Labor Is the Bottleneck, Modularization Is the Solution, and Vertically Integrated Players Capture Value

Overview

The report argues that the pace of U.S. data center expansion is being constrained by insufficient supply of MEP tradespeople. The industry is shifting from a traditional model fully dependent on on-site construction to a prefabricated and modular model, moving repeatable, labor-intensive, and complex work into controlled factory environments, with only relatively simple final assembly performed at the project site. At the same time, non-hyperscale operators are accounting for a rising share of new projects. Their internal engineering capabilities and procurement scale are typically weaker than those of hyperscale operators, so they are more inclined to delegate design, system integration, and equipment selection to equipment vendors or MEP contractors.

Core views

Modularization first expands data center construction capacity by shortening construction cycles and reducing on-site labor requirements; second, it increases the per-project revenue share of integrated equipment vendors and integrated contractors through system-based procurement; and finally, it drives margin expansion through value-based pricing, a higher share of engineering business, and greater execution responsibility. ETN, SU.FP, and VRT, which have electrical, thermal management, and infrastructure management capabilities, as well as LGN and PWR, which can cover the full workflow from design, prefabrication, installation, and commissioning, are best positioned to capture this profit pool migration; companies that only provide on-site labor or a single product face relative pressure.

Analysis framework

Starting from annual additions to U.S. data center capacity, the report estimates the construction ceiling implied by the pace of MEP tradesperson hiring, then compares traditional on-site construction and modular construction in terms of construction period, risk, predictability, and cost. It then combines changes in rack power density, 800V DC architecture, and customer mix to analyze how procurement authority migrates along the value chain, and screens potential beneficiary companies based on product coverage and engineering service capabilities.

Methodology notes

• Supply-Demand and Bottleneck AnalysisConstruction Capacity Constraint Estimation

  Estimating the ceiling of annual data center construction capacity based on the supply of MEP tradespeople

  Links the pace of MEP tradesperson hiring with the labor input required per unit of construction activity, and compares it with potential supply capacity in other links such as power generation and chips to identify the key constraints on data center expansion.

• Scenario AnalysisModularization Penetration Scenario

  Comparing changes in the shares of traditional on-site construction and modular construction

  Combines construction cycles, execution risks, design standardization, rack density, and changes in power supply architecture to infer a possible path for modularization's share to rise from 20% in 2020 to about 60% by the end of the decade.

• Value Chain AnalysisProfit Pool Migration Analysis

  Analyzing how value in design, manufacturing, procurement, and construction migrates from the site to factories and engineering departments

  Modularization reduces on-site labor input but increases the importance of prefabrication, system design, factory testing, and integration, causing more value to flow to integrated equipment vendors and integrated MEP contractors.

• Customer Mix AnalysisProcurement Decision-Making Authority Migration

  A higher share of non-hyperscale operators changes the equipment procurement model

  Compared with hyperscale operators, which have internal engineering teams and prefer best-of-breed individual products, other operators are more likely to use outsourced engineering and complete-system procurement, thereby expanding the decision-making authority and per-project revenue share of equipment vendors and contractors.

• Competitive Landscape AnalysisVertical Integration Capability Comparison

  Screening beneficiaries by product breadth and full-project workflow capability

  Equipment vendors are mainly compared on coverage across electrical, thermal management, and infrastructure management software, while contractors are compared on full-workflow capabilities from conceptual design, prefabrication, and on-site installation to commissioning and delivery.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• ETN

  Core beneficiary electrical equipment vendor, maintained at Outperform

  Strengths

  Can shift from sales of individual electrical products to delivery of modular power and cooling systems, while covering more complete data center infrastructure needs.

  Weaknesses

  System-based delivery increases responsibility for engineering coordination, manufacturing, and project execution.

  Comparison

  Compared with equipment vendors with narrower product lines, it is better positioned to benefit from procurement shifting from best-of-breed individual products to complete solutions.

  Risks

  Modularization penetration falls short of expectations, project execution costs rise, or customers continue to lead individual-product procurement.

• SU.FP

  Core beneficiary integrated equipment vendor, maintained at Outperform

  Strengths

  Has power, cooling, control, and modular data center solutions, enabling it to meet customer demand for prefabricated and pre-tested complete systems.

  Weaknesses

  Complex system delivery may increase customization and fulfillment management difficulty.

  Comparison

  Its integrated product and software capabilities are superior to competitors that cover only a single power or cooling link.

  Risks

  Project cycle delays, system integration risks, and non-hyperscale customer demand below expectations.

• VRT

  Core beneficiary data center power and thermal management equipment vendor, maintained at Outperform

  Strengths

  Power, thermal management, and infrastructure management capabilities fit the modularization needs of high-density artificial intelligence data centers.

  Weaknesses

  Highly affected by high-density data center capital expenditure, delivery capabilities, and the pace of project execution.

  Comparison

  Compared with single-product suppliers, it is better suited to take on system-level procurement and one-stop infrastructure needs.

  Risks

  Slowdown in artificial intelligence infrastructure investment, delays in modularization orders, or cost overruns caused by execution responsibilities.

• LGN

  Core beneficiary MEP contractor, maintained at Outperform

  Strengths

  Can cover engineering design, prefabrication, on-site installation, and commissioning, engaging from the project concept stage and providing turnkey services.

  Weaknesses

  A higher share of engineering and prefabrication requires continued expansion of manufacturing sites, technical personnel, and project management capabilities.

  Comparison

  Its full-workflow capability is superior to contractors that only provide on-site construction or partial MEP services.

  Risks

  Capacity expansion not timely, rising fixed costs, project delays, and execution risks.

• PWR

  Beneficiary of integrated MEP project capability, maintained at Market-Perform

  Strengths

  Can handle complete MEP projects and provide integrated services from design and prefabrication to installation and commissioning.

  Weaknesses

  Although the report recognizes its strategic position, its rating is lower than Outperform-rated names such as LGN.

  Comparison

  Its degree of vertical integration leads on-site labor-type contractors, and it is grouped with LGN among the report's most favored integrated contractors.

  Risks

  Valuation digestion, project cost control, labor and prefabrication capacity expansion risks.

• DLR, EQIX

  Operator-side beneficiaries of modular construction, both rated Outperform

  Strengths

  A faster and more predictable construction process helps shorten capacity delivery cycles and supports data center expansion.

  Weaknesses

  The report does not compare the two companies' respective modular engineering capabilities in detail.

  Comparison

  Compared with equipment vendors and contractors, they mainly benefit from obtaining operable capacity more quickly, rather than directly capturing profit pool migration.

  Risks

  Power access, cost of capital, project approvals, and changes in leasing demand may still constrain expansion.

Key data

• U.S. annual data center construction capacity12 GW/year in 2026, 35 GW/year in 2030Estimated based on the current pace of MEP tradesperson hiring, implying a compound annual growth rate of about 30%.
• Share of modular versus traditional construction20%/80% in 2020, currently about 40%/60%, and potentially 60%/40% by the end of the decadeThe former refers to modular construction, and the latter to traditional on-site construction.
• Potential construction period reductionUp to about 60%Factory manufacturing can proceed in parallel with on-site civil construction, with complex operations completed in controlled environments.
• Hyperscale operators' share of construction55% in 2024, currently about 40%, and potentially declining to about 20% in the coming yearsThis implies that the share of non-hyperscale operators may rise to about 80%, driving outsourcing of procurement decisions.
• Initial labor impact of 800V DC architectureMEP labor hours may increase by about 50% during the adaptation phaseHigher complexity and safety requirements further enhance the economics of adopting factory-prefabricated solutions.
• Emerging artificial intelligence infrastructure and power reserves of mining companiesPlanned power portfolio exceeds 30 GWOver the past two years, more than 8 GW has been contracted with hyperscale operators, neocloud service providers, artificial intelligence labs, and chip manufacturers, with related transaction value exceeding $170 billion.

Impact & implications

Modularization not only increases the buildable scale of data centers, but also changes the distribution of revenue and profit across the value chain. Integrated equipment vendors can shift from selling individual products to selling power, cooling, and control systems, increasing equipment value per megawatt; integrated MEP contractors can increase engineering design and prefabrication business, and expand their share of customer spending through turnkey solutions. Because these companies assume greater responsibility for schedule, quality, and integration, their pricing and margins may also improve. Conversely, equipment vendors with narrow product lines and contractors that rely only on on-site labor face rising competitive barriers and loss of business share.

Risks

• The adoption pace of modularization is lower than expected, as customers continue to choose traditional on-site construction due to customization needs, transportation constraints, or existing processes.
• The share of non-hyperscale operators in new projects does not rise to about 80% as forecast, so procurement authority remains within large operators.
• Equipment vendors and contractors assume more system integration and delivery responsibilities, potentially facing quality, schedule, cost overrun, and warranty risks.
• Insufficient expansion of prefabrication factories, engineering talent, and supply chains may create new capacity bottlenecks.
• The complexity and safety requirements of 800V DC architecture are higher than expected, potentially increasing labor input and delaying projects in the short term.
• Even if the MEP labor bottleneck eases, power access, generation capacity, chip supply, financing, or approval constraints may still depress actual construction volume.
• If relevant beneficiary names have already fully reflected high-growth expectations for artificial intelligence infrastructure, orders or margins below expectations may trigger valuation pullbacks.

What to watch

• Whether annual U.S. data center capacity additions can grow along the path from 12 GW to 35 GW.
• Whether the share of modular construction can rise from about 40% currently to about 60% by the end of the decade.
• Changes in MEP contractors' prefabrication factory area, capacity utilization, and number of engineering personnel.
• Changes in the shares of hyperscale operators and other operators in the project pipeline.
• Whether equipment procurement further shifts from individual product selection to complete-system procurement of power, cooling, and software.
• ETN, SU.FP, and VRT's system business mix, orders, backlog, and data center revenue per megawatt.
• LGN and PWR's engineering and prefabrication revenue mix, project margins, and integrated orders from design to commissioning.
• The implementation pace of 800V DC architecture and its actual impact on on-site MEP labor hours.