Oracle Says Wisconsin Power Rules Could Require More Than $7 Billion in Security
Oracle says Wisconsin's approved data-center power tariffs could require it to arrange more than $7 billion in financial security for the Lighthouse campus, with annual letter-of-credit costs potentially above $100 million. The estimate is not a bill due today: it is Oracle's projection under credit-support rules meant to shield other utility customers if the project cannot cover infrastructure costs.
Oracle has told Wisconsin utility regulators that its Lighthouse data-center campus could eventually require more than $7 billion in financial security under approved electricity tariffs. The company estimates that arranging a letter of credit at that scale could cost more than $100 million a year.
Those figures come from Oracle's June 10 testimony in Public Service Commission docket 6630-TE-113. They are company projections, not a $7 billion invoice, fine, or cash payment that regulators have ordered Oracle to make now.
What the rules require
The commission's May decision set financial-support conditions for very large customers served by Wisconsin Electric, which operates as We Energies. To avoid posting security, a customer must meet an A- rating from S&P and an A3 rating from Moody's, plus a tangible-net-worth or liquidity test.
Oracle said its parent was rated BBB by S&P and Baa2 by Moody's when it filed the testimony, while the Oracle subsidiary taking service for the campus was unrated. Oracle therefore expects the project to fall outside the exemption.
For a non-exempt customer, the required support can cover the remaining book value of subscribed generation resources, dedicated distribution facilities, and 1.3 times two years of specified tariff charges. The approved forms include a cash deposit or letter of credit; a parent guarantee is available only when the guarantor meets the financial tests.
What Oracle wants changed
Oracle, We Energies, Vantage Data Centers, and Cloverleaf Infrastructure asked the commission to restore some utility discretion to waive or modify the requirements, subject to regulatory review. For the Lighthouse campus, Oracle proposed covering about 10% of the required security with a letter of credit—roughly $700 million under its estimate—and the remaining 90% with a parent guarantee.
The commission framed the tariff changes as customer protections. Its April decision also extended the minimum service term to 15 years and required very large customers to cover the full cost of generation resources assigned to them, reducing the risk that infrastructure costs shift to households and other businesses.
Why the dispute matters
The disagreement shows that AI infrastructure economics extend beyond servers and construction. Utility credit rules can create large contingent financing needs before a data center reaches full load. For operators and capacity planners, the relevant cost model therefore includes not only power prices and capital equipment, but also the credit support needed to secure long-term electricity service.
Key Points
- 1Oracle estimates the approved tariff could require more than $7 billion in security and more than $100 million in annual letter-of-credit costs; those are projections, not a current bill.
- 2The exemption requires A-/A3 credit ratings plus a tangible-net-worth or liquidity test, and Oracle says it does not meet those combined conditions.
- 3Oracle proposed a roughly $700 million letter of credit for 10% of the requirement and a parent guarantee for the remaining 90%, with any change still subject to regulatory review.
Scoring Rationale
The dispute could materially affect financing for a large AI data-center campus and shows how utility credit rules shape infrastructure economics. Its impact is substantial but remains project- and jurisdiction-specific, and the $7 billion figure is Oracle's projection rather than an imposed payment.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
