Meta discloses $347 billion in uncommenced data center leases

Big tech debt reached $460 billion as capital expenditures across six major hyperscalers approach $1 trillion next year.

When Meta issued $25 billion in corporate bonds earlier this year, fixed income investors had a clear entry for their spreadsheets.¹ The bond coupons, maturity dates, and overall debt load were public and fixed.¹ Evaluating the balance sheet impact of artificial intelligence infrastructure becomes far more complex when tech companies finance physical expansion through off-balance-sheet operating leases.

In its recent quarterly filing, Meta disclosed $347 billion in future lease obligations on projects that have not yet commenced.¹ That volume dwarfs the company's $84 billion of on-balance sheet debt recorded at the end of 2025.¹ Because these projects represent construction-stage facilities, the contracts have not yet populated the balance sheet as operating lease liabilities.¹ Calculating the present value of those leases once facilities go live will require an accounting discount rate, but whittling that sum below $100 billion would require a double-digit rate.¹

The scale of these contractual commitments is altering how credit analysts evaluate major technology companies. For decades, the dominant corporate models in Silicon Valley relied on asset-light operations focused on software, code, and intellectual property.²³ Generating massive free cash flow required minimal physical infrastructure. The buildout of generative artificial intelligence reverses that dynamic, demanding physical industrial plants packed with power-dense servers and advanced processors.²³

The surge in uncommenced obligations

The capital requirements of this physical pivot are visible across the tech sector. A research note from Moody's Ratings showed that six prominent hyperscalers, including Microsoft, Amazon, Alphabet, Meta, Oracle, and CoreWeave, face capital expenditures reaching $785 billion in 2026 before approaching $1 trillion in 2027.²³ Direct debt across the six tracked businesses has climbed to approximately $460 billion.²³ To avoid loading every dollar of hardware directly onto their balances, companies increasingly turn to long-term leases.

Across the six hyperscalers evaluated by Moody's Ratings, total lease commitments have reached $1.2 trillion.²³ More than $820 billion of that total stems from facilities that remain under construction.²³ Operating leases behave like debt in severe distress because rent payments are generally non-cancellable outside bankruptcy proceedings.¹ In US corporate reorganizations under Chapter 11, commercial landlords routinely sit alongside trade creditors and senior unsecured bondholders on creditor committees, holding equal claims.¹

Accounting standards established under the ASC-842 rule require corporate borrowers to report operating lease liabilities directly on their balance sheets.¹ In standard credit analysis, analysts calculate lease-adjusted leverage by adding the reported balance sheet operating lease liability to outstanding debt, then dividing that total by earnings before interest, taxes, depreciation, and amortization plus rent expense.¹ When hundreds of billions of dollars in future leases sit off the books in the construction phase, that formula misses substantial commitments.

Venture structures and market pricing

Off-balance-sheet structures have expanded beyond simple commercial contracts. Last year, Meta established a joint venture with private capital manager Blue Owl to develop Hyperion, an enormous data center campus situated in Louisiana.¹ Such vehicles allow developers to fund construction off corporate balance sheets, leaving the commitments hidden from standard leverage metrics until the projects finish construction and begin service.

Detailed lease disclosures are legally mandated only in annual regulatory filings.¹ In a year marked by unprecedented construction activity, creditors may encounter unexpected obligations when full annual reports appear. Furthermore, data center lease terms vary dramatically based on location, power purchase arrangements, and proprietary engineering designs, making individual contracts difficult to model from standard commercial templates.

Debt markets already show symptoms of friction. Meta bonds trade with spreads comparable to debt issued by Verizon, a company carrying a high-BBB credit rating, despite Meta retaining an AA-level credit profile.¹ Equity markets have not priced in a matching penalty, with Meta trading at a price-earnings ratio more than double that of Verizon.¹ Stock valuations often overlook off-balance-sheet commitments until cash constraints force an explicit adjustment.

Divergence among technology borrowers

The balance-sheet strain hits technology firms unevenly across different credit tiers. Top hyperscalers like Microsoft, Alphabet, Amazon, and Meta maintain some of the most liquid balance sheets in global corporate history, providing substantial protection against immediate credit rating downgrades.²³ Alphabet raised additional cash by announcing an $85 billion equity sale last month.²³

Pressure remains acute for debt-dependent competitors with lower credit ratings. Oracle carries a Baa2 rating with a negative outlook from Moody's Ratings, leaving its debt two notches above speculative grade.²³ CoreWeave operates directly in the high-yield credit market with a Ba3 rating, using complex private debt arrangements to purchase graphic processing unit clusters.²³

Credit analysts have also highlighted circular structures within the industry financing loop. Major hyperscalers have invested billions of dollars into early-stage artificial intelligence laboratories such as OpenAI and Anthropic.²³ Those artificial intelligence developers subsequently contract to purchase billions of dollars in cloud compute capacity from the exact tech giants providing their venture funding.²³ This web means that hyperscaler order backlogs depend heavily on the financial health and traffic projections of a concentrated circle of venture-backed customers.

Cash flow ceilings and refinancing risks

Even for well-capitalized tech giants, future capital demands threaten to outpace cash generation. Projections from sell-side analysts indicate that Meta could spend more than $500 billion on capital expenditures through 2028, exceeding its projected operating cash flow.¹ The company did not publish explicit multi-year capital expenditure guidance beyond 2026, creating wide variance in projected debt loads.¹

Bridging that cash gap will require tech companies and affiliated infrastructure ventures to tap debt markets repeatedly through the late 2020s. As bond vigilantes drive sovereign benchmark yields higher, funding hundreds of billions in physical campuses will extract steeper coupons. High customer demand for artificial intelligence capacity and massive existing contract backlogs provide revenue stability, yet corporate returns will remain under intense scrutiny as lease commitments mature.

What this rests on

28 sentences trace to 3 sources.

All 3Documents 1Records 0Reporting 2People 0

1. 1 FT Financial Times Document · ft.com · 12 Oct 2026 The huge lurking leverage of AI hyperscalers See the source
2. 2 C CNBC Third-party reporting · cnbc.com · 24 Jul 2026 Moody's says 'unprecedented' AI spending threatens credit quality of Amazon, Meta, Alphabet and others See the source
3. 3 R rocketnews.com Third-party reporting · rocketnews.com · 25 Jul 2026 Moody's says 'unprecedented' AI spending threatens credit quality of Amazon, Meta, Alphabet and others - RocketNews See the source