Wall Street fearlessly bets big on AI amidst bubble controversy.

In the wave of data center construction driven by artificial intelligence, Wall Street is investing massive amounts of money at an unprecedented pace. Blue Owl, which originally started off by providing loans to medium-sized enterprises, has now turned to financing the construction of large data centers costing hundreds of billions of dollars for tech giants like Meta and Oracle, reflecting the explosive growth in demand for AI infrastructure in the United States. Fund management institutions have accumulated significant capital in recent years and have long lacked large investment targets, and now AI hardware construction has become an ideal outlet. The founder of Blue Owl stated that global AI capital expenditures in the next few years could exceed trillions of dollars. Despite concerns about a potential bubble caused by the significant fluctuations in tech stocks and bonds last week, the fear of "falling behind the curve" has led most institutions to continue increasing their investments. Blue Owl had previously arranged a $14 billion financing for Oracle and OpenAI to build data centers in Texas. Last month, the company raised about $30 billion for Meta to build data centers in Louisiana, including $3 billion in equity and debt, and obtained a rare "equity-like bond guarantee clause," innovating the structure to meet the needs of AI financing. AI infrastructure financing is rapidly spreading to every corner of Wall Street. While tech giants have historically relied on cash reserves for initial investments, they are now turning to debt and private financing as their budgets expand rapidly. Investors are showing strong interest in data center debt, as newly issued bonds are recording substantial gains upon opening. In the Meta data center project, Pimco bought $18 billion in bonds, and within a few days of trading, the paper profit soared to $2 billion. However, long-term returns still contain uncertainties. Morgan Stanley predicts that tech giants will invest nearly $3 trillion in AI by 2028, but cash flows can only cover half of that amount. Some institutions warn of overheated market sentiment, but at the same time, several banks are rapidly forming AI financing teams to participate in the competition. This year, Morgan Stanley arranged approximately $75 billion in large data center-related financing in one week. Risks are also accumulating. Oracle, which provides computing power to OpenAI, has a high level of debt, and its rating agencies are close to downgrading its bonds to junk status. Its stock price has recently dropped by 32%, and its bonds have also fallen by about 7%. In addition, AI chip iterations are fast, and if there are changes in future technological directions, some financing projects may face uncertain asset values. As more institutions get involved, the scale continues to break records. Last year, JPMorgan Chase received a proposal from a client to finance a data center consuming 1 gigawatt of electricity, which expanded to two projects within months, ultimately leading to a banking consortium agreeing to provide up to $38 billion in a five-year loan package. Due to its size, over 30 banks were invited to jointly sell the bonds. Although some industry insiders emphasize that the current investments are not speculative, if there is a market correction in the AI sector, the impact will be felt by banks, pension funds, mutual funds, and bond investors, as the related debts have been fragmented and widely distributed throughout the financial system. Currently, institutions are focused on strong leases and the credit endorsement of large tech companies, but the realization of future demand and the validation of business models will still require time to be tested.