I'm LongbridgeAI, I can summarize articles.Morgan Stanley is transforming AI data center financing into a new asset class. By leveraging an innovative model that securitizes the credit of tech giants like Alphabet alongside long-term computing power contracts, its capital markets fees surged by over 60% to $2.3 billion in the first half of the year, propelling it to second place globally. However, as the financing chain extends from data centers to chips, credit risks are quietly accumulating—whether this $10 trillion feast can sustain itself ultimately depends on whether AI demand materializes
Morgan Stanley is transforming AI infrastructure financing into a new asset class in the capital markets, thereby surpassing Goldman Sachs in investment banking business.
According to a report by the UK's Financial Times on Tuesday, the Wall Street bank has become the primary architect of AI data center financing structures, opening up new funding channels for this unprecedented wave of capital expenditure by packaging the balance sheets of tech giants with long-term computing power contracts into securities sellable to mainstream investors.
According to LSEG data, Morgan Stanley's debt and equity capital markets fees increased by more than 60% year-on-year to $2.3 billion in the first half of this year, jumping from fourth to second place in the global ranking of capital markets fees, trailing only JPMorgan Chase.
This model is reshaping the landscape of capital markets. As the scale of AI infrastructure financing leaps from the previous range of $1 billion to $5 billion to $10 billion to $20 billion or even higher, more insurance companies, asset management firms, and pension funds are being drawn into this sector, deepening the tie between the financial system and AI computing power demand.
New Financing Template: Leveraging Tech Giants' Credit to Access Cheap Capital
The TeraWulf bond, spearheaded by William Graham, Co-Head of Leveraged Finance at Morgan Stanley, has become the industry template for AI infrastructure financing.
This structure combines bonds sellable to a broad range of investors with protective covenants typical of project loans, creating a hybrid instrument backed by a guarantee from Alphabet. To further reassure investors, Morgan Stanley introduced a "lockbox" mechanism, directing lease payments directly to bondholders and adding extra collateral. Ultimately, TeraWulf successfully raised $3.2 billion at a yield of 7.75%.
Patrick Fleury, CFO of TeraWulf, stated that this innovative structure allowed the company to bypass the cumbersome stage-by-stage review process of traditional project finance loans while borrowing at a sufficiently low cost to make the business model economically viable. "In effect, we are financing by leveraging the credit strength of Alphabet's balance sheet," he said.
The core logic is that when hyperscale cloud service providers like Alphabet, Amazon, Meta, or Microsoft guarantee data center leases, financing costs are roughly halved. Mo Assomull, Co-Head of Investment Banking at Morgan Stanley, described this gap as a choice between "mid-to-high single digits" and "double that."
From Data Centers to Chips: Financing Boundaries Continue to Expand
Morgan Stanley has not stopped at data centers themselves but has extended this financing logic to the chip level.
In May this year, Morgan Stanley, together with MUFG Bank, arranged a $3.1 billion loan for emerging cloud service provider CoreWeave to purchase and deploy NVIDIA GPUs. This was the first GPU financing completed in the form of a broadly syndicated term loan, bringing a wider range of capital into the chip financing sector. The loan attracted nearly $20 billion in investor subscription demand.
Under this structure, data center buildings and chips are financed separately: the former is supported by leases, while the latter is collateralized by long-term "take-or-pay" contracts. William Graham used a car analogy: "Chips are Ferraris... they need a place to park, so you need a data center to house the chip."
Differences in credit quality are directly reflected in pricing. In March this year, Morgan Stanley assisted CoreWeave in completing an $8.5 billion chip loan, backed by hyperscale cloud service provider contracts, priced at 225 basis points over the benchmark rate; the loan in May was backed by two AI labs with relatively weaker credit profiles, resulting in a pricing spread widening to 450 basis points.
Risk Accumulation: The Credit Chain Stretches Further
As financing structures extend beyond hyperscale cloud service providers, potential risks are quietly accumulating.
Raj Joshi, Senior Vice President at Moody's Ratings, stated that he is closely monitoring the financial health of Anthropic and OpenAI. "This is a capital expenditure investment cycle of massive scale, with no comparable precedent in history," he said. "There is no existing playbook."
Some competitor bankers indicated that they are reluctant to become the top player in this field given the controversies sparked by data centers in communities across the United States. Jeremy Barnum, CFO of JPMorgan Chase, also warned this week that after reviewing the loan terms of some data center financing deals, the bank concluded that "we will not participate."
Morgan Stanley itself expects that AI infrastructure construction will consume $10 trillion in expenditures over the next few years. William Graham is more optimistic about the market's prospects, predicting that AI infrastructure bonds will eventually account for the majority share of annual new non-investment grade debt supply, characterizing it as "the first entirely new segment to emerge in capital markets in the past 20 years." However, whether this financing feast can continue ultimately depends on whether AI computing power demand fulfills its promises.