OpenAI has signed a 20-year lease for an 8-gigawatt data center in Ohio, with Nvidia guaranteeing up to $105 billion for the residual value of the facilities and becoming the exclusive chip supplier. The deal underscores the massive scale of AI infrastructure investments, as nine tech companies hold around $3 trillion in AI commitments that are not on balance sheets.
SynthePulse Insight · AI deep reading
Version 1 · 1 source
OpenAI signs a 20-year lease with SoftBank's SB Energy for approximately 8 gigawatts of IT capacity in Ohio, with Nvidia as the exclusive chip supplier backed by a guarantee of up to $105 billion. The deal highlights the scale and risk of AI infrastructure investment and the swelling of off-balance-sheet commitments by industry giants.
OpenAI has signed a 20-year lease with SB Energy, a SoftBank subsidiary, for approximately 8 gigawatts of IT capacity at the PORTS-Pike site in Ohio. According to the Wall Street Journal, the total project capacity, including cooling and infrastructure, is 10 gigawatts. The site is partly located at a former U.S. Department of Energy uranium enrichment facility, with power coming from a 9.2-gigawatt natural gas plant owned by the U.S. government and financed by Japan under a trade agreement.
Nvidia's guarantee is not a direct payment of rent but covers the residual value of the first 4.25-gigawatt IT capacity facility. If OpenAI exits, SB Energy must first seek a replacement tenant, then attempt to sell the facility, with Nvidia covering only the shortfall, capped at $105 billion. In return, Nvidia becomes the exclusive chip supplier for that portion and invests $1.5 billion in SB Energy.
OpenAI says it will only pay for completed capacity, with the first 800 megawatts expected to come online in 2028. If Nvidia exercises its option for the remaining 3.75 gigawatts, the entire project would grow to about 16 gigawatts, valued at approximately $600 billion.
Jensen Huang introduced the new acronym 'LPS' (land, power, shell) when announcing the deal, arguing these have replaced chips and networking equipment as the primary bottleneck for AI construction. AI labs are growing faster than their balance sheets can support long-term infrastructure contracts, so Nvidia steps in to provide financial support.
Huang estimates each generation of systems will require about 1.5 million GPUs, corresponding to $150 billion to $200 billion in revenue. He estimates OpenAI's Nvidia compute commitments across all sites by 2030 are about 12 gigawatts. If the option is exercised, the Ohio project would grow to 16 gigawatts, valued at about $600 billion.
This strategy transforms Nvidia from a chip supplier to an infrastructure guarantor, but it also assumes significant risk. The guarantee cap was reduced from an initial consideration of $250 billion to $105 billion, reflecting investor concerns about risk.
A Wall Street Journal analysis shows that nine tech companies, including Alphabet, Meta, Microsoft, and Nvidia, hold approximately $3 trillion in AI-related off-balance-sheet obligations. Leases are only recorded when payments begin, and purchase commitments are only recorded upon delivery. Leases not yet started total $1.2 trillion, four times the year-ago period.
Alphabet's purchase commitments jumped from $332 billion to $811 billion in three months. These contracts are nearly impossible to cancel, and both Alphabet and Amazon recently reported negative free cash flow. Morgan Stanley analysts warn that investors are already struggling to assess the actual debt levels of these companies.
The scale of these off-balance-sheet commitments highlights the financial risks of AI infrastructure investment, but investors may underestimate their impact.
Nvidia's guarantee cap is $105 billion, lower than the previously rumored $120 billion, and covers only asset value, not the entire lease. Initially, Nvidia considered a guarantee of about $250 billion but reduced it due to a 5% stock drop and investor pressure.
The project timeline is uncertain: the first 800 megawatts are expected in 2028, but the timing for the remainder is unclear. OpenAI's commitments by 2030 are about 12 gigawatts, but actual execution may be affected by market conditions.
Analysts warn that off-balance-sheet commitments make it difficult for investors to assess debt levels, but the specific impact remains unclear.
This report is based on secondhand reporting from THE DECODER, which sourced information from the Wall Street Journal and public statements from OpenAI and Nvidia. Key figures (such as the $105 billion guarantee and 8 gigawatts capacity) come from the reporting but have not been independently confirmed. Some details (such as the initial $250 billion guarantee) are from the Wall Street Journal and may not be officially confirmed.
This deal marks a new phase in AI infrastructure investment, with Nvidia transitioning from chip supplier to financial guarantor, but the swelling of off-balance-sheet commitments raises concerns about debt transparency among tech companies.
Primary report
Primary source