Oracle lays off 21,000 to fund OpenAI deal; $7B power grid guarantee now blocks project

Oracle became a primary provider of AI computing power after signing a $300 billion（約48兆円） contract with OpenAI, one of the largest deals in the AI infrastructure race. However, constructing the data centers required to fulfill the agreement imposed severe financial strain. To fund the rapid expansion, Oracle enacted a sweeping wave of layoffs driven in part by internal adoption of AI technologies. By the end of fiscal year 2026, the company's workforce had been cut by about 21,000 employees—a decline of roughly 13%—reducing headcount from 162,000 to 141,000 workers.

The financial pressure appears to have weakened Oracle's credit profile. S&P downgraded Oracle's credit rating from BBB to BBB-, citing heavy AI spending and uncertainty regarding the ability to generate profits from it. This downgrade triggered a regulatory constraint in Wisconsin, where Oracle is planning a massive, nearly one-gigawatt data center in Port Washington to supply computing power for the OpenAI contract. Under local regulations governing large electricity consumers, operators whose S&P credit rating falls below A- are required to provide collateral covering the electrical infrastructure built for them. The Public Service Commission of Wisconsin refused to ease these financial collateral requirements, which are designed to ensure that private electricity consumers are not forced to absorb costs in the event of facility failure or closure.

As a result, Oracle now faces a requirement to provide cash collateral or a letter of credit totaling over $7 billion（約1.1兆円） just to connect the data center to the power grid. The ongoing maintenance of this setup will cost the company more than $100 million（約160億円） annually. Oracle petitioned the court against the requirement, arguing that these financing costs would deter future investment in the state and emphasizing its commitment to the project, which carries an estimated value of about $15 billion（約2.4兆円）. Regulators remained firm, clarifying that existing customers should not subsidize data centers. The dispute reflects a broader pattern: at least 24 US states have already approved special rates, minimum conditions, exit penalties, and collateral requirements for heavy electricity consumers, indicating that such regulatory scrutiny is spreading across the country as AI infrastructure demand grows.

The Oracle story illustrates the extreme financial pressure facing technology companies in the AI infrastructure race. With Amazon, Microsoft, Alphabet, and Meta collectively expected to spend about $600 billion（約96兆円） on AI infrastructure during 2026, companies are making massive bets on the assumption that AI investments will eventually generate profits. Oracle took one of the largest single bets by signing a $300 billion（約48兆円） contract with OpenAI, but the capital requirements to build the required data centers created a severe cash squeeze that forced drastic workforce cuts.

The credit rating downgrade from BBB to BBB- appears to have triggered a cascade of financial consequences. Under Wisconsin's existing regulations for large electricity consumers, companies with ratings below A- must provide collateral to protect the public from absorbing infrastructure costs if the facility fails or closes. Oracle's downgrade activated this rule, resulting in a requirement for over $7 billion（約1.1兆円） in collateral plus ongoing maintenance costs exceeding $100 million（約160億円） annually. Oracle has challenged the requirement in court, arguing it will deter investment, but regulators have held firm on the principle that existing customers should not subsidize new data centers.

The tension between Oracle and Wisconsin is not isolated. At least 24 US states have already approved special rates, minimum conditions, exit penalties, and collateral requirements for heavy electricity consumers, suggesting that as companies continue to build power-intensive AI infrastructure, these regulatory and financial barriers may become increasingly common. The $7 billion（約1.1兆円） collateral demand places the $15 billion（約2.4兆円） project in genuine jeopardy and reveals a structural problem: the profitability of massive AI spending remains unproven, yet the financial and operational costs of pursuing it are already substantial and immediate.

No comments yet. Be the first to share your thoughts!