On most construction projects, there’s a rhythm.
Structure goes up. Trades follow. Systems get installed. Then commissioning begins.
But data center construction demands more: from teams, systems, and safety. You’ll have crews installing cable trays next to systems being energized. Commissioning teams testing equipment while other trades are still trying to finish their scope. Day shift hands over to night shift, and the site resets without ever really stopping.
It’s not just busy, it’s layered.
That’s where safety starts to get tested. And when systems are live and crews are still building around them, small gaps don’t stay small for long.
What's Inside |
|
3. Workforce complexity and rotation 4. Repeatable and standardized delivery |
The U.S. sits at the center of the global data center construction boom. This growth drives faster builds, denser systems and puts pressure on timely, safe delivery.
$48.18bn to $112bn: Expected jump (133 per cent) in U.S. data center construction investment from 2024 to 2030 (Arizton)
$258.56bn to $491.99bn: Expected jump (90 per cent) in the global data center construction market from 2025 to 2034 (Precedence Research)
20.8 per cent: Growth in the Dodge Momentum Index (leading indicator of construction starts) in July 2025, driven largely by data center projects (Dodge Construction Network)
Examples of data center projects that entered planning in July 2025:
Fairview Connections Data Center in New Cumberland, PA ($500m)
Jabil AI Data Center in Salisbury, NC ($500m)
Data centers concentrate risk in ways most construction projects don’t.
High-voltage systems
Dense mechanical and electrical infrastructure
Overlapping trades working in tight spaces
Continuous delivery cycles with no clean breaks
These factors exist on other projects.
But the combination and speed of them on mission critical sites changes how safety needs to be managed.
“You have a lot of people working in those environments at accelerated rates and long hours. So, those things put together create a very high risk environment.”
Data centers are built in phases, but those phases run into each other.
One area is being energized while another is still under construction. Prefabrication speeds up installation, but it also means multiple workstreams land on site at the same time.
You don’t get separation. You get overlap.
"These data centers are almost always delivered in phases. It takes a huge coordination effort. You’re building while working around very expensive, live equipment.”
The exposure comes from how work interacts.
An electrical team might be operating under a valid permit. A mechanical crew might be following their JHAs. Both are working correctly within their scope.
But when those scopes share the same space, the risk changes.
That’s where coordination becomes a safety control.
Permits, isolations, and work approvals need to be visible across trades, not just within them. If coordination breaks down between teams, work doesn’t just become unsafe. It slows, stops, and must be re-sequenced.
Data center construction is simply too complex to rely on reactive planning.
Detailed JHAs and PTPs are developed early. Work sequences are mapped out. Dependencies between trades are agreed before mobilization.
Everything looks aligned at the start.
"All safety programs, JHAs, and high-risk activities are submitted, reviewed, and approved before work can start.
Getting that information reviewed and approved before work begins is critical. Poor-quality PTPs have been a root cause behind many of the issues we’ve seen.”
Plans don’t stay static once work begins.
Sequencing shifts. Access changes. Crews adapt to keep progress moving. But the original JHA often stays exactly as it was approved.
That’s where gaps form.
Teams that manage this well build review into the workday:
JHAs and PTPs are checked before work starts
Toolbox Talks focus on what’s happening today
Changes are course corrected early, not after an incident
Because by the time issues show up in the field, the opportunity to fix them without disruption has usually gone.
Data center sites bring together large, specialized workforces that don’t stay static.
Subcontractors move in and out. Crews rotate across shifts. Specialist trades are brought in for short, high-risk scopes.
The makeup of the site changes constantly.
"On most data center projects, you’re managing trade partners rather than self-performing the work. Establishing a clear source of truth for how the project is run is one of the biggest challenges.”
Visibility becomes critical.
Site teams need to know:
Who is on site right now
What they are qualified to do
Whether they’ve completed the required induction and safety submittals
Without that, decisions are based on assumptions.
Someone looks experienced. A crew has “done it before”. A supervisor assumes training is current but new crews need access to safety submittals too so they can get caught up straight away.
Those assumptions hold until they don’t. Clear, accessible workforce data allows teams to focus on managing risk, not verifying basics.
Data centers are one of the most repeatable project types in construction. Similar layouts, systems, and client expectations.
“Data center work is very much rinse and repeat. The standards, the documentation, the way projects are handed over is very consistent. So, the more data we can capture and use, the easier it is for the next team.”
The same risks show up repeatedly. Congestion between trades. Pressure during commissioning. Access constraints in high-density areas.
None of it is new. The difference is whether those risks are carried forward as learning. Without that continuity, teams on each projects have to spend time and energy addressing the same challenges.
Teams that improve treat safety processes as something that evolves:
JHAs are refined based on real project experience
Common issues are addressed and corrected before the next project
Workflows and repeatable are standardized across sites
Commissioning sees systems go live, testing ramps up, and multiple teams converge on the same areas to meet handover deadlines. There is zero room for delay.
Commissioning is where schedule pressure peaks — and where small safety gaps carry the biggest consequences.
Small issues carry more weight.
A missed isolation or unclear permit doesn’t just create risk. It can stop progress entirely. With the average cost of a data center outage hovering around $9,000 a minute (Vertiv), delays during commissioning carry immediate financial impacts.
At this stage:
Permits need to be clear and accurate
LOTO procedures need to be verified, not assumed
Documentation needs to reflect actual site conditions and be referenced during site walks
If teams are chasing information or second-guessing approvals, the pressure moves straight back onto the field.
Structured, traceable workflows keep that pressure under wraps.
Data center clients expect consistency across every project. They’re not just looking at outcomes but how those outcomes are achieved. And they expect visibility.
"Some of our larger data center clients have very strict safetyrequirements. They expect a high level of compliance. Many owners want direct access. They want full visibility into what’s happening on site.”
What does this mean for safety?
Safety performance needs to be demonstrated, not described.
That includes:
Clear records of inductions and training
Visibility into permits and high-risk work
Instant transparency into overall safety performance
When that information is easy to access, conversations with owners stay straightforward.
When it isn’t, teams end up reconstructing what happened after the fact. That’s where confidence starts to slip.
The same patterns show up across most data center projects.
Teams that stay in control tend to focus on a few fundamentals:
Keep safety information current as outdated plans create blind spots
Make work visible across trades as coordination reduces overlap risk
Verify workforce readiness daily, not just when onboarding
Review plans against real conditions i.e. before works, not after issues arise
Standardize what works and carry it forward to the next project
None of this is new.
The challenge is doing it consistently when the job doesn’t slow down.
Data center construction doesn’t give you time to catch up.
Work moves quickly. Conditions change constantly. Multiple teams operate in the same space, often around live systems.
Safety needs to move at the same pace.
“There are more people, more density, and less time than we’ve ever had before. We need to help our teams focus on the most important things.”
That means staying connected to what’s happening on site, not just what was planned at the start. It means visibility across teams, clarity around high-risk work, and processes that hold up under pressure.
Because on these projects, the gap between plan and reality is where risk builds.
Closing that gap is what keeps teams safe.
McKinsey and Company: AI Power - Expanding Data Center Capacity to Meet Growing Demand
U.S. Department of Energy: Best Practices Guide for Energy-Efficient Data Center Design