← Back to News

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

negativeLegalMulti dayYahoo Finance ·4 Aug 2026Original article ↗
Oraklio AI Analysis

The news is focused on Oracle’s disclosure of large off-balance-sheet (uncommenced) lease commitments and cited rating-agency treatment, which can influence credit/leverage perceptions and valuation over the next several sessions.

Article

AI data-centre race builds $1 trillion lease burden for Big Tech 1 / 3 AI data-centre race builds $1 trillion lease burden for Big Tech FILE PHOTO: Microsoft boosts Wisconsin data center spending to $7 billion Patturaja Murugaboopathy Tue, August 4, 2026 at 7:19 PM GMT+2 2 min read MSFT META ORCL AMZN GOOG By Patturaja Murugaboopathy Aug 4 (Reuters) - Microsoft Corp, Meta Platforms Inc, Oracle Corp, Amazon and Alphabet have committed about $1. 09 trillion in future payments under leases that have not yet begun, mostly for data centres ‌needed to power the artificial intelligence boom. The commitments show that a substantial part of Big Tech's ‌AI spending spree has already been locked in, without yet appearing as debt-like lease liabilities on company balance sheets.

If demand for AI computing ​continues to surge, the facilities will underpin the next phase of cloud growth. If it does not, the companies could be left paying for vast amounts of costly, long-lived capacity that is difficult to shed. The total is nearly four times the roughly $285 billion of lease liabilities already recognised on the companies' balance sheets, according to company filings compiled by ‌Reuters.

The gap reflects accounting treatment. Signed leases ⁠generally are not recorded as liabilities until a facility is available for use. Until then, companies disclose the future payments in notes to their financial statements.

The commitments are not ⁠hidden, and rating agencies may already account for some of them. But their scale reveals how much of the AI buildout has yet to enter reported lease liabilities, fixed charges and reported leverage measures. The $1.

09 trillion cannot simply be added to debt. ​Uncommenced ​lease commitments are generally undiscounted payments spread over many years, ​whereas recognised lease liabilities reflect their present value. Oracle ‌has the largest apparent concentration risk.

It disclosed $260 billion of uncommenced commitments, nearly seven times its $37. 89 billion of recognised lease liabilities. The commitments are substantially for data centres, expected to begin between fiscal 2027 and fiscal 2029, and generally run for 15 to 19 years.

Oracle has warned that the duration, renewal terms and pricing of its data-centre leases may not align with customer contracts, leaving it exposed if customers do not renew or cannot perform. Its borrowings ‌equalled about 4. 4 times trailing EBITDA at the end of ​May, according to a Reuters analysis of LSEG data and company ​filings.

Including recognised operating and finance lease liabilities ​lifted that ratio to about 5. 7 times. S&P Global Ratings said it incorporated Oracle's $260 billion of ‌uncommenced leases into its adjusted-debt forecast and expected ​leverage to be around 4.

4 ​in fiscal 2027. Microsoft had the largest disclosed pipeline, at $329. 1 billion, against $88.

52 billion of recognised lease liabilities. Story Continues Meta disclosed $278. 99 billion of uncommenced operating and finance lease payments, then signed a further $68 billion of data-centre leases in ​July, lifting the five companies' known pipeline ‌to about $1.

16 trillion, including those later agreements. Alphabet reported $85. 2 billion of uncommenced leases, while Amazon disclosed $137.

21 billion. ​Amazon's figure is less directly comparable because its lease portfolio also includes warehouses, offices, aircraft and ​vehicles.

Oraklio AI Trading Intelligence

News is just the start.

Oraklio turns news, price data, and market signals into structured BUY / SELL / NO_TRADE calls - updated continuously throughout the trading day.

Get started free

Already have an account? Sign in →