Data center construction doesn't create new hazards. It overlaps familiar ones faster than inspections can track.

Data center builds put electrical, mechanical, and civil crews in the same space on compressed timelines, and the safety plan has to keep pace. This page explains why lagging metrics miss serious injury potential on these sites, and how point-of-work AI with the right guardrails closes the gap.

published

Nobember 4, 2026

Originally covered by

TOPIC

SIF Prevention · AI

As covered in

EHS Leaders

━ Overview

EHSLeaders features a Q&A with Jake Landgraf, Director, Strategic Solutions & Growth at Benchmark Gensuite, on safety in data center construction, from overlapping trades and missed serious-injury warning signs to practical ways of introducing AI on active sites.

This page adds the operational reasoning behind those answers: why compressed schedules outrun inspection-based safety plans, why lagging metrics hide SIF exposure, and how EHS teams can bring AI to the point of work with guardrails and a way to measure results.

━ Benchmark Gensuite Perspective

On data center sites, the safety plan has to move at the speed of the schedule, not the inspection calendar

Data center construction doesn’t introduce exotic hazards. The exposures are the ones heavy-construction teams already manage: electrical contact, falls, lifting, excavation, hot work. What changes is density. Electrical, mechanical, civil, and commissioning crews from different contractors work in adjacent space on a schedule with almost no slack, and every handoff creates an interface that no single contractor’s pre-task plan was written to cover.

That is why scheduled inspections struggle here. A weekly walk-through, or a plan built against the baseline schedule, describes the site as it was planned rather than as it exists by midweek. Safety coverage has to follow the work as it shifts, which means the information must come from the people doing it.

Getting that information is harder than it sounds. Total recordable incident rate and other lagging measures report outcomes after the fact, and because most recordable injuries are minor, a healthy rate says little about fatality potential. Serious injury and fatality (SIF) precursors tend to sit in hazard observations and close calls that never get filed, because under production pressure a worker who stops to report is a worker who falls behind. Lowering that friction matters more than adding forms. Energy-based frameworks such as STKY (stuff that can kill you) and the energy wheel give crews a shared shorthand for recognizing, in seconds, what could seriously hurt them.

“…like having a safety coach in every frontline worker’s pocket.”

— Jake Landgraf, Director, Strategic Solutions & Growth, Benchmark Gensuite

This is where AI earns a place, with a specific job: extending scarce safety expertise to the point of work. A small safety team supports a far larger population of frontline workers and subcontractors, and AI can supply the second set of eyes that is aware of SIF potential when no professional is present.

The design choice that matters is who does the thinking first. A prompt that asks a crew to name the hazards in a photo before the AI adds what was missed keeps worker judgment engaged, while a tool that simply issues a hazard list invites passive compliance. The payoff also reaches beyond the field: when most of an EHS leader’s week goes to reports and training plans, shifting that time toward the job site puts experienced eyes where SIFs actually start. The test of any of it is measurement. Are high-energy hazards being identified more often, and are SIFs falling?

━ Key themes

Three things to understand about data center construction safety

  1. 1
    Overlap creates hazards that no single contractor owns
    Each contractor's pre-task plan covers its own scope. Risk concentrates at the interfaces between trades, such as a lift above another crew's work area or an energized panel beside finishing work, where responsibility is shared and easy to leave unassigned. Site-level visibility has to sit above the individual contractor plans.
  2. 2
    A strong TRIR can coexist with serious exposure
    SIFs are rare, so a project can post a healthy recordable rate while high-energy hazards accumulate unseen. A more useful leading signal is how often crews identify and correct those hazards, a count that rises when reporting is easy and falls when it isn't.
  3. 3
    AI value is set by where it is deployed and how it is measured
    Spreading AI across an entire site dilutes its impact. Starting with high-hazard tasks, coaching people on how to use it, and tracking hazard identification alongside SIF trends turns AI from an activity into a control that can be evaluated.

━ Featured executive

Director, Strategic Solutions & Growth - Benchmark Gensuite

Jake Landgraf is Director of Strategic Solutions & Growth at Benchmark Gensuite, where he leads strategy for the company's frontline safety portfolio, including ANVL® Field Safety and ANVL® Layered Process Audits. He helped design ANVL Mobilize, the 2025 OH&S New Product of the Year in EHS Apps. Over more than five years in frontline safety technology, he has helped utilities, contractors, and industrial operators digitize high-risk field work. He speaks regularly on frontline safety applications of AI, including at the National Safety Council and TVPPA.

Frequently Asked Questions

What EHS leaders are asking

What are the biggest safety risks on data center construction sites?

The hazards match other heavy construction: electrical contact, falls, lifting, excavation, and hot work. What raises the risk is density, with multiple contractors’ crews working in adjacent space on compressed schedules, so exposures overlap in ways individual pre-task plans do not cover.
TRIR is a lagging measure that counts outcomes after they occur, and most recordable injuries are minor. A project can post a strong rate while high-energy hazards go unreported, which is why leading signals such as how often crews identify and correct serious hazards are more useful for SIF prevention.
AI can extend limited safety staff to every crew at the point of work, acting as a second set of eyes that recognizes SIF potential in photos and task descriptions. It also reduces the administrative load on EHS leaders, freeing time for field presence where serious hazards are most likely to be caught.
Start with a few high-hazard tasks rather than the whole site, and have workers name hazards first so the AI adds what they missed instead of replacing their judgment. Provide coaching on how to use the tools, then track whether high-energy hazard identification rises and SIFs fall.

SEE IT IN ACTION

Ready to see Benchmark Gensuite in action?

See how enterprise EHS teams use our platform to manage risk, compliance, and safety outcomes at scale.

Find your enterprise instance of
Benchmark Gensuite

Note: Subscriber instances where you have an active registered user account are listed above. If you need further support, please email us at getHelp@benchmarkdigital.com
Please include your company name and registered email address so we can assist you.