Why Power Is the Binding Constraint

If entitled, powered land is the scarce asset, then within that phrase one word does most of the work. Power is the binding constraint of the modern data-center era, and understanding why reshapes how you weigh every other variable.

Start with scale. A single large training or inference campus can demand as much electricity as a mid-sized city. Where a decade ago a big facility might draw tens of megawatts, leading campuses now plan for hundreds, and announced projects reach into the gigawatts. Demand of that magnitude does not slot quietly into an existing grid. It must be planned for, generated, and physically delivered, and each of those steps runs on the slow clock of heavy infrastructure rather than the fast clock of software.

The clearest evidence of the constraint is the interconnection queue. In the United States, generators and large loads that want to connect to the grid must apply to the regional operator, submit to a series of engineering studies, and wait their turn. Across the major US markets, the volume of capacity waiting in these queues has grown into the thousands of gigawatts, far exceeding what will actually be built, and typical wait times now stretch to several years. For a data-center developer, that queue is not an abstraction. It is the line you stand in to get energized, and your place in it can determine whether your campus opens in 2028 or 2033.

Why is the queue so slow? Because adding a very large load often requires upgrading the shared network around it. New transmission lines cross many properties and require their own permitting, land rights, and construction, frequently taking the better part of a decade. Substations must be built or expanded to step high-voltage transmission down to usable levels. Generation must exist to serve the new demand, and in many regions the retirement of older plants is outpacing the addition of new firm capacity. Each of these is a physical asset with a long lead time, and a data center cannot open until the weakest link is complete. Power is therefore not a line item you procure late in the process; it is the pacing constraint that governs the entire schedule.

Contrast this with the other inputs. Fiber can be trenched in months. Water rights, where genuinely available, can often be secured faster than firm power. Land can be optioned in weeks. Cooling equipment, generators, and even specialized transformers have lead times, but none rival the multi-year, multi-party choreography of getting hundreds of megawatts delivered to a new point on the grid. When one input is dramatically slower and less elastic than all the others, that input governs the outcome. In site selection, that input is almost always power.

This is why sophisticated developers have inverted the traditional search. Rather than picking a location for its market or tax profile and then asking the utility for power, they now start from where power is actually available and work backwards to everything else. They study which regions have generation headroom, which substations have spare capacity, which utilities can commit firm delivery on a workable timeline, and which interconnection queues are moving. The parcel is chosen to fit the power, not the other way around. Faraday's read: this inversion is the single most important behavioral shift in the industry over the past few years, and beginners who still lead with real estate are already a step behind.

The constraint also explains a set of behaviors that look strange from the outside. Developers pay large premiums for sites with existing interconnection rights, because those rights represent time that cannot otherwise be bought. Some pursue "behind the meter" arrangements, siting generation directly at the campus to sidestep parts of the transmission bottleneck. Others target retiring or retired power plants, because a decommissioned coal or gas station leaves behind transmission interconnection and substation infrastructure that a new load can inherit far faster than building from scratch. Each of these tactics is an attempt to solve the same problem: not enough deliverable power, not soon enough, in the places people want to build.

There is a strategic implication worth internalizing early. Because power is the binding constraint, the value of a site is dominated by its power position, and the other variables matter mostly at the margin or as disqualifiers. A site with excellent fiber, generous incentives, and cheap land but no path to firm power is not a good site; it is a trap that consumes capital while the developer waits for energy that may never arrive on schedule. A plainer site with a signed interconnection agreement and firm delivery in two years is worth pursuing hard. When you rank sites, weight power accordingly. Treat it not as one of six equal factors but as the gate that the others cannot open. Everything in the modules that follow builds on this single ordering principle.