Big Tech's Off-Balance-Sheet Debt Isn't Fraud, It's That Nobody Reads the Footnotes
The so-called $1.65 trillion in hidden debt is not Enron-style fraud but long-term leases and purchase commitments for data centers not yet in service — legal, but buried in the footnotes; the real danger is AI giants telling their story through mutual financing and adjusted earnings.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
$1.65 trillion isn't hidden debt, it's stuff not yet delivered
The $1.65 trillion in off-balance-sheet debt reported by Nikkei is mostly long-term GPU purchase agreements and leases on data centers not yet in service. Under standard accounting rules, when goods haven't been delivered or buildings aren't running, you don't have to record a balance-sheet liability — you just disclose it in the footnotes, just as signing a two-year phone contract doesn't mean booking a $1,200 liability that day. Meta added $233 billion in commitments last quarter, of which $96 billion in leases will move onto the balance sheet once the data centers come into service. So this isn't Enron-style fraud, it's a timing gap.
— Patrick BoyleWilling to borrow but not to sell stock means management really believes
If management believed it could turn $1 into $5, it would tend to borrow rather than sell equity, so as to keep the excess profit for itself; only when uncertain about the outlook would it bring in outside shareholders. Alphabet completed nearly $85 billion in equity financing in June, the largest ever, with Berkshire investing $10 billion and buying in at roughly a 6% discount — a company that normally doesn't pay a premium for a story. The signal isn't the method of financing but the record scale of simultaneously using debt, equity and convertibles: the company thinks there's something behind the massive spending worth fighting for.
— Patrick BoyleIgnoring depreciation assumes equipment never wears out
Tech giants prefer to talk EBITDA, but Munger advised reading EBITDA as BS earnings. Depreciation is reverse float: cash goes out first, the expense shows up later; ignoring depreciation assumes equipment never wears out. In the latest earnings season, the four big hyperscalers together generated only $7 billion in free cash flow, and Alphabet went cash-flow negative for the first time since going public. Stock-based compensation shouldn't simply be added back either — it's barter: issuing shares directly is as much a cost as selling shares to pay salaries. To offset dilution, companies have to buy back on a fixed schedule and can't wait for a low price.
— Patrick BoyleWhen they default you lose the customer and the money you lent
Nvidia is pushing more than $750 billion in AI deals: $250 billion of support for OpenAI leases, then financing for $350 billion of chip purchases, plus $5 billion invested in Ilya Sutskever's startup; Google is providing $35 billion in lease support for Anthropic; SoftBank has committed $65 billion to OpenAI and borrowed $40 billion in bridge loans. The giants are investing in each other and buying from each other. It's like carmakers lending customers money to buy cars, but it's also old-fashioned vendor financing. The risk is that on default you lose both the customer and the money you lent the customer.
— Patrick BoyleEveryone is priced as a winner, which is mathematically impossible
Aswath Damodaran and Bradford Cornell call it the big market delusion: every investor treats the company they're betting on as the future winner, but add up the money each is expected to earn and the number exceeds the market itself — mathematically impossible. The Economist estimates about $900 billion of AI buildout spending this year, more than $400 billion of it borrowed; earning that back would require roughly $2.5 trillion a year in AI revenue, more than the entire global tech industry generates today. Actual adoption is modest: US executives use AI about 100 minutes a week on average, and median enterprise spend per person is just $10.66 a month.
— Patrick Boyle93% of SpaceX's market space comes from AI, not rockets
In its IPO prospectus, SpaceX attributed $26.5 trillion of its $28.5 trillion TAM — 93% — to AI/Grok, leaving about $2 trillion for actual businesses like rockets and satellites. Analysts then put out an $800 target price, implying a market cap above $10 trillion, while the company's revenue last year was under $19 billion. The reason: SpaceX disclosed roughly $235 billion in spending commitments through 2030, and IPO proceeds cover only a small part — the roughly $170 billion gap has to be filled by repeatedly issuing stock and debt over many years, which is exactly Wall Street's underwriting feast. Analysts at 18 underwriting banks published bullish calls at almost the same time, and only one independent firm gave a sell rating.
— Patrick BoyleBurying debt on page 83 is a rational bet
Public doesn't mean usable. Robert Bloomfield's incomplete revelation hypothesis holds that the harder information is to extract, the less fully it gets priced; putting debt on page 83 of a 200-page document, scattered across four footnotes, is choosing that most people won't look. Richard Sloan found in 1996 that the market treats cash earnings and accruals the same, even though accruals are less persistent. Limits of arbitrage then explain that someone who reads the footnotes can short, but may be blown up before the market corrects, so smart money often stays away from hype stocks. Companies burying inconvenient numbers in the footnotes is a rational bet: it disclosed them, it's legal, and it often works.
— Patrick BoyleThe AI winner is the person paying $20 a month
This round of "hidden debt" panic isn't the discovery of the next Enron; it's people finally reading the footnotes and realizing the true scale of spending. AI isn't fraud, it's obviously useful, but the biggest returns don't go to the giants burning hundreds of billions — they go to the founders using AI to start one-person businesses at about $20 a month. For the renter it's a great deal; for the side that spent enormous sums building and is still waiting for $20 monthly fees to accumulate into $2.5 trillion in annual revenue, it's terrible. As Jamie Dimon said: AI may pay off like the internet, but not on the timetable you expect.
— Patrick BoyleIn their own words · checked verbatim
So it isn't Enron-like fraud, it's camouflage which only really works on people who aren't paying much attention.
Patrick Boyle7:21
every time you read the word EBITDA you should replace it in your head with BS earnings.
Patrick Boyle8:22
If options aren't a form of compensation, what are they? If compensation isn't an expense, what is it? And if expenses shouldn't go into the calculation of earnings, where in the world should they go?
Patrick Boyle9:29
The story is doing all the work and nobody's minding the numbers.
Patrick Boyle15:46
The largest capital investment in the history of the species is being justified by an hour and a half per executive per week, so somewhere between lunch and a drive home.
Patrick Boyle16:51
Will AI pay off? Probably the way the internet did. Will it pay off the way you expect on the timeline you expect? Definitely not.
Patrick Boyle31:37
Figures
| Reported off-balance-sheet debt of the five big tech companies | $1.65 trillion | 0:00 |
| Alphabet equity financing (largest ever) | nearly $85 billion | 5:12 |
| Meta's new AI commitments in a single quarter, and the lease portion | $233 billion, of which $96 billion is leases | 6:17 |
| Total AI-related deals Nvidia is pushing | more than $750 billion | 11:29 |
| The Economist's estimate of total AI buildout spending this year | about $900 billion, of which more than $400 billion is borrowed | 16:51 |
| The Economist's calculated AI annual revenue threshold | about $2.5 trillion | 16:51 |
| Ramp's median enterprise AI spend per person per month | $10.66 per person per month | 16:51 |
| Bank of England survey: executives saying AI hasn't raised company productivity in three years | 9 out of 10 | 17:57 |
| Market cap implied by SpaceX IPO underwriter target price | more than $10 trillion ($800 per share) | 20:06 |
| SpaceX's disclosed 2030 spending commitments and financing gap | about $235 billion, with a gap of about $170 billion | 21:09 |
Glossary
- off-balance-sheet vehicle
- A special-purpose entity that keeps debt off a company's books, though it still must be disclosed in the footnotes.
- vendor financing
- A seller lending to a buyer so the buyer can purchase its product, locking in the customer while taking on default risk.
- big market delusion
- Investors price every company in a new sector as the eventual winner, pushing total valuation beyond the truly addressable market.
- incomplete revelation hypothesis
- The harder information is to extract, the less fully the market prices it; facts buried in hard-to-read footnotes get undervalued.
- limits of arbitrage
- Even when mispricing is spotted, shorting can blow up because the market stays irrational, keeping smart money out.
How to listen
Investors who care whether AI capex is forming a bubble, and founders and analysts who need to read tech company financials.
Listeners already familiar with the Enron case can skip the first 3 minutes of background.