Our method for data center studies
We model construction and operations as separate phases and never add them into a single annual figure. Construction is a one-time shock spread over the build schedule. Operations recur every year once the facility is running.
For construction, we split the capital budget into site work and shell, mechanical and electrical systems, IT equipment, and professional services, because each category has a different local share. Most servers, switchgear, generators and chillers are manufactured outside the host county. In the regional model we enter only the local portion of those purchases, such as installation labor, freight and distributor margins. Entering the full equipment budget as local spending is a frequent cause of overstated data center results.
For operations, the direct shock is on-site payroll plus local purchases of maintenance, security and other services. Electricity needs care: its local effect depends on where the power is generated and how the utility is structured.
Fiscal results are reported by tax type and jurisdiction, net of any abatements in the incentive agreement, and kept in tables separate from the economic results.
Typical inputs we ask for
- Capital budget by category (site and shell, MEP, IT equipment, soft costs)
- Construction schedule and expected peak on-site workforce
- Planned operating headcount and wage levels
- Annual operating budget excluding depreciation
- Expected power demand and the serving utility
- Parcel and assessed-value details, plus the terms of any incentive or abatement
Questions clients ask
Why are operations jobs so much smaller than construction jobs?
Data centers are capital intensive. The building and equipment cost a great deal relative to the staff needed to run them, so construction usually supports far more jobs than operations. Construction jobs last for the build; operations jobs recur every year. We report both, label construction as one-time, and show the supply-chain and household-spending effects of operations so the recurring footprint is visible.
Can the study support a tax incentive application?
Yes. Our project studies for Cielo Digital Infrastructure supported incentive applications and local government negotiations, with fiscal results by tax type modeled net of abatements. If a jurisdiction needs to weigh new revenue against the cost of public services, that is a fiscal impact question, and we scope a fiscal module alongside the economic impact study.
Should the study area be the county or the state?
It depends on who will read the results. A county planning commission or board of supervisors needs county figures, which will show more leakage because less equipment and fewer services are sourced locally. A state incentive program may want statewide results, which capture more of the supply chain. We can report both levels, labeled clearly, so neither figure is quoted in the wrong setting.
Related reading
- Cielo Digital Infrastructure case study
- U.S. data center construction study
- Multipliers explained
- Fiscal impact vs. economic impact
- How to evaluate an economic impact report
Sources
- Bess and Ambargis, Input-Output Models for Impact Analysis: Suggestions for Practitioners Using RIMS II Multipliers, BEA working paper WP2012-3 (on producer prices, margins and entering only net local final demand) Source
