Wellington Passes on Data Center Debt, Questioning Long-Term Value (Podcast)

Wellington Management says data-center debt returns are falling while risk is rising, with uncertainty about what properties may be worth in the long run. The firm notes it focuses on “replacement value” when evaluating data centers, suggesting tighter underwriting standards as future valuations become harder to underwrite. The piece is a cautious read-through for private credit exposure to data-center real assets, rather than a specific market-wide policy shift.

Analysis

The market mechanism is a repricing of the financing layer, not a demand shock. When debt underwrites are getting less generous, the first-order losers are speculative developers and any lender that treated data centers like low-risk infrastructure; the second-order winners are the scaled incumbents that already control powered land and can finance growth internally. That creates a widening moat for EQIX/DLR-style assets over the next 1-3 months as new supply gets harder to pencil, even if headline AI demand remains intact.

The more important spillover is into the build-cycle ecosystem. Tighter credit slows greenfield starts and refinancing, which can flatten backlog growth for power, cooling, and electrical equipment vendors before it shows up in revenue; the earnings inflection for those names is usually 1-2 quarters after the credit impulse. If cap rates keep moving out, private-markets marks can force asset sales and reset comps, which would matter for anything levered to replacement-value assumptions.

Consensus is probably underestimating how concentrated the winners become if external financing stays scarce. The true scarcity premium accrues to operators with existing campuses and balance sheets, while the rest of the chain becomes more cyclical than the AI narrative implies. The thesis breaks if construction lending reopens quickly or if hyperscalers keep funding capex from operating cash flow, making credit less relevant than feared.

Key Decisions for Investors

• Long EQIX vs. short VRT on a 3-6 month horizon: express the view that tighter data-center credit slows new-build activity more than it hurts incumbent leasing; target 2:1 upside/downside if EQIX keeps comping on scarcity while VRT backlog growth softens.
• Buy EQIX or DLR on a 5-7% pullback rather than chasing here: the risk/reward is better if refinancing stress forces supply discipline and supports occupancy/rent spreads over the next 6-18 months.
• Reduce or avoid high-beta data-center construction beneficiaries until financing spreads stabilize: watch VRT/ETN for order-book deceleration as the first falsifier of the build-cycle thesis.
• Set an alert for private-credit and REIT marks tied to data-center collateral; if cap rates expand another 50-100 bps or refinancing is delayed, expect a broader de-risking in adjacent infrastructure credit.