Stock Markets August 4, 2026 01:22 PM

Unstarted Data-Centre Leases Push Big Tech Commitments Past $1 Trillion

Five major cloud providers have locked in long-term payments for AI infrastructure that are not yet recorded as lease liabilities

By Maya Rios
Share
Twitter Reddit Facebook LinkedIn
MSFT META ORCL AMZN GOOGL

Microsoft, Meta, Oracle, Amazon and Alphabet have disclosed roughly $1.09 trillion in future lease payments for facilities that have not yet become operational. Most of the commitments are for data centres intended to support accelerating artificial intelligence workloads. Because accounting rules generally defer recognition until assets are available for use, these obligations mostly appear only in financial-statement notes rather than as recognised lease liabilities on balance sheets.

Unstarted Data-Centre Leases Push Big Tech Commitments Past $1 Trillion
MSFT META ORCL AMZN GOOGL
Summarize with
ChatGPT Perplexity Claude Grok Gemini

Key Points

  • Five major tech firms have disclosed about $1.09 trillion in uncommenced lease payments, largely for data centres to support AI workloads.
  • Accounting rules typically defer recognising lease liabilities until facilities are available for use, so much of this pipeline appears only in financial-statement notes and not as recognised lease liabilities.
  • Oracle, Microsoft and Meta account for large portions of the pipeline - Oracle disclosed $260 billion; Microsoft $329.1 billion; Meta $278.99 billion plus an additional $68 billion signed in July - while Alphabet and Amazon disclosed $85.2 billion and $137.21 billion respectively.

Overview

Five of the largest technology firms - Microsoft, Meta, Oracle, Amazon and Alphabet - have together committed about $1.09 trillion in future lease payments on facilities that have not yet begun operations, according to company disclosures. The bulk of these commitments are for data-centre capacity targeted to serve rapidly growing artificial intelligence computing needs.


Accounting treatment and scale

The commitments represent a substantial tranche of the AI infrastructure buildout that has been contractually locked, but which has yet to be recorded as debt-like lease liabilities on company balance sheets. Under typical accounting rules, signed leases are not added to recognised lease liabilities until the leased asset is available for use. Until that point, firms disclose the undiscounted future payment obligations in the notes to their financial statements.

That timing difference helps explain why the uncommenced commitments - nearly $1.09 trillion - are almost four times the roughly $285 billion of lease liabilities already recognised by the five companies. The uncommenced figures are commonly presented as undiscounted multi-year payments, whereas recognised lease liabilities reflect the present value of those payments.


Firm-level breakdown

  • Microsoft disclosed the largest pipeline of uncommenced commitments, at $329.1 billion, compared with $88.52 billion of recognised lease liabilities.
  • Oracle reported $260 billion of uncommenced commitments - nearly seven times its $37.89 billion of recognised lease liabilities. Much of Oracle's pipeline is for data centres expected to commence between fiscal 2027 and fiscal 2029 and typically carry terms of 15 to 19 years. The company has cautioned that the duration, renewal provisions and pricing of those leases may not align with customer contracts, creating exposure if customers do not renew or cannot perform.
  • Meta disclosed $278.99 billion of uncommenced operating and finance lease payments, and later signed an additional $68 billion of data-centre leases in July. Including those subsequent agreements brings the five firms' publicly known pipeline to about $1.16 trillion.
  • Alphabet reported $85.2 billion of uncommenced leases.
  • Amazon disclosed $137.21 billion of uncommenced commitments, though that figure encompasses a broader set of leased assets, including warehouses, offices, aircraft and vehicles, making it less directly comparable to the other companies' largely data-centre-focused totals.

Leverage implications

The disclosed commitments mean a sizable portion of the AI-related capital build is already contractually committed, even if it is not yet reflected in recognised lease liabilities and conventional leverage metrics. For Oracle in particular, borrowings were about 4.4 times trailing EBITDA at the end of May based on company filings and market-data compilations. When recognised operating and finance lease liabilities are included, that leverage measure rose to about 5.7 times. Credit analysts have already factored Oracle's uncommenced lease pipeline into some models; S&P Global Ratings indicated it incorporated the $260 billion of uncommenced leases into its adjusted-debt forecast and expected leverage to be around 4.4 in fiscal 2027.


Potential outcomes

If demand for AI computing continues its rapid expansion, the completed data-centre capacity underlying these lease commitments could become the foundation for the next phase of cloud growth. Conversely, if demand does not materialise as projected, the companies could be left with extensive, costly and long-lived leased capacity that is difficult to exit.


Disclosure and market treatment

The uncommenced lease commitments are disclosed in financial-statement notes and are not hidden. Rating agencies and other analysts may already account for portions of the commitments when assessing adjusted leverage and credit metrics. Still, the magnitude of the uncommenced pipeline highlights how much of the AI infrastructure rollout has yet to be incorporated into standard reported lease liabilities, fixed charges and leverage calculations.


Conclusion

The companies' contracts for future data-centre capacity reveal that a substantial share of Big Tech's AI investment is already locked in, though not yet reflected as recognised lease liabilities. The ultimate financial and operational impact will depend on how demand for AI computing evolves and how closely lease terms align with customer contract performance.

Risks

  • If demand for AI computing falls short of expectations, the companies could be obliged to pay for substantial, long-lived leased capacity that is difficult to reduce - affecting cloud providers, data-centre operators and commercial real estate markets.
  • Term, renewal and pricing mismatches between long-term data-centre leases and customer contracts - a specific concern noted for Oracle - could create earnings and cash-flow risk for affected technology and hosting businesses.
  • The accounting timing gap means conventional reported lease liabilities and leverage metrics may understate future fixed obligations until facilities are placed into service, which has implications for credit assessment and corporate finance measures in the tech sector.

More from Stock Markets

Voyxact Preserves Kidney Function at Two Years in IgA Nephropathy, BTIG Says Aug 4, 2026 Goldman Sachs Sticks With Buy Ratings on Samsung, SK Hynix After Sharp Pullback Aug 4, 2026 Appeals Court Bars EPA From Halting Roughly $20 Billion in Clean Energy Grants Aug 4, 2026 Thryv Shares Collapse After Disappointing Q2 Results and Weaker SaaS Outlook Aug 4, 2026 Jefferies Lowers Cigna to Hold Citing Mounting 2027 Earnings Risks Aug 4, 2026