HSBC Weighs Insurer Cover for Data Center Loans as AI Finance Expands

HSBC Holdings Plc is exploring whether insurers will cover part of the credit risk on project loans, including financing tied to artificial intelligence infrastructure. The discussions are preliminary, and any coverage remains uncertain. The move offers a view into how lenders are managing exposure as data center financing expands and insurance capacity faces competing demand.

People familiar with the talks said HSBC has approached insurers about a series of project finance deals. The bank had earlier assessed insurer interest in covering a portfolio focused exclusively on data centers. A representative for the London-based lender described the work as part of routine portfolio management and said its data center exposure is small.

How credit insurance shifts risk

Under credit risk insurance, a lender pays premiums in exchange for coverage of an agreed share of losses on specified loans. The arrangement can transfer some risk away from the bank while leaving it exposed to losses outside the covered portion and to the terms of the policy.

That structure differs from a significant risk transfer, or SRT. In an SRT, investors typically provide liquid assets upfront to absorb losses on a portfolio’s first-loss tranche and receive coupon payments in return. Both approaches can help banks manage credit exposure and release capital for other lending, but the risk transfer works through different parties and payment arrangements.

For insurers, the decision rests on the loans and the amount of risk they would be taking on. For banks, insurance can add a layer of protection as they compete to finance projects. Whether it makes a material difference depends on the coverage available, its cost and the losses the contract actually covers. HSBC’s talks have not yet produced a deal, and the people familiar with them said they may not do so.

Borrowing needs are growing with AI infrastructure

AI infrastructure is expected to require trillions of dollars in spending over the coming years. Companies building that infrastructure are likely to seek financing through several channels, including asset-backed securities, private credit and syndicated loans. The source article’s assessment is that company cash flows and equity alone will not meet the expected funding needs.

Data centers sit at the intersection of real estate, technology and credit markets. Financing them through project loans can concentrate lender exposure in assets whose repayment depends on a project’s ability to generate cash flow. As borrowing grows, banks and insurers will have to assess how much capacity they are willing to commit and how they price the risks.

There are already signs that markets are having difficulty absorbing the volume of AI-related credit. Yields on bonds and loans linked to the infrastructure buildout are rising, while some insurers have indicated that their capacity is close to exhausted. Those conditions could make new protection harder or more expensive to secure, even as borrowers seek more debt financing.

HSBC’s position and peer activity

HSBC has also been working on an SRT tied to a portfolio of more than $1 billion of data center-linked loans. The bank is assessing ways to hedge its exposure despite having a relatively limited role in this segment of lending.

Chief Financial Officer Pam Kaur said during a call this week that HSBC is “not a big player at all in the US in data center lending, same in Asia,” adding that the business “is not a significant part of our growth story.” Her remarks frame the effort as management of a contained portfolio exposure rather than evidence that data center lending is central to HSBC’s expansion.

Other large banks are also seeking ways to manage exposure to the sector. Barclays Plc is looking for insurance on more than $10 billion of AI infrastructure credit. Deutsche Bank AG has been working on an SRT linked to about $2 billion in project finance debt, including data center loans.

The scale of those efforts points to a financing market where banks are trying to accommodate large infrastructure needs while controlling the amount of risk they retain. Yet insurance and SRTs do not remove the underlying credit exposure from the system; they redistribute it among lenders, insurers and investors under specific contract terms. The next signal to watch is whether preliminary insurer discussions translate into coverage, and on what terms, as AI-related borrowing grows.