AI data center costs face new uncertainty! Stricter regulations may require Oracle to provide over $7 billion in guarantees

Credit rating downgrade combined with tightened US regulatory environment is pushing Oracle's AI infrastructure construction costs even higher.

Oracle has encountered a major funding obstacle in its large-scale data center project in Wisconsin, USA. The state's power regulator upheld its credit collateral requirements for the local utility, meaning Oracle may need to provide over $7 billion in guarantees—a ruling that further intensifies the tech giant’s financial pressure in building out AI infrastructure.

The Public Service Commission of Wisconsin told the Financial Times on Monday, July 20, that it had "declined to take action on the application," thereby upholding its collateral requirements for utility company We Energies. The rules require Oracle to provide about $7 billion in guarantees in the form of cash or letters of credit, with additional annual costs exceeding $100 million. Oracle stated that it hopes the Commission will reconsider its stance after evaluating the employment and economic growth opportunities brought by the $15 billion project, and reaffirmed its "commitment to provide the necessary financial assurances to ensure Wisconsin ratepayers bear no risk."

This data center is a critical project for Oracle to fulfill its $300 billion computing contract with OpenAI, with a total investment of $15 billion.

The ruling is an initial setback for Oracle. Meanwhile, S&P this month downgraded Oracle’s credit rating to BBB-, just one notch above junk, citing unclear profitability prospects due to heavy AI investments. The deterioration in credit rating could also impact the tens of billions in construction debt underwritten by Wall Street banks for Oracle’s data center expansion, further increasing overall financing risk.

Facing regulatory hurdles, Oracle filed a lawsuit in county court last month, seeking to overturn the regulation and allow an exemption for We Energies. In its filing, Oracle argued that the rule "threatens to impose heavy financing costs on Oracle and may hinder future investments in Wisconsin."

The regulator’s refusal to reconsider means Oracle’s legal battle will continue. Analysts point out that this case reflects an increasing caution among state regulators that large-scale AI infrastructure projects could become "stranded assets"—especially when these projects lack sufficient financial support.

This month’s downgrade by S&P places Oracle at BBB-, just one step from speculative grade. The ratings agency cited growing uncertainty about profitability due to Oracle’s major investments in AI as the main reason for the downgrade.

The impact of the credit rating drop goes beyond that. According to a recent report by the Financial Times, Wall Street banks that have extended tens of billions in construction loans for Oracle’s data centers are now seeking new ways to shift the lending risk associated with massive data center borrowing. Further downgrades would inevitably make such financing arrangements more difficult and expensive.

The Wisconsin case is not an isolated event, but part of a broader regulatory tightening trend across the US. Amid rising concerns that power users will ultimately shoulder the costs of data center-related grid upgrades, 24 states have now adopted "large user tariffs" that set terms and pricing for data centers and other major industrial users.

Such provisions typically require data centers to commit to minimum contract terms, pay exit fees, and provide collateral guarantees. Large tech firms and manufacturers have objected, arguing that these terms are discriminatory and impose excessive costs.

Oracle’s experience in Wisconsin signals that regulatory negotiations with local authorities will be a significant variable in the ongoing race to expand AI data center capacity.