Data centre business cases are usually presented as real-estate models with an IT flavour: land cost, shell cost, fit-out, lease-up curve. But having spent nearly four decades around digital infrastructure, I can tell you where these projects are actually won or lost: in four power decisions that are often made too late, by the wrong people, with too little information.
1. The grid connection
Everything starts with the interconnection: what capacity, at what voltage, from which substation, on what timeline, with what firmness. Two otherwise identical sites can carry radically different business cases because one has 60 MVA available at 132kV within eighteen months and the other is behind a network upgrade with an unfunded completion date. Due diligence here is not a checkbox — it is the core of the investment decision.
2. The substation
Whether you build a dedicated substation or share utility infrastructure shapes capex, programme and risk in one stroke. A dedicated 33–132kV substation adds meaningful cost and a long procurement tail — transformers and switchgear carry some of the longest lead times in the industry — but it buys control, expandability and often a faster path to full load. The right answer depends on phasing: a business case built on a 4 MW day-one load and a 40 MW year-five ambition needs the substation conversation on day one.
3. The backup architecture
2N, N+1, distributed redundant, or something leaner for AI training loads that can tolerate interruption? Backup architecture is one of the largest single capex lines and one of the most workload-dependent. Traditional enterprise loads demand full concurrent maintainability; some AI training campuses are consciously trading redundancy for density and cost. Designing the backup architecture before you understand the workload mix is how projects end up gold-plated and uncompetitive — or fragile and unlettable.
4. The renewable strategy
Power is the dominant operating cost, and increasingly a licence-to-operate issue. Solar PPAs, green tariffs, on-site generation and battery storage each change the effective cost per kilowatt-hour and the sustainability story you can credibly tell customers and regulators. In markets like the UAE and India, well-structured renewable procurement is now a genuine competitive lever — the difference between winning and losing an anchor tenant whose own commitments require green power.
“The spreadsheet decides nothing. The grid connection, the substation, the backup architecture and the power contract decide everything the spreadsheet reports.”
The discipline
None of these four decisions belongs to a single discipline. They sit at the intersection of engineering, procurement, finance and strategy — which is exactly why they are so often mishandled. The projects that move fastest are the ones that put an experienced owner's team around these decisions early, before the business case hardens around assumptions that the grid will not honour.
About the author
Saandeep V. Dandekar
Executive Director — Infrastructure Strategy & Delivery, MetaDecrypt AI Atlas