Managing Risk on Data Center Megaprojects

Managing Risk on Data Center Megaprojects

Hyperscale safety compliance means proving that thousands of workers, dozens of subcontractors, and multiple general contractors on one campus are meeting OSHA rules, owner mandates, and wrap-up carrier requirements at the same time, with records that survive an audit. The money behind that challenge keeps growing: Dell’Oro Group now projects global data center capex will pass $1 trillion in 2026, with the four largest US hyperscalers raising their spend 78 percent in a single quarter. Every one of those dollars lands on a jobsite where a single unverified worker can put the schedule, the insurance program, and the owner relationship at risk.

This piece is the roll-up of our data center series, written for the COOs, CFOs, risk leaders, and safety executives who answer for megaproject exposure. BiltOn is a Safety Intelligence platform that gives enterprise GCs one verified workforce and compliance record across every gate, shift, and joint venture partner on a hyperscale campus.

Two terms anchor this guide:

Safety Intelligence: the practice of turning verified, field-captured safety data into risk decisions that hold up to an audit and to an insurance underwriter.

Predictive Safety Management: using verified leading-indicator data to score risk and act before an incident happens, rather than reacting after a claim is filed.

What makes hyperscale different from an ordinary data center job?

Scale changes the category of the problem. An ordinary data center build is a large commercial project with strict access rules. A hyperscale campus is an industrial program: Meta’s Richland Parish campus in Louisiana now represents more than $50 billion of investment with 7,500 construction jobs at peak, building 5 gigawatts of compute capacity across multiple phases running for years.

The sector’s own cost experts confirm the pressure. Turner and Townsend’s Data Centre Construction Cost Index 2025 to 2026, drawing on 250 industry experts across 52 markets, reports that 73 percent of industry leaders still view the sector as recession-proof even as delivery challenges mount. Demand is not the constraint. Delivering safely at this scale is, a theme we opened in our review of the top data center construction challenges.

What does the hyperscale compliance stack include?

Four layers of requirements sit on every hyperscale GC at once, and each layer keeps its own scorecard. Passing one layer proves nothing to the next.

• OSHA: the federal baseline, including the multi-employer rules that let citations reach controlling contractors, covered in our data center regulatory playbook.

• Owner safety mandates: hyperscalers protect their brand with requirements above code, from site-specific orientations and drug testing to badge-level access control and incident reporting inside 24 hours.

• Wrap-up carrier requirements: the insurer behind the OCIP or CCIP sets enrollment, payroll reporting, and claims documentation duties that follow every subcontractor.

• The local AHJ: the authority having jurisdiction, meaning the local building and fire officials who inspect and approve the work, with rules that shift at every county line on a multi-state program.

• Multi-jurisdiction overlap: a portfolio spanning Virginia, Texas, and rural markets answers to different inspectors, labor rules, and emergency services capacity, a gap we mapped in the rural data center safety playbook.
  

How do wrap-up insurance programs change safety obligations?

A wrap-up program consolidates insurance for everyone on the project into one policy, purchased by the owner in an OCIP or by the contractor in a CCIP. In plain language, the sponsor buys one big policy so a hundred subcontractors do not bring a hundred small ones. Nearly every hyperscale campus runs one, because owners want unified coverage over a multi-billion-dollar asset.

The trade is administrative weight. IRMI’s 2025 guidance on wrap-up programs notes that enrollment is mandatory but not automatic, and that contractors must keep submitting certificates, payroll reports, and incident documentation for the life of the project. On a campus with hundreds of subs, that paperwork becomes its own risk: a sub working outside enrollment is an uninsured exposure standing inside your fence.

Beyond the paperwork, a wrap-up changes the incentive math. Because the sponsor holds the deductible, every claim on the program hits the sponsor’s ledger, so the carrier and the owner both want proof of who was on site, what they were trained for, and what the crews did before each shift. Carriers price the risk they can see, and on a wrap-up program they are looking at the whole campus, not one contractor’s loss runs.

How do you govern credentials when thousands of workers churn through the gate?

Credential governance at hyperscale means verifying identity, training, and enrollment status for a workforce that turns over constantly, without slowing the gate. Deloitte’s 2026 engineering and construction outlook, summarized by Construction Executive, pegs the industry’s shortfall at roughly 500,000 workers, which guarantees heavy churn as trades chase megaproject wages from campus to campus.

Badges and paper rosters were designed for hundreds of workers, not thousands. A shared badge or a misspelled name is invisible in a spreadsheet and obvious in a verified identity system, which is why we treat access control as the foundation of site compliance. Broadway Construction Group runs about $900 million under construction management on one access and credentialing standard with a central office of roughly six people, and standardizes every name entry against city-issued cards. When a prevailing-wage claim arrived, BCG produced clock-in and clock-out records showing the claimants were absent or on other sites for many of the days claimed.

The throughput question answers itself with the right architecture. BiltOn clients report 90 percent faster worker check-in and 15 to 20 hours per week of admin saved, client-reported across the platform, using 3D facial verification tied to each worker’s credentials, an approach we detailed in the benefits of 3D facial verification on data center sites.

What is the incident exposure math on a megaproject?

Exposure scales with worker hours, and a hyperscale campus produces more worker hours in a month than most projects produce in their lifetime. A campus running 6,000 workers at 50 hours a week generates about 300,000 exposure hours weekly, roughly 15 million a year, before the second phase mobilizes. Small probabilities multiply into near certainties at that volume.

The consequences are not theoretical. In October 2025, a worker was struck and killed by an articulated truck at a 440-acre data center project in Medina County, Texas, and all work on the site was suspended during the investigation. On a campus burning millions of dollars a week, a suspension is a program-level financial event, and the National Safety Council puts the average medically consulted injury at about $48,000 before indirect costs.

Around-the-clock schedules add a quiet multiplier. A 2025 peer-reviewed review of shift work fatigue found accidents cluster on night shifts and after three or more consecutive shifts, with fatigue risk rising as hours stretch. A 24/7 hyperscale schedule is a fatigue-management program whether the GC runs one deliberately or not. As Corey Jones, VP of Corporate Safety at Hunter Roberts, puts it: “When you consider the cost of an incident, the recovery is astronomical.

How do you standardize safety data across JV partners and multiple sites?

One standard, enforced at the gate, is the only model that survives a portfolio. Hyperscale programs are delivered by joint ventures and parallel GCs, and each partner arrives with its own forms, thresholds, and naming conventions. Left alone, the owner receives five versions of the truth and the carrier receives none, the fragmentation problem we described in managing a data center workforce at scale.

The fix is treating safety data like the owner treats power and cooling: one specification everywhere. That means one identity standard for every worker regardless of employer, one taxonomy for observations and incidents, one completion metric for pre-shift plans, and one dashboard the JV board and the wrap-up carrier can both read. From a system of record to an intelligence platform, the portfolio stops being a stack of binders and becomes a single queryable asset, which is what solving compliance challenges on data center projects ultimately requires.

What does an audit-ready portfolio record look like?

Audit-ready means any stakeholder, at any time, can pull verified answers about any worker, shift, or document across the portfolio in minutes. The test is not whether the GC believes it is compliant. The test is how fast it can prove compliance to an owner, a carrier, or an inspector without a records hunt.

Hunter Roberts shows what that looks like in production. The firm lifted pre-shift plan completion from 67 percent to 98 percent and cut observation time-to-resolve from 9 days to 4.2, turning two of its most audited metrics into numbers it can defend in real time.

Customer proof point: Hunter Roberts. Pre-shift plan completion rose from 67 percent to 98 percent, and observation time-to-resolve fell from 9 days to 4.2. “The difference between knowing you’re compliant and being able to prove it in real time is everything. BiltOn gave us that proof.” Corey Jones, VP of Corporate Safety, Hunter Roberts

As Omer Slavin, Co-Founder and CEO of BiltOn, puts it: “See every site, verify every worker, prove every record.”

Request a demo and we will show you how BiltOn puts this to work on your sites. Book a demo at bilton.tech/demo.

Executive takeaway

Global data center capex is projected to pass $1 trillion in 2026, and hyperscale campuses like Meta’s $50 billion Richland Parish program peak at 7,500 construction workers, so compliance now operates at industrial scale. A hyperscale GC answers to four scorecards at once: OSHA, owner mandates, wrap-up carrier requirements, and the local AHJ, and each demands its own evidence. Wrap-up programs concentrate claims on the sponsor’s ledger, so OCIP and CCIP carriers expect verified proof of who was on site and what crews did before each shift. A 6,000-worker campus generates roughly 15 million exposure hours a year, and one fatality can suspend an entire site, as the October 2025 Medina County incident showed. Hunter Roberts lifted pre-shift plan completion from 67 to 98 percent and cut observation resolution from 9 days to 4.2, the working definition of an audit-ready record.

Frequently Asked Questions

1. What is the difference between an OCIP and a CCIP?

Both are wrap-up insurance programs that cover everyone working on a project under one policy. In an OCIP the project owner sponsors the program, and in a CCIP the general contractor sponsors it. The sponsor holds the deductible and controls the claims process, which is why sponsors demand verified site records from every enrolled contractor.

2. Why do hyperscale owners add safety requirements beyond OSHA?

A hyperscaler’s brand, uptime commitments, and community relationships all ride on its construction sites, so owners impose mandates above code, such as enhanced orientations, badge-level access control, and rapid incident reporting. Meeting OSHA alone does not satisfy these programs. GCs are scored on the owner’s standard, and that score decides future awards.

3. How many workers are on a hyperscale data center site?

The largest active campuses report thousands of construction workers at peak. Meta’s Richland Parish campus in Louisiana projects 7,500 construction jobs at its height, and multi-phase programs keep thousands on site for years. At that scale, identity verification and credential governance become the backbone of safety compliance.

References

• Construction Executive. (2026). Heads or Tails: Deloitte’s 2026 Construction Industry Outlook

• Data Center Dynamics. (2025). Construction Worker Killed at Data Center Project in Medina County, Texas

• Dell’Oro Group. (2026). AI Infrastructure Buildouts and Memory Cost Inflation Drove Data Center Capex Higher in 1Q 2026

• IRMI. (2025). Navigating Wrap-Up Programs in Construction: Overcoming Challenges with Best Practices

• Meta. (2026). Deepening Our Investment in Richland Parish, Louisiana

• National Safety Council. (2024). Work Injury Costs

• Revista Brasileira de Medicina do Trabalho. (2025). Physical and Mental Fatigue in Shift Work and Mitigation Strategies: An Integrative Review

• Turner & Townsend. (2025). Data Centre Construction Cost Index 2025-2026

• BiltOn. (2026). Data Center Construction Compliance: The Regulatory Playbook

• BiltOn. (2026). Data Center Construction Compliance and Safety

• BiltOn. (2026). Construction Site Access Control: From Check-In to Verified Identity