Everyone Can Count the Trillions. Nobody Can Tell You Who Gets Paid Back.
Forbes toured the trillions. We answer what the scale tours never do: the $4.2 trillion revenue hole, the circular money problem, the five layers of the buildout ranked by who gets paid first, what breaks first, and the trade, with our tracked positions as receipts.
Everyone has learned to count the trillions. Forbes toured the spectacle this week: estimates for the American AI data center buildout now run from $2.8 trillion by 2030 to $10.3 trillion by 2032, and Anthropic's leaked prospectus, per that reporting, pencils in $518 billion of infrastructure spending from one company over the next decade. The numbers are so large they have stopped meaning anything. A trillion here, a trillion there; eventually the reader's eyes glaze and the point is lost.
Counting is the easy part. The questions that actually matter for your money are the ones the scale tours never answer: who is paying, who gets paid, in what order, and what breaks first if the revenue arrives late. This briefing answers them.
Start with what is verifiably happening right now, because the present is dramatic enough without the 2032 projections. The four hyperscalers, Microsoft, Alphabet, Amazon, and Meta, are on track to spend roughly $725 billion of capital expenditure in 2026, up around 77 percent from last year, a figure that now exceeds 100 percent of their combined cloud revenue. Data center construction spending has been compounding at a 149 percent annualized pace since March. Data centers have become the single largest segment of American office construction, which is a sentence that would have read as satire three years ago.
And here is the fact that reframes all of it, the one we flagged for members in September: the buildout stopped paying for itself. In fiscal 2024, debt funded about 9 percent of hyperscaler capex. By mid-2026 it funded 32 percent. Alphabet's debt load has jumped to roughly $100 billion alongside an $80 billion-plus equity raise. Meta's free cash flow has thinned to near zero with buybacks on hold. And Oracle, the most aggressively levered builder of them all, has been cut to BBB-, one notch above junk, while its stock has fallen 56 percent from its 52-week high even as Nvidia trades at records.
Read that last sentence again, because it is the whole story in miniature. The market is no longer pricing "AI" as one trade. It is grading balance sheets, layer by layer, and it has already started failing some of them. In 18 months, this went from the most self-funded capital boom in corporate history to a bet that credit markets stay open on friendly terms indefinitely.
What follows, for members: the $4.2 trillion hole at the center of the math, the circular money problem nobody wants to name, the five-layer map of who actually keeps the margin, what breaks first and the exact signposts to watch this month, and the trade, with our tracked positions as receipts.
Get the next briefing free in your inbox: Join AlphaBriefing
Behind the paywall: Bain's $6 trillion revenue requirement and the $4.2 trillion annual gap, why 98 percent of American households paying nothing for AI is the number that decides everything, the vendor-financing loops that make reported demand look stronger than it is, the five layers of the money map ranked by who gets paid first and who gets hurt first, the three ways this resolves, and the names we already hold on the tracked Watchlist, with live performance since entry.
AlphaBriefing Paid gets every intelligence brief, the weekly Top 10, and the live tracked Watchlist.