Market Intelligence
Perspectives
Market commentary, transaction analysis, and perspective on the forces reshaping data center investment.
AI Demand
Project Visibility
Product Specs
AI Infrastructure Demand: Why Visibility Matters More Than Location
The data center market has never seen demand like this. AI compute requirements are driving leasing inquiries at a pace that would suggest any developer with available capacity should be fully leased within months. The reality on the ground is considerably more complicated.
AI tenants will commit — but only when they can see clearly to the finish line. The threshold question for any AI customer isn't whether the space exists, it's whether the project has sufficient visibility on delivery. Power availability confirmed, permitting advanced, equipment lead times understood, sponsorship and capitalization in place. When those elements are aligned, AI tenants move decisively. When they aren't, the tenant walks — regardless of how compelling the location or the relationship.
This creates a dynamic that catches many institutional investors off guard. The assumption coming into the market is that AI demand is so strong that a credible development site with a capable sponsor will attract a tenant commitment early in the process. In practice, the sequence doesn't work that way. AI customers won't commit to a project that has too many open variables, because their own obligations to their end users depend on certainty of delivery. The developer needs enough work done to give the tenant genuine confidence — and that work requires capital and time before any lease is signed.
The chicken-and-egg problem is real: investors are reluctant to advance a development without a tenant commitment, and tenants are reluctant to commit without the level of project advancement that gives them delivery confidence. Navigating that gap — knowing how much development work is necessary to reach the tenant commitment threshold, and structuring the capital to get there — is where experienced advisory makes the difference.
The product requirements have also shifted materially. A well-located Tier 3 air-cooled facility at standard density was a highly leasable product for most of the past decade. That playbook is largely obsolete for AI workloads. Today's AI tenants require higher power density, liquid cooling readiness, and infrastructure that is forward-thinking enough to accommodate the next generation of compute requirements — not just today's. Getting the product specification right from the outset is no longer optional.
For institutional investors entering or expanding in this market, the implication is straightforward: AI demand is real and durable, but capital deployed into data center development without sufficient project visibility and the right product specification will not attract the tenants the underwriting assumes.
Market Concentration
Auction Dynamics
Off-Market Value
The Institutional Data Center Trap: Why the Best Opportunities Aren't in the Auction
The institutional data center investment market has a concentration problem. A significant share of institutional capital is chasing the same small set of assets — fully marketed portfolio transactions run by top-tier brokers and investment banks, with established cash flow, professional management teams, and ambitious growth projections built into the offering memorandum. When fifteen qualified bidders compete for the same asset, the math is straightforward: price goes up, returns go down.
The valuation dynamic in these processes follows a predictable pattern. Bidders systematically overvalue projected growth — future leasing velocity, expansion pipeline, power capacity that hasn't yet been permitted — and undervalue the execution risk embedded in achieving it. Timing slippage on a hyperscale lease, equipment lead time uncertainty, power availability constraints in a competitive market — these risks are real, they are material, and they tend to get compressed in a competitive auction where the pressure is to win.
The more interesting opportunities rarely appear in a formal marketing process. They don't check every institutional box on first review. The management infrastructure may be thin, the cash flow may be lumpy, or the asset may require a more hands-on approach to unlock its value. For investors without deep sector experience, these characteristics read as liabilities. For investors who can evaluate the true risk and opportunity through rigorous, operationally-informed diligence, they often represent exactly the profile that generates superior returns.
The under-the-radar opportunity tends to offer something the auction process cannot: a realistic entry price. And when the asset is properly capitalized, stabilized, and positioned — with institutional-quality management, a credible growth story, and the right tenant relationships in place — it becomes precisely the kind of asset that commands unrealistically low cap rates in a competitive sale process. The entry discount and the exit premium are two sides of the same opportunity.
For institutional investors serious about data center returns, the question worth asking is not which marketed portfolio to pursue — it's what the rest of the market is overlooking.
Power Strategy
Utility Relations
Market Geography
Power Is the New Location: What Data Center Investors Need to Understand
For most of the past two decades, power was a checkbox in data center underwriting. You acquired a site, engaged the utility, received a commitment based on your projected load, and moved forward. Utilities routinely over-committed and found that sponsors consumed only a fraction of their reserved capacity — an inefficiency that masked the underlying constraints for years. That era is over.
Today's utility environment is fundamentally different. Power studies that were once routine and low-cost now require substantial deposits before a utility will even begin assessing what can be delivered, on what timeline, and at what cost. The results of those studies increasingly include requirements for the sponsor to contribute to infrastructure upgrades and equipment costs that utilities historically absorbed themselves. And the timelines are no longer measured in months — in many markets, meaningful power availability is years away, a reality that renders otherwise attractive development sites effectively unleasable for AI and hyperscale workloads in the near term.
The instinct of many investors is to interpret constrained power as a binary problem — either the power is there or it isn't. The more useful frame is that power availability is a negotiation, and the outcome depends heavily on who is doing the negotiating. Utilities have real constraints but they also have flexibility, and understanding what a utility needs — how a sponsor's project can align with the utility's own infrastructure priorities and capital planning — can unlock capacity that isn't visible in a standard power study. The right combination of an experienced data center advisor and a power engineering team that knows how to work with utilities constructively will find capacity that less sophisticated sponsors cannot.
Geography is also shifting in ways that create opportunity for investors willing to look beyond the established Tier 1 markets. Northern Virginia, Silicon Valley, Chicago, and the other traditional absorption centers are experiencing the sharpest power constraints precisely because they have been the most active markets. AI, hyperscale, enterprise, and colocation customers who would historically have defaulted to Tier 1 locations are increasingly open to alternative markets — not by preference, but out of necessity. For investors and developers who move into those markets early, with the right power strategy and local utility relationships, the competitive dynamics are considerably more favorable than anything available in a mature market today.
Power is no longer a checkbox. It is the project.