The Data Center Is Approved. Is the County Ready?

Somewhere this month, a county board will approve a data center. The vote will turn on tax base, power, water and jobs. It will rarely turn on a harder question: who responds on that site's worst day, and are they ready?

I have spent more than thirty years on the response side of that question. My work has run from tactical EMS to the DHS Center for Domestic Preparedness, where I was a primary instructor, to FEMA Region VI, where I served as Planning Section Chief on an Incident Management Assistance Team. A pattern I saw again and again in federal disasters is now showing up in data center country: the hazard arrives faster than the capability to manage it.

A hyperscale campus is not simply a large building. It is an industrial site with utility-scale electrical infrastructure, large stores of energy and fuel, and a construction workforce that can outnumber the town around it. The local fire department, EMS agency and emergency manager inherit that risk the day ground is broken. Most were never sized for it.

Sited for power, not for response

The readiness problem begins before the first permit is filed, because site selection now starts with power. Developers chasing gigawatts look first for generation, transmission and an interconnection they can secure. That pulls campuses toward power plants and substations, which often sit in open country, far from the population centers where fire, EMS and hospital capacity are concentrated.

Consider a proposal of the kind now moving through rural counties across the country. The campus would sit beside an existing power plant, on a two-lane highway miles from the nearest town. The applicant projects several gigawatts at full build-out, battery storage measured in gigawatt-hours, and a multi-year build. The closest fire district is small, and the nearest trauma center is more than an hour away by ground.

Then add the people. A hyperscale build can put close to 5,000 construction workers on site at peak, and in many host communities that is more than the town's entire population. Those workers drive the same roads residents use, fill the same motels and clinics, and get hurt at the rates any heavy construction site produces. Their calls land on an EMS agency staffed for the town as it was, not for the workforce that just arrived.

The logic that makes a site attractive for power is the same logic that makes it demanding for response. None of that is a reason to say no. It is a reason to hold the response conversation now, at the preconstruction and permitting stage, while the conditions of approval are still being written.

What arrives with the campus

The risk profile changes twice: once during construction and again when the site goes live.

Construction brings people, and lots of them. That workforce can stay on a rural site for two years or more, working around cranes, trenching, heavy equipment, summer heat and a daily traffic surge on roads built for farm vehicles. That is a steady source of trauma, cardiac and heat-illness calls, and a standing potential for a mass-casualty incident at a location many responders have never seen.

Operations bring energy. Substations and medium-voltage switchgear create arc-flash and electrocution hazards. Rooms of lithium-ion batteries behind the UPS systems can go into thermal runaway, release toxic and flammable gases, and reignite hours after the fire appears to be out. Backup generators and their diesel tanks add fuel fires, spills and regular tanker deliveries. Add secured perimeters, layered access control and large windowless buildings, and the first-arriving crew faces a complex structure that may take precious minutes just to enter.

The capability gap

The communities that host these campuses are often served by volunteer or combination fire departments, a small EMS agency with one or two ambulances in service, and a part-time emergency manager. The nearest trauma center may be an hour away by ground. None of that is a criticism. Those agencies were built for the town as it was, not for an industrial site the size of a small city.

The gap usually shows up in the paperwork first. Mutual aid agreements were written for structure fires and highway crashes, not for a battery-room fire that needs hours of water supply and specialized monitoring. The county emergency operations plan rarely names the site at all. The 911 center may not have current gate codes, site maps or a working phone number for the operator's emergency response team. The hospital has never been asked how it would absorb twenty construction injuries arriving at once.

What federal disaster work teaches

On a FEMA Incident Management Assistance Team, the Planning Section's job is to make sure no one is improvising the plan in the middle of the incident. We wrote objectives, assigned resources and set the operational period before conditions forced our hand. The teams that performed best were the ones that had already worked together before the disaster.

The same principle applies to a data center. The operator will have its own security force, facilities staff and often an internal emergency response team. Local government will have fire, EMS, law enforcement and the emergency manager. On the worst day, those two worlds need to function as a unified command with shared objectives, a common picture of the site and a clear chain of decision. That does not happen at 2 a.m. in a parking lot. It is built months earlier.

Oil spill responses such as Deepwater Horizon and the Enbridge Line 6B release showed what happens when a private operator and government responders meet at scale. Response capacity matters, but so do relationships, information sharing and agreed decision-making. The lesson for host counties is direct: get the operator to the planning table now, while there is still leverage to set the terms.

Five steps before the ribbon cutting

1. Make readiness a condition of approval. The special use permit conditions, development agreement or community benefit agreement should require the operator to disclose site hazards, share pre-incident plans, and fund the training and equipment local responders will need. A county has the most leverage before the vote, not after.
2. Build the pre-incident plan together. Map access points, gate procedures, utility shutoffs, battery and fuel locations, water supply and staging areas. Put it in the hands of 911, the first-due company and mutual aid partners, and update it with every construction phase.
3. Train for the site's actual hazards. Local crews need specific training on lithium-ion battery fires, high-voltage electrical emergencies, and technical rescue from heights and confined spaces. Command staff on both sides should complete ICS-300 and ICS-400 together, so they share a vocabulary before they need one.
4. Exercise in stages. Run a tabletop during construction, a functional exercise before commissioning, and a full-scale exercise in the first year of operations. Include the hospital, the 911 center, the utility and the neighboring departments that would actually respond.
5. Update the plans that already exist. Add a site annex to the county emergency operations plan. Revisit mutual aid for specialized resources and hospital surge. Plan how neighbors would be told to shelter in place or evacuate if smoke or a hazardous release reaches beyond the fence.

The worst-day question

Data centers can be good neighbors. Many operators take safety seriously and will invest in the communities that host them, especially when the request is specific and arrives early. The counties that come out ahead are the ones that ask the worst-day question before the vote and put the answer in writing.

Ask it now, and the first serious incident becomes a plan the community has already rehearsed. Wait, and it becomes an after-action report.

Also published on LinkedIn. Hear the discussion on the Resiliency and Recovery Forum.