Data Center Construction Costs: The Staffing Math Behind Them
Share this article
Key takeaways
- The U.S. data center development pipeline is approaching $2.3 trillion, and average greenfield construction costs have climbed to $17.6 million per megawatt, up 21% year over year.
- MEP systems (mechanical, electrical and plumbing) absorb 30 to 50 percent of a typical data center budget, which is why skilled labor shortages move the price tag more than almost any other line item.
- A single Michigan campus, publicly described as a $16 billion project, illustrates what that spending looks like on the ground, including more than 1 million logged construction work hours on one site alone.
- Rising costs and a projected shortfall of up to 1.9 million construction workers by 2033 are pushing AEC firms to rethink how they staff documentation and coordination work, not just how they price it.
- Firms that scale drafting, BIM and coordination capacity through outsourcing can take on data center work without committing to permanent headcount for a single project cycle.
Land, power and labor keep pushing data center construction costs higher, and the staffing question behind every megaproject rarely makes the budget line.
Data Center Construction Costs Keep Climbing Toward a $2.3 Trillion Pipeline
The U.S. data center development pipeline is approaching $2.3 trillion, and it keeps getting more expensive to build a piece of it. Cushman & Wakefield's 2026 data center development cost guide puts average greenfield costs at $17.6 million per megawatt for the most modern facilities, up 21% per MW since the firm's prior report, with $492 billion of that pipeline already under construction and the remaining $1.8 trillion planned or precommitted.
Other benchmarks land in a similar range. Standard facilities run $10 million to $12 million per megawatt, while AI-optimized builds with high density racks and liquid cooling can exceed $20 million per megawatt, according to TrueLook's cost breakdown. Whichever guide a firm reads, the direction is the same: the price of building one of these facilities keeps moving up, not down.
Land, Power and Labor Drive the Price Tag Higher Every Year
Three forces explain most of the increase. The first is land, since grid connected sites are scarce, so powered parcels command a steep premium over land without secured power, and developers increasingly buy in secondary and tertiary markets to keep pace with demand.
The second is power itself. Cushman & Wakefield calls grid connected power the gating factor for new projects, ahead of land or even construction labor, because a site without secured power cannot move forward regardless of budget.
The third, and the one that touches every AEC firm working on these projects, is labor. Mechanical, electrical and plumbing (MEP) systems consume 30 to 50 percent of a typical data center construction budget, more than any other category, and TrueLook's research points to a construction industry labor shortfall of up to 1.9 million workers by 2033. AI data center work intensifies that gap because it demands electrical contractors, HVAC specialists and network installation teams with specific data center experience, not general construction skills.
A $16 Billion Michigan Campus Shows What That Spending Looks Like on the Ground
Numbers on a cost guide are one thing. A widely viewed construction video that circulated online this year put the scale in front of a much larger audience, using on screen captions to describe a specific project: a campus outside Saline, Michigan, publicly described as a $16 billion development built by Related Digital for Oracle and OpenAI, planned around three buildings at 550,000 square feet each.
The same captions noted that by late August, construction firm Walbridge reported its crews had logged more than 1 million work hours on that site alone. Whatever the final figures turn out to be once the campus is complete, the order of magnitude lines up with what the cost guides above describe: a small number of projects this size can absorb a meaningful share of the national data center construction pipeline, and each one needs a large, coordinated documentation effort behind it, covered stage by stage in how data centers get built.
Rising Costs Are Also a Staffing Problem for the Firms Documenting These Builds
Every dollar in the cost guides above eventually becomes a drawing, a coordination model or a set of as built documents that an architecture or engineering firm has to produce. A labor shortfall on the construction side does not stay contained to the trades; it shows up as pressure on the firms drafting, modeling and coordinating the MEP systems that consume half the budget, on the same accelerated timelines the project owner is paying a premium to hit.
This is the same dynamic behind the broader hiring pressure many firms report, examined in Architect Shortage or Hiring Bottleneck? What Firms Should Know: the people are often available, but a firm's own hiring process, not a true absence of talent, decides how fast it can staff up for a project this size. The cost side of that equation, including what a firm actually pays for BIM and drafting capacity, is broken down in what drives the cost of hiring a BIM modeler.
How AEC Firms Scale Documentation Capacity Without Overhiring for One Project
A single data center project, even a modest one, can require months of concentrated drafting, BIM coordination and clash detection work that does not repeat once the building is handed over. Hiring full time for that peak, then carrying the headcount afterward, is exactly the kind of overhiring firms are trying to avoid while construction costs climb everywhere else in the budget.
WorldTeams places dedicated architecture and engineering professionals inside a firm's own workflow and timezone for as long as a project needs them, matched through a defined vetting process before a firm ever sees a profile. That structure lets a firm add documentation capacity for a data center's construction phase without turning a one time staffing spike into a permanent cost. Firms scoping a project of this scale can see how that staffing model works on the architecture and engineering staffing page.
- data-center-construction-costs
- aec-staffing-capacity
- hyperscale-data-centers
Frequently asked questions
Answers to the questions readers ask most about Data center construction costs.
Why are data center construction costs rising so fast?
Costs are rising because the pipeline itself is enormous and every project competes for the same scarce land, power and skilled labor. Cushman & Wakefield puts the U.S. development pipeline at close to $2.3 trillion, with average greenfield costs up 21% per MW in a year, covered in the cost drivers section of this article and in their 2026 cost guide.
How much does it cost to build a data center per megawatt?
Standard facilities run $10 million to $12 million per megawatt, and AI-optimized builds with liquid cooling can exceed $20 million per megawatt, according to TrueLook's cost breakdown.
What share of a data center's budget goes to MEP systems?
Mechanical, electrical and plumbing systems typically consume 30 to 50 percent of the total budget, more than any other line item, which is why the staffing pressure this creates matters as much as the price of land.
Is the construction labor shortage actually affecting data center projects?
Yes. Industry estimates point to a shortfall of up to 1.9 million construction workers by 2033, and AI data center work adds pressure because it needs specialized MEP and network installation skills. The same shortage shapes hiring for architecture firms more broadly, covered in Architect Shortage or Hiring Bottleneck? What Firms Should Know.
How can an architecture or engineering firm staff up for a data center project without overhiring?
Most firms scale drafting, BIM and coordination capacity through outsourcing rather than hiring permanently for a single project cycle. See WorldTeams' architecture and engineering staffing page for how dedicated remote professionals plug into a firm's existing workflow.
