Eldridge President: Data Center Demand ‘Not Going Away’

Tony Minella said Eldridge is leaning into AI as an enabler and argued that demand for AI data centers is not going away, reinforcing a constructive view on the AI infrastructure buildout. He also highlighted the undeniable strength of equity markets. The remarks are broadly supportive of risk assets and AI-related investment themes, but the article contains no hard financial metrics or company-specific catalyst.

Analysis

The important signal is not a fresh AI demand thesis, but a capital-allocation one: large private allocators are still willing to fund the physical buildout even as public-market multiples have already rerated. That tends to extend the cycle in the picks-and-shovels complex because the bottleneck shifts from “is this real demand?” to “who can actually deliver power, cooling, interconnect, and land on time.” The second-order winners are less the obvious GPU vendors and more the grid-adjacent names, electrical equipment suppliers, and infrastructure developers with multi-quarter backlog visibility.

The market implication is that AI capex is becoming less sensitive to near-term sentiment and more tied to financing conditions and power availability. If equity markets stay firm, the cost of capital for private infrastructure stays manageable, which reinforces the buildout; if rates reprice higher or secondary equity markets weaken, the lag will show up first in smaller data-center developers and venture-backed enablers, not in the hyperscalers. That asymmetry favors quality over beta: the market is likely underestimating how much of the incremental spend gets captured by firms selling capacity and connectivity rather than software narratives.

The contrarian risk is that consensus is treating AI infrastructure demand as linear, when in practice it can pause abruptly if compute utilization disappoints or if permitting/grid constraints delay revenue conversion. In that case, the overhang is not a collapse in demand but a digestion phase where orders remain strong while bookings-to-revenue timing stretches, pressuring the more levered private-market names first. Over a 3-6 month horizon, the setup is still supportive; over 12-24 months, the key reversal catalyst is a capital-spend discipline shift once investors demand proof of monetization rather than just capacity growth.

Key Decisions for Investors

• Go long a basket of AI infrastructure beneficiaries with operating leverage to power and data-center buildout (e.g., VRT, ETN, APH) for 3-6 months; target 15-20% upside if capex revisions continue, with a 8-10% stop if AI spend expectations roll over.
• Pair long quality infrastructure equities vs. short higher-duration, valuation-sensitive AI software names that depend on future monetization (e.g., long VRT/ETN, short a basket of unprofitable AI software proxies) for a relative-value trade over the next 1-2 quarters.
• Use call spreads in utility/grid equipment names on 6-9 month tenor to express the view that power bottlenecks, not model breakthroughs, are the binding constraint; risk/reward is attractive because backlog should de-risk earnings before sentiment fully prices it in.
• Avoid chasing small-cap data-center developers after strength; wait for any 10-15% pullback tied to rate volatility before entering, since these names are most exposed to financing conditions and execution slippage.
• If public markets weaken materially while AI capex commentary stays firm, rotate from venture-like enablers into hyperscaler supply-chain leaders; that is the cleaner expression of the theme as risk appetite fades.