Morgan Stanley has emerged as the chief architect of the financing structures underpinning the AI boom, devising new debt and equity models that funnel tens of billions into data‑centre build‑outs. The bank’s portfolio includes a $3.2bn bond for data‑centre developer Tera Wulf backed by Google, a $27bn debt package for Meta’s Hyperion tie‑up, and a $35bn chip financing for Broadcom.

The surge in AI dealmaking helped it edge past Goldman Sachs in the first half of the year, with capital‑markets fees rising to $2.3bn. AI is reshaping markets by packaging long‑term computing contracts and hyperscaler balance sheets into securities that can be sold to mainstream investors, dramatically expanding the capital pool.

Key to low rates is tying，则 hyperscalers—Google, Amazon, Meta and Microsoft—into guarantees that roughly halve financing costs. Morgan Stanley’s hybrid bonds combine project‑finance protections with a broad investor base; the Tera Wulf deal lifted the yield to 7.75% and attracted insurers and pension funds.

Beyond data‑centres, the bank has pioneered GPU financing, arranging a $3.1bn loan for Core Weave to purchase Nvidia GPUs, illustrating a model that relies more on long‑term take‑or‑pay contracts than on chip valuations. The shift away from hyperscalers to AI labs introduces tighter credit risk.