Amid the AI-driven boom in data center construction, Wall Street is pouring unprecedented capital at record speed. Blue Owl, once known for lending to mid-sized businesses, now finances multi-billion-dollar data centers for tech giants like Meta and Oracle, reflecting explosive demand for AI infrastructure in the U.S.

Investment funds, flush with capital but lacking large-scale opportunities, see AI hardware as an ideal outlet. Blue Owl’s founders project global AI capital expenditures could surpass $1 trillion in coming years. Despite recent volatility in tech stocks and bonds sparking bubble fears, most institutions continue doubling down, driven by fear of "missing the cycle."

Blue Owl recently arranged $14 billion in financing for Oracle and OpenAI’s Texas data center. Last month, it raised ~$30 billion for Meta’s Louisiana facility—including $3 billion in equity/debt—with innovative "debt-like equity guarantees" tailored for AI projects.

AI infrastructure financing is spreading rapidly across Wall Street. Tech giants, traditionally reliant on cash reserves, now turn to debt and private funding as budgets balloon. Investor appetite is robust: Meta’s $18 billion bond offering by Pimco saw paper gains hit $2 billion within days.

Yet long-term returns remain uncertain. Morgan Stanley estimates tech firms’ AI investments could near $3 trillion by 2028, with cash flows covering only half. While some warn of overheating, banks are racing to form dedicated AI financing teams. Morgan Stanley alone structured ~$75 billion in data center deals within a week this year.

Risks loom. Oracle, which powers OpenAI, faces potential junk-bond downgrades as debt mounts. Its shares fell 32%, with bonds down 7%. Rapid AI chip evolution also threatens asset valuations if technology shifts.

These projects mirror leveraged buyouts—Meta’s Hyperion used $27 billion debt at 6.58% interest. Scale keeps expanding: JPMorgan initially planned a 1-gigawatt data center loan, later expanded to two facilities, culminating in a $38 billion five-year syndicated loan requiring 30+ banks for distribution.

In Abilene, Texas, Blue Owl is building a 4-million-square-foot campus housing 500,000 Nvidia chips, consuming 1.2 gigawatts, with ~$10 billion in loans. AI’s funding needs now dwarf the shale industry’s peak.

Though insiders deny speculation, an AI downturn could ripple across banks, pensions, and bondholders given widespread debt securitization. While strong leases and tech-company guarantees reassure investors, validating demand and business models will take time.