--- title: Big Tech's $1.65 trillion in off-balance-sheet AI debt,... description: A Nikkei study reveals that Alphabet, Microsoft, Amazon, Meta, and Oracle collectively carry $1.65 trillion in off-balance-sheet debt tied to AI infrastructure... canonical: https://daily.dev/posts/big-tech-s-1-65-trillion-in-off-balance-sheet-ai-debt-explained-ypcfcopqb twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Big Tech's $1.65 trillion in off-balance-sheet AI debt, explained | daily.dev og:description: A Nikkei study reveals that Alphabet, Microsoft, Amazon, Meta, and Oracle collectively carry $1.65 trillion in off-balance-sheet debt tied to AI infrastructure... og:url: https://daily.dev/posts/big-tech-s-1-65-trillion-in-off-balance-sheet-ai-debt-explained-ypcfcopqb og:image: https://api.daily.dev/og/posts/yPcfCoPQB.png og:image:alt: Big Tech's $1.65 trillion in off-balance-sheet AI debt, explained og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Big Tech's $1.65 trillion in off-balance-sheet AI debt, explained **[Collections](https://daily.dev/sources/collections)** · 6 min read · 1 upvotes · 0 comments ## Summary A Nikkei study reveals that Alphabet, Microsoft, Amazon, Meta, and Oracle collectively carry $1.65 trillion in off-balance-sheet debt tied to AI infrastructure — exceeding their officially reported $1.35 trillion in on-balance-sheet liabilities and representing an eightfold increase over four years. The obligations are structured through data center leases, GPU supply contracts, special purpose vehicles, and joint ventures, all technically compliant with current accounting rules but largely invisible to investors reading standard financial statements. Meta's off-balance-sheet liabilities alone are estimated at $420 billion, nearly triple its reported debt, while Oracle's off-book obligations have grown thirtyfold. Credit agencies including S&P, Morgan Stanley, and Moody's have flagged the risk, particularly around timing: when data centers go live, leases roll onto balance sheets, and if AI demand falls short of projections, lenders and insurers may face losses they haven't fully priced in. ## Content ## The hidden leverage behind the AI buildout If you've been reading Big Tech earnings reports and thinking the numbers look manageable, you're only seeing part of the picture. A Nikkei study found that Alphabet, Microsoft, Amazon, Meta, and Oracle collectively carry $1.65 trillion in off-balance-sheet debt tied to AI infrastructure — more than the $1.35 trillion they officially report. That ratio is striking on its own. What's more striking is how fast it got there: the figure has grown roughly eightfold in four years. Meta is the most extreme case. Its off-balance-sheet liabilities sit at around $420 billion, nearly triple its reported debt. Oracle's equivalent obligations have grown thirtyfold in four years. These aren't rounding errors. The mechanism is legal. Companies use special purpose vehicles and joint ventures to build data centers without recording the full cost on their own books. Accounting rules allow future lease commitments to stay in footnotes as long as lease renewal isn't "reasonably certain." Moody's estimates the five largest US hyperscalers hold $969 billion in future lease commitments, with $662 billion not yet appearing on any balance sheet. Analysts at S&P, Morgan Stanley, and Moody's have all flagged the risk: when data centers go live, leases roll onto balance sheets, and if AI demand disappoints, losses fall on lenders and insurers. This is not Enron. The structures are disclosed, technically. But disclosed in footnotes is not the same as visible. --- ## How the deals actually work: Meta and BlackRock in El Paso The Meta-BlackRock joint venture in El Paso is a clean example of how this works in practice. The two companies have formalized a $14 billion deal to build a one-gigawatt AI data center campus. BlackRock — channeling investment through Global Infrastructure Partners and HPS Investment Partners — owns 80% of the campus. Meta owns 20%, acts as construction manager, and is the sole tenant. BlackRock is leading a $12 billion bond sale to finance the project. That $12.5 billion in project debt stays off Meta's balance sheet. Meta books the cost as rent rather than capital expenditure, which matters a lot given the company's $125–145 billion capex guidance for 2026. The campus is expected online in 2028, employing around 4,000 construction workers at peak and roughly 300 permanent staff. This mirrors Meta's earlier Hyperion campus in Louisiana, where Blue Owl held 80% and sold $27 billion in bonds — the largest private-debt deal on record at the time. For BlackRock CEO Larry Fink, the El Paso deal validates a $25 billion acquisition spree that brought Global Infrastructure Partners and HPS in-house. BlackRock now both originates these assets and sells debt against them. The key risk analysts flag: the bonds are long-dated, AI chips depreciate quickly, and leases reportedly run shorter than the campuses they finance. --- ## Google's $44 billion in other people's rent Google has taken a slightly different approach. The company disclosed $44 billion in lease guarantees on data centers it neither owns nor occupies, up from $6.5 billion just nine months ago. The logic is almost elegant: Google guarantees a data center developer's borrowing costs, making the project financially viable. The developer builds. Tenants — including Anthropic — move in and buy Google's TPU chips. Google's guarantee is effectively a sales tool that doesn't show up as a liability in the conventional sense. It's a financial structure designed to sell hardware while keeping the associated risk in the footnotes. --- ## Nvidia's circular financing problem The most discussed version of this dynamic right now involves Nvidia. The company announced over $750 billion in AI infrastructure deals, including a $500 billion-plus initiative with SK Group in South Korea. The deal drawing the most scrutiny: Nvidia is reportedly in talks to provide $250 billion in financing guarantees for OpenAI's planned 10-gigawatt data center campus in southern Ohio, developed by SoftBank's energy subsidiary. Because OpenAI lacks an investment-grade credit rating, Nvidia's balance sheet would backstop the lease and construction debt. A separate tranche of up to $350 billion for chip purchases could push the total project cost above $500 billion. Nvidia has already committed over $40 billion to AI equity positions, including roughly $30 billion in OpenAI. The circular nature of this is hard to ignore: a chip supplier underwrites its customer's ability to keep buying its own chips. Critics including Michael Burry and tech commentator Ed Zitron have made exactly this point. The IMF and BIS have both flagged circular financing as a systemic risk. Jensen Huang has called the concern "ridiculous," arguing the equity stakes are small relative to total capital raised. Markets are less dismissive. Nvidia's five-year credit default swap spread hit a record 82 basis points after the announcements. Nvidia shares fell nearly 5%, wiping roughly $250 billion from its market cap. No agreement on the OpenAI deal has been signed. Talks are ongoing. --- ## What the market signals are saying Beyond the CDS spread on Nvidia, there are other signals worth noting. CoreWeave — one of the "neoclouds" that exists primarily to rent GPU capacity — carries junk-level bond spreads. Google issued a century bond that has already lost 10% of its value. Oracle received a credit downgrade. The broader concern some analysts are raising is structural. Hyperscalers have collectively spent over $1 trillion in capex on AI infrastructure. None have disclosed meaningful AI revenues. The majority of apparent AI compute demand is concentrated in a handful of companies — OpenAI and Anthropic chief among them — neither of which is profitable. The SPV structures used to finance data centers resemble, at least superficially, the CDO structures used to finance subprime mortgages before 2008. The exposure runs through pension funds, insurance companies, and private credit markets. If the data centers fail to generate sufficient customer revenue, the losses don't stay with the hyperscalers — they flow to whoever holds the bonds. Whether that comparison is fair or alarmist depends on how much real AI demand actually materializes. That's the question nobody has a clean answer to yet. But investors reading quarterly earnings this week are seeing less than half the leverage picture. ## Similar posts on daily.dev - [Big Tech's AI debt hits $350bn and heads to Europe](https://daily.dev/posts/big-tech-s-ai-debt-hits-350bn-and-heads-to-europe-zyvvvckzp) · The Next Web · 0 upvotes · 0 comments --- Tags: [#ai](https://daily.dev/tags/ai), [#cloud](https://daily.dev/tags/cloud), [#big-tech](https://daily.dev/tags/big-tech) [View this post on daily.dev](https://daily.dev/posts/big-tech-s-1-65-trillion-in-off-balance-sheet-ai-debt-explained-ypcfcopqb) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Big Tech's $1.65 trillion in off-balance-sheet AI debt, explained","url":"https://daily.dev/posts/big-tech-s-1-65-trillion-in-off-balance-sheet-ai-debt-explained-ypcfcopqb","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/big-tech-s-1-65-trillion-in-off-balance-sheet-ai-debt-explained-ypcfcopqb"},"datePublished":"2026-07-21T17:40:11.496Z","dateModified":"2026-07-28T16:49:58.457Z","description":"A Nikkei study reveals that Alphabet, Microsoft, Amazon, Meta, and Oracle collectively carry $1.65 trillion in off-balance-sheet debt tied to AI infrastructure...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/813bec8bee0f60c508ffd605485433c8?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/813bec8bee0f60c508ffd605485433c8?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/big-tech-s-1-65-trillion-in-off-balance-sheet-ai-debt-explained-ypcfcopqb","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai,cloud,big-tech","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"Big Tech's $1.65 trillion in off-balance-sheet AI debt, explained"}]} ```