Per-MW pricing, regional variance, and cost drivers for owners scoping hyperscale & AI builds.
Salary benchmarks across the 14 mission-critical disciplines.
The short answer: hyperscale sites are usually built under tight control from one cloud or AI owner, while colocation sites are usually built by an operator that leases space to many tenants.
If I strip the article down to what matters most, it comes to this:
This split affects who leads design, who manages utilities, who handles tenant work, and what kind of people you need to hire. For example, hyperscale teams often need people with 30–50 MW campus exposure and $500 million+ project history. Colocation teams often need people who can work in live sites, manage 5–10 MW phases, and hit 6- to 18-month schedules.
At the market level, demand is high. U.S. primary markets had 6,350 MW under construction at the end of 2024, and many hard-to-fill data center roles take 60+ days to staff. So if you hire for the wrong model, cost and schedule can slip fast.
Quick comparison
If you want to know who builds each type, I’d say it this way: hyperscale is owner-led construction at campus scale; colocation is operator-led construction built around lease demand and tenant timing.
Hyperscale vs Colocation Data Centers: Builder Stack & Hiring Guide
The main difference comes down to who runs delivery: a single owner in hyperscale, or the facility operator in colocation.
Hyperscale projects are single-operator campuses built for one cloud or AI owner. Companies like AWS, Microsoft, and Google design and fund these campuses to support their own workloads.
What makes these projects different is sheer size. Many new hyperscale builds start at about 40 MW and often go past 50 MW of IT load per facility. Across several buildings, campuses often total 100 MW to 300 MW or more. AI-heavy campuses can reach 500 MW to 1,200 MW, which puts major demand on utility systems.[10][11]
At that size, the work has to start early with utilities. It also puts pressure on long-lead equipment buying and phased campus planning.[3][5][8]
Because one operator owns the roadmap, delivery is owner-led from the start. The hyperscaler sets the design standards, picks the site, and works directly with utilities on high-voltage transmission and substation infrastructure. In plain terms, the campus plan shapes the whole job from design through commissioning.
That kind of control also changes who leads the work. The owner-side team stays close to the action, alongside EPCs, major MEP trades, and utility partners. So hiring has to line up with utility work, MEP scope, and campus-scale decisions.
Colocation operators build and lease powered space to many tenants while keeping the base facility up and running. Operators like Equinix and Digital Realty serve a mix of enterprise, cloud, financial, and government tenants in the same building.[4][6][7]
This is a portfolio-driven model. These operators roll out repeatable building types across several U.S. metros, and many are set up as REITs or infrastructure funds.[4][9] The model depends on repeatable delivery, pre-lease demand, and phased expansion across markets.
That setup shifts more day-to-day responsibility to repeat-build GCs, trade partners, and tenant-facing PM teams. Since the operator leads delivery, hiring has to support repeat builds, live-site work, and tenant coordination.
Those delivery differences shape which builders, PM layers, and trade partners take the lead on each project.
The team setup changes based on the model. And even when job titles look the same, the work behind those titles can be very different on a hyperscale campus versus a colocation project.
On hyperscale projects, the owner usually stays close to the work. The chain often runs from the owner to a developer or JV partner, then to a mission-critical general contractor or EPC, and then to major MEP trades.
A lot of hyperscale owners use a build-to-suit developer or JV partner to handle land, entitlements, and early site work. After that, they bring in a mission-critical GC or EPC to deliver buildings in parallel. That setup makes sense when speed matters and multiple buildings are moving at once.
Common GCs on hyperscale campuses include Turner Construction, Mortenson, and DPR Construction. On the MEP side, firms like Rosendin Electric and Bel-Aire Mechanical take on electrical and mechanical scope with large craft crews and fast-track schedules. That owner-led setup also shapes hiring and project controls later in the job.
Colo delivery works a bit differently. It's more repeat-build and more driven by tenant needs.
On colo projects, the operator usually runs the delivery program through long-term relationships with repeat-build general contractors such as DPR, HITT, and JE Dunn. The chain typically runs from the operator to the repeat-build GC, then to a tenant-facing project management team or owner's rep, and then to trade partners.
Because more than one tenant shares the same building, operators often keep internal or outsourced owner's rep teams in place. Those teams help manage tenant improvements, cage fit-outs, and cross-connects without disrupting active halls. Trade partners like Rosendin and Titan Electric show up here as well, but the focus shifts. Instead of greenfield speed, the work leans more toward phased activity in live sites, tenant improvements, and tightly coordinated turnarounds.
The contract is what tells you who is actually calling the shots on design, procurement, and change control. Put simply: the contract model decides who leads.
EPC and design-build are common on hyperscale projects. CMAR shows up more often in colo.
Under EPC, one entity handles design coordination, procurement, and construction. That gives hyperscale owners a way to standardize delivery across regions while keeping day-to-day execution off their plate. On the colo side, CMAR lets operators stay closer to design decisions and shift scope as tenant demand changes. That's useful when one tenant wants different power densities, cage layouts, or phased hall turnovers. The flip side is simple: the operator takes on more coordination work.
Hiring usually follows the way the project gets delivered. Hyperscale builds need campus-level leaders. Colo builds need teams that can work in live facilities and deal with tenant needs without missing a beat. Even the same job title can mean very different day-to-day work depending on the model.
Start with hyperscale. At that level, campus scale shapes almost every hire.
On an owner-led campus, with EPC partners and major MEP trades already in place, the hiring focus shifts toward leaders who can run work across an entire campus, not just one building.
A hyperscale project executive may oversee several phases at once, line up procurement with construction, manage MEP and commissioning, and keep a regional capacity roadmap on track.
The core roles usually include campus program managers and MEP directors with deep experience in medium-voltage distribution, generator yards, chilled water systems, and advanced liquid cooling across standardized designs used in more than one region. Commissioning directors sit alongside construction leaders, not behind them. Cost and schedule controls managers handle multi-year capital portfolios that can stretch across several phases and sites.
With U.S. primary markets hitting 6,350 MW under construction at the end of 2024 - more than double 2023 levels - the race for this talent is intense.[2][12]
Hyperscale reporting is also more layered. Campus MEP and commissioning leaders usually report into program leadership, while controls teams answer to both project leadership and PMO oversight.
That setup creates a hiring pattern that looks very different from the colo side.
Colo hiring is built around repeat delivery, tight schedules, and live-site execution. These projects are often operator-led, which changes the mix of people you need on the ground.
Project managers often run several smaller phases at once, usually in the 5–10 MW range, using standard design templates. Superintendents and field managers need direct experience with method-of-procedure shutdowns, often done at night or on weekends, without affecting active data halls. That kind of work leaves no room for sloppy handoffs.
Commissioning managers work on shorter and more frequent cycles, usually tied to tenant move-in dates. MEP coordinators take standard designs and adjust them for tenant-specific power densities. And one role stands out here: the operations-linked construction leader. This person helps construction teams and active facility operations work side by side.
Because tenant SLAs are always sitting in the background, experience with tenant-facing change management can separate one candidate from the pack.
Those differences show up fast when recruiters start screening resumes.
The first filters should be simple: scale delivered, site conditions, and contract side.
For hyperscale roles, look for people who have handled:
For colo roles, the better signal is different. Prioritize candidates with multi-phase delivery history, live-site experience, and 6- to 18-month fast-track schedules.
Contract background matters too. Owner-side and EPC experience usually lines up better with hyperscale. Repeat-GC and CM-at-risk backgrounds tend to fit colo more cleanly.
A simple way to think about it:
Greenfield work means site development and utility buildout. Live-site work means protecting an active data hall.
Those delivery differences should shape the hiring plan before any requisition opens.
A job title by itself doesn't tell you much. Before you start the search, lock down the role's scope, reporting line, site condition, and who owns commissioning. The builder stack matters too. Who owns delivery? Who manages the EPC or GC relationship? How is the contract set up? Those details should shape the search brief before the first resume gets a look.
Staffing also needs to line up with the project phase. Preconstruction calls for development and entitlement talent. Multi-phase campus work calls for program and controls leadership. Live-site retrofit work calls for operations-minded PMs with MOP experience.
That same thinking should carry straight into the recruiting brief.
Once the role profile is clear, the search can line up with the project's delivery model.
Only about 15% of applicants meet minimum qualifications for data center roles, and many critical roles take 60+ days to fill.[1][13] iRecruit.co supports owners, developers, GCs, owner's reps, and mission-critical trades with hiring for construction management, MEP, commissioning, scheduling, cost, and project executive roles. The focus stays on the delivery model. Hyperscale campus searches need campus-scale governance experience. Colocation searches need live-site coordination and tenant-facing experience.
Bring the search in early, at preconstruction or funding approval, so staffing stays in step with the schedule.
The builder stack is the clearest sign of the talent a data center project needs. Hyperscale campuses are owner-driven, single-operator programs. Colocation facilities are operator-led, multi-tenant assets built to lease capacity.
That setup shapes everything: where risk sits, who makes the calls, and which hires matter most. In hyperscale, the focus is campus-scale controls. In colocation, it’s live-site coordination and tenant-facing execution.
A title like Project Manager or MEP Lead doesn’t tell you much by itself. The real question is where that role sits in the builder stack, what risk it carries, and whether the site is greenfield or occupied. Hiring to match the delivery model - not just the job description - is what separates teams that deliver from teams that miss the mark.
Start with the delivery model. Then build the search brief, screening process, and hiring plan around it.
Look at the scale, ownership model, and pace.
Hyperscale projects are large, single-operator campuses built for one client. They usually range from 20 MW to more than 1.2 GW of IT load, and project costs often run above $200 million.
Colocation projects serve multiple tenants. They tend to span 10,000 to 50,000 square feet and usually cost $30 million to $150 million. The focus is different too: these builds lean more toward tenant-specific needs and phased turnover.
They differ because hyperscale and colocation projects focus on different things: speed, scale, risk, and control.
Hyperscale builds often use Design-Build, EPC, or multi-prime models to move faster and get key partners involved early. Colocation builds more often lean toward tighter owner oversight and more flexibility for tenant-specific fit-outs, especially in live environments.
Direct mission-critical experience matters most. A general construction background often falls short because data center work depends on uptime, redundancy, and tight systems integration.
For hyperscale, look for teams that have handled campus-scale delivery, repeatable programs, and high-density power and cooling. For colocation, put the focus on tenant-facing turnover, SLA performance, and coordination inside live, occupied spaces.
In both cases, check certifications, review BMS and EPMS experience, and confirm the team is ready for shift work.
