AI Data Center Growth Rate Forces Investors to Evolve

The capital markets infrastructure underpinning the AI buildout is maturing fast – but panelists speaking at Deal Catalyst’s Digital Infrastructure Finance conference in New York City this morning cautioned that complexity, concentration risk and power constraints are shaping the next phase of financing.

Andrew Giudici, Global Head of Corporate, Project and Infrastructure Finance at KBRA drew a comparison to how other asset classes evolved over time, noting the speed at which data center financing has developed. Solar took years to move beyond vanilla structures, he said, while data centers have done it in a fraction of that time.

In 18 months, the AI infrastructure market has evolved from projects in tier 1 markets with triple-net lease structures and top-quality tenants to ones featuring assignability provisions, tighter debt service coverage ratios and lease extension provisions.

John Servidea, Global co-Head of Investment Grade Debt at JPMorgan noted the U.S. corporate bond market is the largest and most liquid in the world, and if AI, AI-adjacent and utility issuance were treated as a single sector, it would already rank as the largest segment of that market.

Servidea contextualized the financing boom within a broader convergence between public and private credit. Banks are fulfilling a core role, he said – providing capital and intermediating other sources – while also educating corporate bond investors on structures that they may be less familiar with.

Valentina Berger Jimenez, VP Debt Capital Markets at DigitalBridge described her firm’s progressive approach. It has moved beyond a simple flywheel of project finance followed by bank takeout and ABS execution and is now looking at borrowing base facilities, 144a and private placement markets and CMBS, which she said has grown substantially. The goal, she said, is to understand each market and tap them in a thoughtful manner.

Structure Follows Complexity

Jiri Honajzer, a vice president at Brookfield Infrastructure Debt at Brookfield Asset Management, said that the bond market is emerging as a source of construction capital, with structures evolving to accommodate the full lifecycle of a leased asset. The question facing investors is not just who the tenant is, he said, but what the underlying structure looks like across the full term of the agreement.

Kenneth Irvin, a partner at Sidley Austin flagged an accumulation of promise-makers in a single transaction, pointing to the data center developer, the tenant and the power provider, saying that even with creditworthy counterparties, ensuring everything functions as intended remains a key diligence challenge.

Giudici described a dynamic in which data centers and power projects are sometimes built in parallel while being underwritten in silos. Some structures include provisions where a tenant begins making payments on a fixed date regardless of whether the facility is operational.

The risk, he said, is whether termination-for-convenience provisions in a lease are sufficient to pay off the debt if a tenant stops paying. He also noted that in some deals, even where a tenant is on the hook and an investment-grade counterparty backstops payment, the underlying question of tenant intent remains.

Mary Kogut, a partner at Kirkland and Ellis, raised concentration risk as a systemic issue as hyperscaler issuance scales up. However, Servidea said he does not believe the market is close to hitting single-issuer concentration limits while Berger Jimenez responded that institutional buyers are not approaching concentration through a single lens, rather they are differentiating by type of deployment, market tier and developer.