Five major tech giants have committed over $1 trillion to AI data center leasing, sparking concerns over future capacity risks.

According to Reuters, Microsoft, Meta, Oracle Cloud, Amazon, and Alphabet have committed approximately $1.09 trillion in future lease payments, primarily for building data centers to support artificial intelligence (AI) development. Data shows this figure is nearly four times the $285 billion in total confirmed lease liabilities currently held by the five firms. Due to accounting rules, data center leases that have been signed but not yet put into operation are not typically recorded as balance sheet liabilities immediately; instead, they are disclosed as future payment obligations. Reuters notes that a large share of current AI infrastructure investment spending has been locked in advance. If future AI computing demand continues to grow, these data centers will support cloud business expansion; however, if demand falls short of expectations, tech companies may face risks of long-term high rental payments and idle capacity. Among them, Oracle’s exposure is most prominent: the company disclosed its unused lease commitments reach $260 billion, nearly seven times its confirmed lease liabilities of $37.89 billion. These leases are mainly for data center construction, expected to launch between fiscal years 2027 and 2029, with typical lease terms of 15 to 19 years. Microsoft reported the largest unused lease commitments at $32.91 billion; Meta disclosed $27.899 billion, and further signed a $68 billion data center lease agreement in July, bringing the five firms’ total known related commitments to approximately $1.16 trillion. Alphabet and Amazon disclosed unused lease commitments of $85.2 billion and $13.721 billion respectively. Analysts point out that the $1.09 trillion cannot be simply viewed as corporate debt, as unused leases are usually multi-year, undiscounted payments, while lease liabilities on the balance sheet reflect present value. Nevertheless, this scale shows that tech giants are locking in massive infrastructure resources in advance for the AI era.