• The commitments are nearly four times the companies’ recognised lease liabilities of about US$285 billion
  • Oracle’s US$260 billion pipeline largely covers data centres expected to open between fiscal 2027 and 2029

The fact

Microsoft, Meta, Oracle, Amazon and Alphabet have committed about US$1.09 trillion in future payments under leases that have not yet begun, according to company filings compiled by Reuters. Most of the agreements cover data centres being built for AI and cloud computing.

The total is nearly four times the roughly US$285 billion of lease liabilities already recognised on the companies’ balance sheets. Signed leases are generally recorded as liabilities only when the facility becomes available for use. Until then, the future payments are disclosed in notes to company accounts.

Microsoft reported the largest pipeline at US$329.1 billion, followed by Meta at US$278.99 billion and Oracle at US$260 billion. Oracle said most of its commitments relate to data centres expected to become available between fiscal 2027 and 2029, with leases generally running for 15 to 19 years. Amazon’s US$137.21 billion figure also includes warehouses, offices, aircraft and vehicles.

The assessment

The companies are reserving data-centre space years before they can use it. This helps them secure capacity for future AI growth, but it also locks them into long payment schedules before the buildings open. The timing creates the main risk. Customer demand and hardware needs can change much faster than a large data centre can be built.

Oracle has warned that some leases may run longer than the customer contracts supporting them. Other companies may face the same problem, although their filings provide less detail. For BTW readers, these companies are making large capacity decisions years before they know exactly how much space they will need. If demand falls short, they could be left paying for data centres that take longer to fill.

What to watch

Watch when the leased facilities open, how quickly the companies fill them and whether cloud revenue grows alongside lease costs. Changes to construction plans, rating-agency treatment, or efforts to sublease or renegotiate space would show where these long-term commitments are becoming harder to manage.