Interview on ‪@Bloomberg-News‬ on September 8th, 2026

Hyperscalers are financing the AI companies without any stated limit, and those companies are nowhere near profitability, so for now the only financial return flowing back up the chain is an accounting one. Alphabet booked $98.0 billion of other income in the second quarter, "primarily the result of net unrealized gains on our equity securities," against $40.8 billion of operating income, in the same quarter it posted negative free cash flow for the first time in its history and raised 2026 capex guidance to $195 to 205 billion. Amazon booked $53.4 billion of other income "primarily from our investments in Anthropic" against $27.5 billion of operating income. Magnificent Seven earnings growth of 118.5% in the quarter falls to 43.2% once those two marks are removed, on published consensus data. The mechanism is simple: the hyperscaler funds the round, the round lifts the valuation, and the valuation is booked as income, none of which is cash and none of which repeats without a higher round next time. Meanwhile the company being marked has a $965 billion valuation, a disclosed run-rate of $47 billion and no published loss figure, and the other one's only public profit and loss is a leaked document showing a $20.9 billion operating loss for 2025. The second leg is that the build is increasingly debt-financed, with hyperscalers having issued $219 billion of investment-grade bonds this year to 20 August, roughly nine percent of all dollar investment-grade supply, and that duration lands on a long end already carrying a $40 trillion gross federal debt and a Treasury that is doubling its long-dated buybacks while borrowing $739 billion net this quarter. The 30-year closed at 5.24% on 4 September after an August high that was widely reported as the highest since 2007, and although the Dallas Fed puts AI duration supply at about an eighth of Treasury's, so the fiscal borrower remains the first-order pressure, the two are now competing for the same marginal buyer. Which brings me to the question I keep turning over: if the funding is unlimited and AI is the future, why are these companies in such a hurry to list? A business that expects to be profitable in three years, with adoption far more advanced, would fetch a better price and face fewer questions by waiting. The rush suggests the people financing the rounds want a public mark, and a public exit, before the private round-to-round remeasurement has to meet a buyer who is not also a supplier and a shareholder. I wonder.