Per-MW pricing, regional variance, and cost drivers for owners scoping hyperscale & AI builds.
Salary benchmarks across the 14 mission-critical disciplines.
If you size a data center delivery team by square footage, you can miss the mark. I’d use IT load MW first, then map staffing by phase, because a 5 MW edge build, 20 MW colo job, and 60–100 MW hyperscale program do not need the same team shape.
Here’s the short version:
In simple terms, I’d start with these peak delivery ranges:
The phase pattern matters just as much as the ratio:
For a 36 MW U.S. greenfield project, that can mean roughly:
The role mix also shifts as projects get larger or denser. I’d usually put more weight on electrical, mechanical, controls, commissioning, and project engineering, while CSA is more front-loaded. And if the job uses liquid cooling or phased energization, staffing often moves up again.
Budget-wise, a rough planning rate of $15,000–$20,000 per month per mid-to-senior engineer can turn headcount into a delivery labor forecast. On a 24-month, 36 MW project, direct engineering labor can land near $8.6 million before travel, relocation, and pay premiums.
Below, I’d break the model into the three things you need to act on fast: ratio, phase, and role mix.
Data Center Delivery Staffing Ratios by Project Scale (Engineers per MW)
Start with a baseline ratio. You need a working estimate for how many delivery-team FTEs per MW of IT load the project will need during delivery.
That number will move as the project takes shape. But starting without one is how teams get caught short when pressure is highest. Use this baseline to size each project phase in Step 2.
Don't size delivery teams using steady-state operations ratios.
Delivery staffing covers roles like design engineers, construction project engineers, commissioning engineers, and owner's technical representatives. Those jobs support a different phase of the asset life cycle, so they need a different staffing model than operations.
Delivery is temporary and front-loaded. Operations is steady-state. Treat them that way.
Build a separate model for delivery staffing, fund it directly in project capex, and keep it separate from future ops headcount.
Not every job starts from the same point. A 5 MW edge build, a 20 MW colocation deployment, and a 60–100 MW hyperscale program each need a different staffing density per MW.
A 5 MW edge facility usually has less standardization, more site-specific issues, and smaller local contractor teams. That tends to push peak delivery staffing higher, to about 0.8–1.2 delivery-team FTEs/MW at peak across all delivery functions.[1][4]
A 20 MW colocation deployment gets some help from repeatable MEP topologies, standardized hall layouts, and established playbooks. That usually brings the range down to about 0.5–0.8 delivery-team FTEs/MW.[1][4]
At the 60–100 MW hyperscale end, reference designs, modular power rooms, and seasoned owner PMO teams can push delivery density down to about 0.35–0.6 delivery-team FTEs/MW.[1][4]
A few factors can move those numbers in either direction:
Use these ranges as planning benchmarks, then adjust for your risk profile, schedule, and delivery model.[1][4]
These phases don't all peak at once. Design hits hardest early. Construction and commissioning overlap in the middle. Owner oversight runs across the full life cycle.
So the total delivery E/MW at any one moment will be lower than the sum of all four columns. Think of these peaks as your hiring waves. Step 2 converts them into phase-by-phase headcount.
Use the baseline from Step 1 to turn MW into headcount by phase. The key point is simple: apply a ratio to each delivery stage, not one blended ratio for the whole job. That gives you hiring numbers for design, construction, commissioning, and owner oversight that line up with how data center delivery actually works.
For a typical U.S. greenfield data center, a good planning range is about 0.3–0.6 design engineers per MW. That usually covers electrical, mechanical, CSA, controls, and project engineering from concept design through construction documents and procurement support.
In most cases, electrical takes the biggest slice. Mechanical comes next, then controls/BMS, CSA, and project engineering.
One choice can shift the math in a big way: whether design is run from one central campus team or split out by building.
On a multi-building hyperscale campus with standardized reference designs, a central team of 12–15 engineers can support 60–120 MW across three or four similar halls. That can bring the effective campus-level ratio down to about 0.12–0.20 engineers per MW.
Project-dedicated engineers make more sense when halls use different cooling strategies, when design and construction overlap on a tight schedule, or when the project needs fast, hands-on RFI turnaround.
Design teams usually peak first. Field teams and startup teams hit their peak later.
Construction-phase engineering support usually lands around 0.2–0.4 engineers per MW. Early in construction, especially during MEP rough-in, staffing often sits closer to 0.15–0.25 engineers per MW. As equipment setting and startup prep ramp up, that can move toward 0.20–0.35 engineers per MW.
At that point, the work shifts more toward field coordination, issue resolution, and startup readiness.
A typical construction support mix looks like this:
Commissioning is where staffing intensity jumps again. Plan for 0.25–0.50 commissioning engineers per MW during startup, IST, and turnover. A commissioning authority or lead Cx manager usually runs the team, backed by electrical, mechanical, and controls/BMS specialists. Commissioning labor usually bunches up in a 3–6 month window [5][6].
Owner-side oversight is lighter, but it stays in place through the full delivery timeline. A normal range is 0.05–0.15 engineers per MW for owner's reps, QA/QC engineers, and technical PMs. That number tends to move up when acceptance criteria are strict or reporting demands are heavy.
A worked example makes the hiring waves easier to see. For a 36 MW single-phase U.S. greenfield project, mid-range ratios translate like this:
If the schedule gets compressed, push the design and commissioning ratios toward the high end.
Use the headcount by phase from Step 2 to divide staffing by discipline. Headcount per MW tells you how many FTEs to plan for. Role mix tells you which people you need and when you need them.
That second part matters more than many teams expect. A project can look fine on total staffing and still miss key milestones if the discipline mix is off. One weak spot in the wrong phase can knock the delivery team off the critical path.
Six roles sit at the center of most U.S. data center delivery teams, and each one tends to scale with MW capacity.
Electrical engineers own the power side of the project: utility intake, medium- and low-voltage distribution, UPS, generators, PDUs, busway, protection coordination, and fault/arc-flash studies. Their work runs from early design through energization and integrated systems testing. When electrical coverage is too thin, the warning signs show up fast: RFIs stack up, submittals slow down, and late one-line revisions can hold up energization. [7][8][11]
Mechanical engineers own the cooling plant. That includes chillers, CRAHs/CRACs, pumps, hydronic systems, airflow, and liquid cooling loops. Their busiest stretch usually lands in design and startup. Thin mechanical coverage often leads to incomplete sequences of operation, which then drives late changes and retesting that squeezes the commissioning window. [7][8][11][16]
Controls/BMS engineers own sequences of operation, points lists, cause-and-effect matrices, alarms, trend logs, and the links between BMS, EPMS, and client monitoring platforms. Their scope runs across FAT, SAT, and commissioning tuning. If you don't have enough controls support, commissioning can stall before it even gets going because BMS vendors are waiting on engineering direction before programming starts. [7][8][10][11][13]
Commissioning leads and engineers own test plans, field execution, issue tracking, and turnover documents from Level 1–5 checks through integrated systems testing (IST). When this team is too small, test windows pile on top of each other, retests go up, and turnover documents aren't ready when the client asks for them. [7][9][12][13][14][15][17]
CSA engineers (civil, structural, architectural) are loaded toward design and early construction. They handle site layout, grading, drainage, foundations, structural steel, equipment pads, and roof loading. Their workload usually drops after the superstructure is done, but on multi-building campuses they often stay busy as new halls and equipment yards come online.
Project engineers keep the machine moving. They handle RFIs, submittal reviews, clash resolution, change management inputs, and equipment tracking logs across design and construction. When this role is stretched too far, RFI backlogs grow, field workarounds spread, and on tight schedules, slow turnaround can delay both procurement and installation. [8][11]
Use the table below as a starting point, not a hard rule. Then adjust for cooling approach, redundancy setup, and schedule speed.
Schedule compression puts the most pressure on electrical, controls, commissioning, and project engineering. When design and construction overlap, you need parallel workstreams. One electrical engineer may be closing out design and utility interfaces while another handles field RFIs and construction support. On a compressed 20 MW project, that can push electrical staffing from 1–2 engineers to 3–4. Commissioning teams also need to come in sooner. Bringing a commissioning lead on during construction helps make sure test plans and pre-functional checklists are ready as soon as equipment is ready for startup. Same MW, different schedule, very different staffing need. In these roles, a compressed timeline can push demand to 1.5–2x the baseline engineers per MW. [8][18]
Liquid cooling and high-density IT loads shift the mix toward mechanical and controls. Direct-to-chip cooling, rear-door heat exchangers, and immersion systems add pumps, manifolds, heat exchangers, and fluid-handling scope that a standard HVAC mechanical engineer may not be able to cover alone. As rack density goes up, teams often need a dedicated liquid cooling mechanical engineer alongside the main HVAC lead. Controls scope grows too. High-density zones call for tighter monitoring, adaptive control logic, and closer links between BMS, EPMS, and client monitoring platforms. [11][16]
Multi-building campuses and phased energization keep every role active for longer and pull commissioning engineers into the project earlier across several live tracks at once. That role mix then feeds directly into hiring order, budget timing, and recruiting priorities in Step 4.
Use the phase-based role mix from Step 3 to decide who to hire first, when budget needs to hit, and which roles need the most recruiting attention.
Time hiring around project milestones, not just calendar dates. That keeps the right people in place before big design and delivery choices get locked in.
A simple way to think about it: early hires shape decisions, mid-project hires keep the build moving, and late-stage hires protect turnover and startup.
After you map out the headcount curve, turn it into monthly burn. That way, budget peaks line up with the heavy design, construction, and commissioning phases instead of catching finance off guard.
Start by assigning each role a fully loaded monthly cost. Based on 2026 U.S. benchmarks, data center MEP engineers often earn about $95,000-$140,000 per year, mission-critical project managers commonly earn $120,000-$180,000, and commissioning engineers typically range from $98,000-$155,000 at mid-level to $140,000-$210,000+ at senior level [19][20]. A practical planning number is $15,000-$20,000 per month fully loaded for a mid-to-senior engineer.
For a 24-month, 36 MW project, a rough cash-flow model could look like this:
That comes to about $8.6 million in direct engineering labor before contract premiums, travel, or relocation.
Those extra costs can add up fast. Relocation for greenfield sites in secondary U.S. markets can add $15,000-$25,000 per engineer for full moves, or $2,000-$5,000 per month for rotational travel and lodging. On top of that, mission-critical engineers with hyperscale or Tier III/IV experience often command 10%-25% pay premiums.
Once the budget is mapped out, the next problem is simple: can you fill the hardest roles before the schedule starts slipping?
Specialized recruiting support can take an MW profile and milestone schedule and turn them into a role-by-role hiring plan. That usually includes sourcing project engineers, MEP leads, commissioning managers, controls specialists, schedulers, and owner-side technical PMs.
For multi-phase programs, this kind of support can also keep a live pipeline in place over several years of delivery. That matters because recruiting works best when it follows the budget and build schedule from the start, not when the team is already short and trying to catch up.
Engineers per megawatt (E/MW) gives teams a simple way to turn IT load MW into the headcount and role mix needed for data center design, construction, commissioning, and owner-side technical oversight.
One point matters from the start: delivery staffing is not the same as operations staffing. Delivery is tied to the project phases. Operations starts after handover. Mixing the two can muddy planning and throw off hiring.
The ratio also shouldn’t stay fixed from start to finish. Teams need to recalibrate it based on project type, complexity, and schedule pressure at each phase gate. A fast-track build won’t need the same staffing shape as a more standard program, especially when mitigating schedule risks.
As MW goes up, the staffing mix should lean more toward electrical, controls, commissioning, and project engineering. At the same time, CSA roles tend to stay more concentrated in the early phases. And when complexity goes up, the answer isn’t just adding more designers. It usually means adding more integration and commissioning leadership. This shift requires construction managers with specialized skills to oversee the technical handoff.
Phase gates help teams time hiring with a lot more precision. From there, the staffing curve can be turned into budget plans and recruiting plans, which makes the whole model much more useful than a rough headcount guess.
That’s also where specialized recruiting support comes in. It can help pre-build pipelines for commissioning, electrical, controls, and project engineering roles before those jobs turn into critical-path risks.
Put simply, E/MW is a common planning standard that converts MW capacity into staffing, recruiting, and budget decisions for data center delivery.
Use milestone-based workforce planning instead of a fixed E/MW multiplier. In plain English: staff the job around what has to happen next, not just how many seats are open.
That matters most for key roles like MEP leads and commissioning leads. Bring them in when the project hits the phases where their decisions shape schedule, quality, and handoff. If you wait until headcount models say it’s time, you’re often already behind.
For fast-track builds, a blended team usually works best: core staff plus contract specialists. Keep the core group steady, then add targeted outside help where the schedule gets tight or the work gets more technical.
A few staffing moves matter more than others:
That last point is easy to overlook. On a simpler build, a looser ratio may hold up for a while. On a hyperscale job, though, too many direct reports can slow decisions, weaken field oversight, and let small issues turn into schedule problems.
The transition needs to start well before the facility goes live if you want a smooth handoff.
That means hiring operations-facing roles - like facilities managers and operations leads - 6 to 9 months ahead of the planned IT load date.
Those people should be on board before Integrated Systems Testing (IST) starts. Why? Because they play a direct part in shaping the turnover package, including:
If they come in too late, the team can miss details that matter once the facility is up and running.
To cut risk and keep the job moving, bring key leaders and project controls people on at project kickoff - ideally 4 to 6 months before site mobilization. Start with the Project Executive, Development Manager, Owner’s Representative, Cost Manager, and Project Controls Lead or Scheduler.
Loop in MEP and commissioning leads during design and preconstruction. That gives the team time to spot space clashes, confirm long-lead equipment needs, and map out integrated systems testing early.
