AI data-centre race builds $1 trillion lease burden for Big Tech

Big tech firms Microsoft, Meta, Oracle, Amazon, and Alphabet have collectively committed $1.09 trillion towards future lease payments, largely for data centers supporting the AI surge. Most of these commitments have not yet materialized as lease liabilities on their balance sheets. If AI computing demand remains robust, these data centers will fuel the next phase of cloud growth.

However, if demand falters, companies might be burdened with vast, costly, long-term facilities that are hard to dispose of. The $1.09 trillion is nearly four times the $285 billion of recognized lease liabilities on their balance sheets. The difference is due to accounting practices; signed leases are typically not recorded as liabilities until the facility is ready for use, whereas the $1.09 trillion figure is disclosed in financial notes.

Oracle leads with the largest apparent concentration risk, with $260 billion in uncommenced commitments, nearly seven times its recognized lease liabilities. Oracle's data center leases are expected to kick off between fiscal 2027 and fiscal 2029 and last for 15 to 19 years. Oracle's debt-to-EBITDA ratio stood at 4.4 times at the end of May, rising to 5.7 times when including recognized lease liabilities.

Microsoft follows with a disclosed pipeline of $329.1 billion, against $88.52 billion in recognized lease liabilities. Meta disclosed $278.99 billion in uncommenced lease payments, which climbed to $348.99 billion after signing $68 billion more in data center leases in July. Alphabet and Amazon reported $85.2 billion and $137.21 billion in uncommenced lease commitments, respectively. Amazon's figure is less comparable due to its broader lease portfolio including other facilities.