Before adding compute, ask when power can arrive

When people discuss a data center project, server counts, rack capacity, and future demand can take up most of the conversation. They deserve attention. Yet a less exciting question often determines when the plan can become real: when the first equipment arrives, will the promised power connection be ready? A capacity figure alone cannot answer that.

The International Energy Agency's Energy and AI report identifies grid connection queues, transmission construction timelines, and the delivery of key equipment as constraints on data center expansion.^[1] Demand for compute and the provision of electricity may therefore follow different calendars. A convincing view of the industry's direction still leaves the delivery conditions of an individual project to be established.

The judgment needs another layer of detail. A region may have substantial energy resources without a particular site having usable capacity on a particular date. An expression of interest, capacity included in a plan, and completed connection requirements represent different stages of progress. Putting them in the same spreadsheet does not give them the same degree of certainty.

A more useful discussion brings the evidence into view. What supports the capacity commitment? Who is responsible for the remaining steps? Can the customer's installation schedule be coordinated with the arrival of power? If one condition changes, how much room remains to adjust purchasing and construction? Answering these questions helps reveal the actual limits of an otherwise attractive plan.

This is a useful place to examine the relationship between recognition and judgment. Seeing rising demand is a beginning; making sense of the conditions that allow demand to become revenue takes further work. If we move directly from a growth forecast to a decision, we can miss the constraints in between. Understanding those constraints gives us a better basis for judging the project in front of us.

The eventual outcome also deserves careful review. A smooth delivery may owe something to preparation, coordination, and favorable external conditions. A delay cannot automatically be attributed to the original judgment. Keeping a record of the evidence available at the time makes it easier to distinguish the reasoning worth retaining from the assumptions that need calibration once the outcome is known.

The next time you read a compute capacity plan, leave a line beside the scale figures: when will this capacity be usable, and what confirms that date? Filling in that line can bring the discussion closer to the reality the project will have to face.