Building on the Edge of a Cliff: The Hidden Debt Google and Meta Don't Dare Disclose

In August 2025, someone registered seven companies in Delaware, all with "Beignet" in their names.

Beignet, a fried dough pastry common in New Orleans, is covered in thick powdered sugar, some of which always falls on your clothes when you pick one up.

No matter how hard you look, you wouldn't see any connection between this dessert and AI.

A month after the Beignet companies appeared, Meta built a data center in Louisiana called Hyperion, covering an area equivalent to four New York Central Parks.

To build this mega data center, Meta borrowed a total of $27.3 billion.

But if you dig into Meta's financial reports, you'll find that on its balance sheet, the total record related to this project is only a $2.37 billion investment.

The remaining over $200 billion in debt has disappeared.

This is not an isolated case.

On July 22, the Nikkei published a report stating that US tech giants have hidden up to $1.65 trillion in debt where people can't see it, a figure surpassing the $1.35 trillion in total liabilities on their balance sheets.

We went through the filings these five companies submitted to the SEC from beginning to end and found the reality is even more exaggerated than that report suggested.

On July 23, the day after the report was published, Google's parent company, Alphabet, submitted filings for the new quarter. Its purchase commitments jumped from $332.4 billion three months earlier to $811 billion. A year ago, that number was $62.1 billion.

One year, a 13-fold increase. This also pushed the $1.65 trillion figure cited by Nikkei up to $2.13 trillion.

Over the past year, the debt created by the five companies—Microsoft, Google, Amazon, Meta, and Oracle—on data centers grew from $710.8 billion to $1.55 trillion. If you include procurement and construction contracts for hardware like GPUs, it rose from $1.02 trillion to $2.86 trillion. An almost two-fold increase in one year.

And the portion actually sitting on their balance sheets is only a quarter of this final number. A total of $2.13 trillion in "data center debt" has vanished from the giants' balance sheets.

Where has this debt gone?

The Bank for International Settlements (BIS) noticed this early on. In a quarterly report from March 2026, it gave this practice a name: "Shadow Borrowing." The report stated that these arrangements are economically equivalent to debt but mostly remain off the corporate balance sheet.

Three months later, in its annual report, the BIS unusually listed AI泡沫 and circular financing alongside sovereign debt as primary risks to the global financial system.

Going through the various filings of the five companies reveals at least five different techniques: SPVs, credit wrappers, finance leases, residual value guarantees, and future-dated leases.

And the people helping them create this shadow debt are turning this into a brand new business.

Tech Giants' Collective "Return to Poverty"

For the past two decades, tech giants have been the most comfortable type of company in the US capital market. They are highly profitable, have cash on their books, and buy back their own stock.

In Q4 2021, Microsoft, Google, Amazon, Meta, and Oracle collectively repurchased $48 billion in stock, with Meta alone spending $20 billion. With so much money, shareholders didn't worry about them ever running short of cash.

But in Q1 2026, the total buybacks from these five companies plummeted to $4.6 billion.

Over the past two years, tech companies' AI-related capital expenditures have more than doubled (1.5x increase), but their operating cash flow has grown by less than 60%. According to this growth rate, by mid-2027, tech giants will "collectively return to poverty," falling back into an era of losses.

Morgan Stanley estimates that by 2028, tech companies will need to spend approximately $2.9 trillion on AI, but can generate only about $1.4 trillion internally. The $1.5 trillion gap must be sourced from outside cash flow.

So they started borrowing. From 2020 to 2023, the five companies issued an average of about $31.3 billion in bonds annually. By July 2026, this figure reached $189.7 billion, six times the historical average.

The massive amount of corporate bonds is starting to strain the market. Over the past nine months, Amazon's bond oversubscription ratio has been consistently declining. It was 5.3 times in November 2025, but dropped to just 1.6 times by July 2026.

The cost of issuing bonds has also increased. This year, in the entire investment-grade bond market, an average of 4 basis points extra is needed to sell a new issue. But for Amazon's July issuance, they needed to offer 18 to 21 basis points to get it done.

Google and Oracle went even further, raising funds directly through stock offerings. After raising nearly $80 billion, they launched another $60 billion in ATM (At-The-Market) offering programs.

Financial media outlets collectively complain that tech giants have broken the "unspoken covenant" with investors. In the past, buying these companies' stocks was implicitly buying into net cash, low debt, and continuous buybacks. Now they need to issue massive debt and suspend buybacks.

But the real trouble is written on the giants' balance sheets.

The more debt on the balance sheet, the more cautious rating agencies become, and the fewer people can buy their bonds. The first among the five to hit this wall was Oracle. Its capital expenditure jumped from $21.2 billion to $55.7 billion within a year. In July 2026, S&P downgraded Oracle from BBB to BBB-, just one notch above junk status. If downgraded further, global insurance companies and pension funds would be forced by regulations to dump Oracle's bonds.

So the giants don't just need more money; they need money that is more hidden, longer-term, and carries less onerous reporting responsibilities. And this kind of money simply cannot be found in the public market.

After SaaS "fails," Wall Street is also looking for new business

Just as tech companies were worrying about money, another part of Wall Street was also searching for an exit.

In the first half of 2026, a fund managed by Blue Owl received redemption requests approaching 40% of its assets for two consecutive quarters, but actual payouts were only slightly over 10%. The company's stock price fell from $24 to $9.

Blue Owl is one of the world's largest private credit institutions, managing over $310 billion in assets. This fund, which faced a redemption run, specializes in the software sector, with software loans accounting for over 60% of its portfolio.

In 2026, software loans happened to be the last thing Wall Street wanted to hold onto.

From its peak in October 2025, software stocks have fallen by nearly 40%. The market's explanation is simple: AI will kill software. Traditionally, investors were willing to give software companies high valuations because customers would renew subscriptions annually, making revenue seem perpetually contract-bound. Now, whether customers will continue to renew has suddenly become a question.

But in reality, the fundamentals of software companies aren't that bad. Microsoft's 365 subscription revenue growth rate rose from 15% to 19%. The revenue growth rates for software companies like ServiceNow, Salesforce, and Snowflake have accelerated for five consecutive quarters. Gartner has even raised its forecast for global software spending.

However, in the financial world, confidence often trumps fundamentals.

Blue Owl acknowledged in a shareholder letter that the market's concerns about AI disrupting software companies have significantly affected how investors view software credit risk.

Wall Street urgently needs a new narrative to get investors interested again. That narrative is data centers.

Software loans bet on whether customers will renew next year. Data center loans bet on whether AI companies will need computing power. The former question is getting harder to answer, while the latter seems almost rhetorical. The stronger AI gets, the more valuable the server rooms become.

Currently, data center loans are the hottest business on Wall Street. In December 2025, Blue Owl rejected Oracle's data center project in Michigan, citing it didn't meet underwriting standards. However, PIMCO quickly outbid them and secured the deal. Such instances of snatching deals away happen practically every month on Wall Street.

On one side, you have cash-strapped tech giants; on the other, asset management institutions with nowhere to invest their money. So, the two sides came together.

Their first major creation was Beignet, the dessert mentioned at the beginning.

How did Meta hide $28 billion in debt inside a dessert?

That Hyperion hyperscale data center in Louisiana, co-developed by Meta, was initially registered under Laidley LLC. This entity operates the campus and is the one that signed a 15-year power supply contract with the local utility company.

Hyperion's footprint compared to Manhattan, NYC. Source: Bloomberg

Laidley belongs to Project Beignet Holdings, a joint venture. The property rights for the data center reside here.

The major shareholder of the joint venture is called Beignet Investor. The $27.3 billion in bonds were issued through it.

The debt is not placed on the company that owns the data center, but on its shareholder.

Above that is Beignet Pledgor. In legal terms, a pledgor is someone who offers collateral. It wholly owns Beignet Investor and pledges 100% of Beignet Investor's equity to a trustee.

This gives creditors a very straightforward form of collateral. If something goes wrong, the trustee doesn't have to go to Louisiana to appraise a data center, sell it, and wait through a lengthy lawsuit. They can just execute the equity contract. The campus continues operating, the lease continues running, but the person collecting the rent changes.

Above Beignet Pledgor, there are four more companies, including one called Beignet Net Lease Aggregator. At the very top is a net lease real estate investment trust (REIT) under Blue Owl called OSNL, along with co-investors.

All seven companies are registered in Delaware, where limited liability companies don't have to publicly disclose their members or capital contributions.

The $27.3 billion debt was also not publicly issued. It went through Rule 144A, sold only to qualified institutional buyers, without a public prospectus. To see the terms of the transaction, you first need to sign a non-disclosure agreement. It's a "shelf 144A," meaning it will never be converted into a registered public bond.

Searching for "Beignet" in the SEC's full-text filing system (EDGAR) yields only one narrative hit: a subsequent events footnote in Blue Owl's quarterly report. By the annual report, it disappears along with "Meta" and the name of the project's parish, leaving only a single line aggregating "net lease data centers."

This is already dizzying, but it's only the legal isolation.

SPVs are not new. Real estate and infrastructure industries have used them for decades. Accounting standards have long known that companies pack debt into these structures and have set up two thresholds. Whether an entity needs to be consolidated into a company's financial statements depends not only on ownership percentage but also on who directs the most significant activities and who bears most of the losses and reaps most of the benefits.

Meta is Hyperion's sole tenant. It provides the capital, the credit, and is responsible for construction and property management. By this standard, this debt should logically be consolidated onto its own balance sheet.

But it left out 20%.

Blue Owl's OSNL fund and its co-investors, through Beignet Pledgor and the other four holding companies, wholly own Beignet Investor. Beignet Investor then holds 80% of the joint venture. Meta holds the remaining 20%.

80 and 20 are the recurring numbers in this story.

Meta's explanation in its financial reports is that it does not have the power to direct the activities that most significantly impact the joint venture's economic performance, and therefore it is not the primary beneficiary and does not consolidate. Since the joint venture is not on Meta's books, the $27.3 billion in debt naturally doesn't appear either.

The debt hasn't vanished. It has just been moved to a different location.

Converting Mortgages to Leases: The "Debt Repayment Art" of Silicon Valley Giants

Debt not being on Meta's books doesn't mean Meta doesn't have to pay.

Meta has a leasing company called Pelican Leap. It signed a four-year lease with Laidley. Starting in 2029, Pelican Leap pays rent to Laidley monthly. The money flows from Laidley to the joint venture, then to Beignet Investor, and is used to repay principal and interest to bond investors.

Ultimately, the rent still comes out of Meta's pocket.

The $27.3 billion bond carries a coupon of 6.581%, maturing in May 2049, with a fully amortizing structure. Unlike a typical corporate bond that repays the principal in one lump sum in 2049, this one amortizes gradually over 24 years.

It looks more like a mortgage.

The total rent Meta pays in the first four years is $12.3 billion, averaging $3.08 billion annually. This amount just covers the year's principal and interest payments, with the remaining 12% going to the equity investors.

The real twist starts with the lease term.

The bond lasts 24 years. The initial lease is only 4 years. Starting in 2029, there are renewal options attached, extending up to 20 years maximum. This means Meta could theoretically choose not to renew the lease in 2033 and simply walk away.

So what happens to the over $20 billion in debt that hasn't been repaid yet?

The answer is hidden on another page of the lease agreement. Besides the monthly rent, Meta provided a residual value guarantee, with an upper limit of approximately $28 billion—slightly higher than the debt amount itself. This amount decreases over time. If Meta doesn't renew the lease, Meta must make up the difference if the campus's value falls below this threshold.

Putting $28 billion and $27.3 billion together makes it hard not to connect the two things.

Therefore, what truly backs these bonds is not just the building or the machines inside.

It relies on Meta's credit.

This also explains the rating. S&P rated this bond A+, while Meta itself is rated AA−. The rating agency didn't price it based on a yet-to-be-operational building but simply downgraded Meta's own credit by one notch.

Meta provides the money, the credit, handles operations, and is the sole tenant, yet states in its financial report that it is not the primary beneficiary.

This clean balance sheet doesn't come cheap. If Meta issued debt of the same maturity in the public market, the cost would be around 5.5%. Through this structure, the cost rises to 6.581%. For the same amount of money, Meta pays nearly $300 million more in interest per year.

What Meta buys with this extra cost is clearly more than just a building.

In July 2026, a second similar project appeared, named "Sopaipilla," another fried pastry from the southwestern United States. The project is located in El Paso, Texas, with a bond issuance starting at $12 billion, also with an 80/20 split. The difference is that the 80% holder changed from Blue Owl to BlackRock.

Meta has used internal codenames for its next-generation large model—one was Avocado, another Mango. The financial side of naming clearly has more appetite.

Meta's structure is the most intricate, but not the only approach.

Microsoft doesn't use shell companies, nor has it seen a massive increase in book debt. Over the past two years, its total debt decreased from $44.9 billion to $40.3 billion. However, during the same period, finance lease liabilities more than doubled from $27.1 billion to $62.9 billion, over $20 billion more than its total debt.

The $62.9 billion is indeed on Microsoft's balance sheet, but it's not under the "debt" line. It's split into "Other current liabilities" and "Other long-term liabilities." From the most prominent row, it looks very quiet.

The meaning of a finance lease is quite straightforward. In name, it's a lease