Under the AI infrastructure construction frenzy, an unprecedented scale of debt expansion is quietly taking shape—and the most dangerous part of it has never appeared on any balance sheet.
Goldman Sachs' latest report predicts that by 2027, mega-scale cloud computing enterprise capital expenditure will reach $1.1 trillion to $1.4 trillion, far exceeding market consensus. However, according to Morgan Stanley's in-depth research, this already jaw-dropping figure is just the tip of the iceberg.
Nearly $1 trillion in procurement commitments, over $800 billion in unrecognized leasing contracts, and billions of dollars in supplier financing arrangements together constitute an off-balance-sheet exposure of around $1.8 trillion—these liabilities, off the balance sheet, are effectively locking in future cash outflows.
The market has not yet fully priced in the above risks.
Morgan Stanley warns that the leverage ratio of mega-scale cloud enterprises has surged from 0.9 times to 1.8 times in just two quarters, with capital expenditure growth continuing to outpace revenue and free cash flow growth, while the real impact of depreciation pressure has yet to come.
Meanwhile, private credit institutions represented by Apollo and Blackstone are transferring leverage to the supply chain level through SPVs (special purpose vehicles), forming a highly cyclical and opaque financing structure. If the AI commercialization process falls short of expectations, or enterprise customers massively shift to low-cost alternative solutions, the vulnerability of the entire financing chain will be exposed.
Debt Issuance Frenzy: AI Has Become the Biggest Wild Card in the Public Market
According to Morgan Stanley's latest "AI Debt Financing Tracker Report," as of the end of May 2026, the global AI-related bond issuance has reached $236 billion, a 357% surge from the same period in 2025.
Morgan Stanley expects that the total annual issuance of AI debt will surpass $570 billion, and in the second half of the year, as capital expenditure financing demand is concentrated, the issuance pace will further accelerate.
In April, monthly issuance of AI-related bonds exceeded $74 billion, hitting a new high for the year. Project financing for data center construction accounted for 85% of the high-yield bond supply and 40% of the investment-grade bond supply. Meanwhile, the five hyperscale cloud giants—Amazon, Meta, Google, Microsoft, and Oracle—currently represent 4% of the entire investment-grade bond index.
On the leverage front, the overall gross leverage ratio of hyperscale cloud giants has risen from 0.9 times in Q3 2025 to the current 1.8 times, increasing by approximately 0.3 times each quarter, surpassing the leverage levels of the entire energy sector.
Morgan Stanley pointed out that due to supply pressure, relevant credit spreads have shifted from the AA range to the A range and may further widen. Meta's credit spread is currently wider than the CDX IG benchmark.
On the free cash flow front, Morgan Stanley predicts that Amazon and Meta will reach near-zero or negative free cash flow in 2026, at which point incremental financing will rely almost entirely on new debt.
$1.8 Trillion Off-Balance-Sheet Exposure: Invisible Liabilities, Locked Cash Outflows
Todd Castagno from Morgan Stanley's Global Valuation, Accounting, and Tax team highlighted in a report that focusing solely on capital expenditure figures would significantly underestimate the true financial commitments of the AI build-out cycle. In addition to disclosed capital expenditures, there are three key off-balance-sheet exposures:
Procurement Commitments of around $982 billion. Hyperscale cloud giants and NVIDIA have long-term procurement contracts totaling nearly $1 trillion. Under accounting standards, unless a loss is expected on the contract, these obligations are not recognized as liabilities until the goods are delivered. Therefore, nearly a trillion dollars of future cash outflows are currently not reflected as any liability on the balance sheet.
Of note, NVIDIA's own inventory and procurement obligations have risen to around 32% of consensus revenue estimates for the FY2027, well above the historical range of 15% to 20%, extending supply chain commitment risks to the chip supplier side.
Operating Lease Commitments of approximately $822 billion. Over $800 billion of lease agreements have been signed but not yet commenced, not included in current lease liabilities. Additionally, arrangements such as variable lease payments, renewal options, residual value guarantees, and similar terms remain off the balance sheet.
Morgan Stanley estimates that if finance leases are included, Microsoft's capital expenditure as a percentage of sales will increase from 33%/50% (FY2026/FY2027) to 44%/64%, while Oracle could rise from 76%/115% to 101%/189%.
Unpaid capital expenditures in accounts payable amount to approximately $110 billion. The Days Payable Outstanding (DPO) of mega-scale cloud companies has significantly extended — Oracle increased by 370% YoY, Meta by 73%, Microsoft by 69% — indicating that the entire supply chain is effectively financing AI development, with suppliers shouldering liquidity pressures that should have been borne by the buyer.
SPV and Recurring Financing: Leveraging Shifted into the Shadows
Another core dimension of off-balance-sheet risk is the use of SPVs to create revolving financing structures.
This week, Apollo and Blackstone jointly completed a $35 billion "chip-backed" private credit transaction for Anthropic, showcasing the operational logic of this model:
Broadcom provided endorsements to this SPV, Anthropic used the raised funds to purchase Google chips manufactured by Broadcom, with Google holding a 14% stake in Anthropic; Morgan Stanley, arranging this transaction, also provided loans to the participating investors.
Morgan Stanley's AI ecosystem financing network map shows multiple rounds of financing and buybacks between OpenAI, Oracle, NVIDIA, Microsoft, CoreWeave, AMD, and Amazon, with funds circulating among a few entities, and the SPV being a core tool to enable this circulation.
Reportedly, Apollo's insurance subsidiary Athene is particularly active in the above structure — raising funds by selling annuities to retirees, then channeling the funds into an SPV for AI infrastructure financing.
This model shifts leverage from the visible balance sheet of mega-scale cloud enterprises to suppliers and the private credit ecosystem, making the true systemic risk exposure difficult to be identified and aggregated by external observers.
Depreciation Cliff and Monetization Gap: The Postponed Impact
The current financial data shows a systematic optimistic bias. A significant amount of capital expenditures is currently capitalized as Construction in Progress (CIP), without having started depreciation, leading to artificially inflated reported profit margins and underestimated future expense pressure.
Oracle, Meta, and Google's CIP balances have increased by approximately 200%, 90%, and 55% year-on-year, respectively.
Once these assets start depreciating, the impact will be released in a concentrated manner.
Morgan Stanley predicts that the cumulative depreciation of Microsoft, Oracle, Meta, and Google over the next three years will exceed $520 billion. For example, Oracle's depreciation as a percentage of revenue may increase from the current 7% to 28% by the 2028 fiscal year; Meta may increase from 9% to 19%.
Against this backdrop, the only path to maintaining profit margins is through synchronized significant revenue growth—yet the current upward revisions of revenue forecasts significantly lag behind the upward revisions of capital expenditure forecasts.
Data shows that consensus capital expenditure forecasts for Google in 2026 have been revised upwards by 139% compared to a year ago, while Meta and Amazon have seen increases of 85% and 81%, respectively. Oracle has seen the largest upward adjustment of 175%.
Meanwhile, revisions to revenue forecasts significantly lag behind, and the structural mismatch where capital expenditures precede commercialization is becoming increasingly clear.
Furthermore, over $2 trillion in Remaining Performance Obligations (RPO) is highly concentrated in a few large, long-term contracts, highlighting the concentrated risk of counterparties—a problem with any major participant in the ecosystem could potentially trigger a chain reaction.
Mismatch of Timing Rather Than Immediate Solvency Crisis
Morgan Stanley's conclusion is that the aforementioned risks do not currently pose an imminent solvency crisis but rather a combination of timing mismatches and information disclosure gaps: deferred depreciation pressure, capital expenditures outpacing monetization progress, leverage shifting to suppliers and the private credit layer, as well as significant discounts in comparability of capital intensity between companies due to accounting classification differences.
The ultra-large-scale cloud companies clearly recognize the limited window of current market sentiment and are seizing the opportunity to maximize financing.
Goldman Sachs analyst Ryan Hammond points out that if AI infrastructure investment reaches 2% to 3% of GDP, analogous to the historical construction cycles of railways and the automotive industry, capital expenditures in 2027 could reach $11 trillion; in an extreme scenario, combining the cash flows of ultra-large-scale cloud companies with the capacity of the investment-grade credit market, the upper limit could reach $14 trillion.
However, all of this hinges on Large Language Models (LLMs) being able to continue driving token valuations and maintaining sufficient enterprise customer stickiness. More and more enterprises are now turning to AI products that offer performance close to LLMs but at a significantly lower price.
Once a structural shift in demand occurs, this carefully constructed financing system will face a fundamental stress test.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
