Banks Rebalance Data Center Lending as AI Infrastructure Boom Tests Risk Limits

A planned secondary sale of a Hong Kong data center loan is drawing attention to a broader trend in global banking, as financial institutions seek to manage their growing exposure to artificial intelligence infrastructure projects.

According to people familiar with the matter, Crédit Agricole CIB is seeking to sell approximately HK$150 million of a HK$1.6 billion loan extended to ESR Group for a data center project in Hong Kong. The bank has reportedly approached other lenders after reaching its internal lending limit for data center-related financing, illustrating how rapidly expanding demand for AI infrastructure is reshaping banks' risk management strategies.

The financing was originally provided in 2023 to support the conversion of a cold storage facility in Kwai Chung into a modern data center. While it remains unclear how much of the original loan was held by Crédit Agricole CIB, the transaction reflects increasing efforts by banks to rebalance their portfolios rather than reduce support for the sector.

The move mirrors a broader trend across global financial markets. As investment in AI infrastructure accelerates, banks and institutional investors are exploring new ways to distribute risk. Morgan Stanley has reportedly considered transferring part of its data center loan exposure through significant risk transfer structures, while asset manager Voya Financial has limited investments tied to large technology companies with substantial long-term AI infrastructure commitments.

Demand for financing continues to surge alongside the global expansion of artificial intelligence. In Asia-Pacific, the AI boom has driven a wave of large financing transactions, including discussions surrounding a potential $7 billion loan for DayOne Data Centers, which would become the largest data center financing completed by an Asian company if finalized.

Credit rating agency Moody's estimates that at least $3 trillion will be invested globally in data center-related projects over the next five years, with debt expected to finance a significant portion of that spending. Major technology companies, including Meta Platforms and Alphabet, have substantially increased borrowing to fund AI infrastructure, reflecting the industry's aggressive expansion plans.

The rapid pace of investment has also prompted questions about financial sustainability. While demand for AI computing capacity remains strong, investors have become increasingly focused on whether massive capital expenditures can generate consistent long-term returns. The scale of borrowing has raised concerns that infrastructure spending may be outpacing proven commercial demand for AI services.

Bloomberg data illustrates the speed of the financing boom. Companies have issued at least $334.5 billion in bonds and loans to support AI infrastructure projects so far this year, nearly double the $185.5 billion raised during all of 2025.

As AI infrastructure becomes one of the fastest-growing areas of global capital investment, banks are increasingly balancing two competing priorities: supporting one of the world's most attractive growth sectors while ensuring that concentrated exposure to data centers remains within prudent risk limits. The Hong Kong loan sale demonstrates that risk redistribution, rather than reduced lending, is becoming an important feature of the next phase of AI infrastructure financing.