomes the sole supplier of compute infrastructure at the site. The company also said it will invest $1.5 billion in SB Energy. According to documents filed with the SEC, the facility could scale from an initial 4.25 gigawatts to 8 gigawatts.

Land and power

The site's history is notable: the land is owned by the US Department of Energy and previously enriched uranium for the US nuclear arsenal and for US Navy submarines.

On the power side the picture is as follows:

  • A 9.2-gigawatt natural gas power plant will be built on the site.
  • The plant is expected to cost $33 billion.
  • According to BloombergNEF, the cost of building natural gas plants has risen 66 percent in the last two years.
  • The plant is owned by the US government, with part of the financing provided by Japan under a trade agreement.

TechCrunch also reports a warning: by the time SB Energy's plant and others are completed, they will be competing for natural gas with export markets. That confluence could triple natural gas prices in some parts of the country.

Huang's new acronym

Nvidia CEO Jensen Huang announced the deal with a new acronym: LPS, standing for land, power and shell. According to Huang, these basics have replaced chips and networking gear as the real bottleneck in the AI buildout.

The phrase captures where the constraint in the sector has moved. Chip supply was the only topic for a long time; now finding a gigawatt of electricity, finding land that can draw it, and getting the building up on time have become more binding.

The part that does not appear on the balance sheet

The financial side of commitments at this scale is a separate debate. As the Wall Street Journal reports, major tech companies hold around $3 trillion in AI commitments off their balance sheets. Analysts warn that investors can barely gauge these companies' actual debt levels anymore.

The mechanism works like this: labs such as OpenAI are growing faster than their balance sheets can support long-term, so capacity is secured through long leases and supplier guarantees rather than bought outright. The obligation is real but does not show up in the accounts. Nvidia occupying three separate roles in the same project — supplier, lender and investor — is a concrete example of that entanglement.