Morgan Stanley Pitches Data Center Clients on Leveraged Loans

The debt market that usually backs leveraged buyouts is Morgan Stanley's new pitch to AI data center clients. The Information reports that the bank, described as one of the most active lenders to data center developers, is pitching some of those same clients on tapping the leveraged loan market, a corner of credit that has historically served private equity deals rather than infrastructure developers.

The numbers behind the pitch are large. Morgan Stanley estimates about $20 billion of AI-related deals will reach leveraged finance markets in 2026, while JPMorgan projects $150 billion over the next five years. Those figures are still a fraction of the broader financing wave: the hyperscalers, tech giants such as Microsoft and Meta and their joint ventures, are expected to issue $250 billion to $300 billion in debt in 2026 alone to build out computing capacity that will require gigawatts of power. The full spectrum of credit markets, from secured to unsecured to structured and securitized instruments, is now being drawn into AI-related infrastructure financing.

Bringing data center developers into the leveraged loan market alongside project finance and bonds would give those developers access to a broader and more liquid pool of institutional capital. Leveraged loans have established buyer infrastructure, particularly in CLOs and credit funds, that could absorb new issuance volumes at scale if the asset class takes hold.

There is a tension worth noting. Morgan Stanley has separately been exploring significant risk transfer deals to offload portions of its existing data center loan exposure, according to earlier reporting, suggesting the bank is managing real concentration risk in the sector at the same time it is working to expand lending in it. The Information's reporting does not address how the bank is navigating that balance sheet question.

What the reporting does not give you is the covenant and pricing detail that would let investors assess the asset class seriously, or clarity on whether rating agencies have a settled view on how to classify these loans. Whether the $20 billion estimate for 2026 actually materializes will depend on how institutional buyers price the risk and whether hyperscaler buildout timelines hold.