--- title: "Morgan Stanley puts the AI financing gap at $1.5 trillion, and nobody can see where it sits" description: "Nearly $3 trillion of global AI infrastructure investment through 2028, with about half of it needing outside money. A Fortune commentary argues the Fed does not know who is providing it. Our own reporting says the same thing from four directions." category: "Economy" category_url: https://boursel.com/category/economy author: "Sofia Marchetti" published: 2026-08-25T10:37:49.000Z updated: 2026-08-25T10:37:49.000Z canonical: https://boursel.com/article/morgan-stanley-puts-the-ai-financing-gap-at-1-5-trillion-dollars-and-nobody-can tags: ["ai capex", "private credit", "federal reserve", "leverage", "financial stability"] --- # Morgan Stanley puts the AI financing gap at $1.5 trillion, and nobody can see where it sits Nearly $3 trillion of global AI infrastructure investment through 2028, with about half of it needing outside money. A Fortune commentary argues the Fed does not know who is providing it. Our own reporting says the same thing from four directions. Morgan Stanley projects [nearly $3 trillion of global AI-related infrastructure investment through 2028, with an estimated $1.5 trillion external financing gap](https://fortune.com/2026/08/25/the-fed-doesnt-know-whos-financing-the-3-trillion-ai-boom/). That second number is the one worth holding: roughly half the spending is not expected to come from the cash flows of the companies doing it. The argument built around it, in a Fortune commentary, is that the Federal Reserve does not understand how the investment is being financed: the growing role of private markets, the increasingly complex links among borrowers and intermediaries, and where leverage, maturity risk and ultimate exposures actually sit. That is a commentary rather than a Fed statement, and we report it as an argument. It happens to match what our own reporting has run into repeatedly, from four different directions. ## The four places we have hit the same wall **Commitments in the notes.** A Wall Street Journal analysis put nine companies at [roughly $3 trillion of AI commitments](/nine-companies-have-signed-3-trillion-dollars-of-ai-commitments-that-sit-off-the), about $1.9 trillion of purchase obligations and $1.2 trillion of leases signed but not started. All disclosed, none on the balance sheet, and we wrote a separate piece on [how to find them](/how-to-read-the-commitments-note-the-part-of-a-filing-where-3-trillion-dollars-w) because most readers never open that note. **The bond market.** US technology companies have issued [at least $220 billion of debt this year against $12.5 billion in the same stretch of 2025](/the-bond-market-has-started-charging-more-for-ai-even-to-the-safest-borrowers), and the price has moved: technology now trades wider than the investment grade market it used to trade through. **Equity instead of debt.** Alibaba raised [$10.2 billion of equity](/alibaba-is-asking-shareholders-for-10-2-billion-dollars-to-keep-spending-on-ai) rather than borrowing, which we used to set out [how the two instruments fail differently](/big-tech-borrowed-for-the-ai-build-alibaba-just-sold-shares-instead). **Borrowed money in the shares.** The SEC is [subpoenaing the banks that funded an AI hedge fund](/the-sec-is-asking-the-banks-who-funded-an-ai-hedge-fund-that-nearly-blew-up) that took margin calls in July and was forced to liquidate. That is leverage against AI equities rather than AI assets, and it is disclosed to prime brokers rather than to markets. Four channels, four different disclosure regimes, and no single place where the total is visible. That is the gap the commentary is describing, and it is not hypothetical. ## Why a central bank would care Not because AI investment is bad. Fed chair Kevin Warsh has made the case that AI could raise productivity and productive capacity, and if that is right the investment is exactly what an economy should be doing. The concern is about the transmission of policy and the location of risk. A central bank sets one interest rate and relies on knowing roughly how it reaches borrowers. When a large and growing share of credit runs through private funds rather than banks, the map gets worse: the intermediaries are not reporting to the same regulators, the leverage is embedded in fund structures rather than on bank balance sheets, and the maturity mismatch is harder to observe. The commentary cites its author's own research with Sergey Sarkisyan arguing that credit spreads carry policy-relevant information about financing distortions that inflation and the output gap miss, which is an argument for watching prices in credit markets when the quantities are not observable. ## The Greenspan comparison, and its limit The piece reaches for the 1990s, when Alan Greenspan entertained the possibility that faster productivity growth had raised the economy's speed limit and largely resisted further rate increases, and unemployment fell without inflation following. It is the right analogy for the productivity question and the wrong one for the financing question. The late-1990s technology build was funded predominantly through public equity markets, where the excess was visible in real time, priced daily and reported quarterly. When it unwound, everyone could see what they owned. The current build is funded through purchase commitments in footnotes, private credit funds, vendor financing and prime brokerage. The productivity payoff may well be larger this time. The visibility is worse. ## What would close the gap Nothing that exists today. Private funds report to their investors, prime brokers to their regulators, and issuers to the SEC on quarterly cycles, and none of those aggregate. The practical version for a reader is narrower and more useful: when a company describes its AI spending, the question is not how much, it is who is funding it and on what terms. That is answerable from filings for public companies, and it is the part the headline number never contains.