🧠 Food for thought

Implications, context, and why it matters.

AI buildout lifts deals amid unclear profits

• Data center deals hit just under US$61 billion through November 2025. Goldman Sachs sees data center power demand rising 50% by 2027 and up to 165% by 2030 from AI computing 1. It is unclear if these workloads are driving pre-leases (leases signed before construction is complete) or lease rates that justify today’s valuations.
• Private equity is holding high-quality assets rather than flipping them, which signals confidence in long-term value. Rising power needs and build costs must be passed through to tenants such as cloud platforms or AI model developers. The lag between invested capital and cash flow clouds the outlook for deal volume.

Operators and investors can target power-rich regions with strong incentives

• Operators, along with digital infrastructure investors, should target markets that pair grid capacity with fast permitting and tax breaks. Thirty-six U.S. states run statutory data center incentives tied to investment and jobs 1. Projects with at least $500 million in capital spending or an incremental 100 MW or more load can tap federal loans, grants, and tax incentives under a recent Executive Order 2.
• Site selection should start with available megawatts and a predictable rulebook. Useful tools include cost segregation (an accounting method that reclassifies assets to accelerate depreciation) with 100% bonus depreciation for immediate expensing of many assets. Add Section 179D (a U.S. tax deduction for energy-efficient commercial buildings) that can exceed $5 per square foot and renewable energy Investment Tax Credits of 30% or higher 3.