Why Data Centres Changed the Economics of Construction

General contracting is normally a terrible business: low margins, high fixed-price risk, many bidders and clients who choose on price. Data centres suspend all four conditions simultaneously, which is why they have transformed the results of the contractors positioned to build them. This analysis explains the economics and the risks. It is part of the Spain Company Stories hub.

What actually makes a data centre difficult to build?

Everything except the building. The structural shell is straightforward; the difficulty is in power distribution at enormous scale, cooling systems capable of removing heat from densely packed equipment, redundancy engineering so that failures do not interrupt service, and commissioning that proves all of it works before the client’s equipment arrives.

The electrical scope is where the value concentrates. A hyperscale facility requires substation-scale power infrastructure, uninterruptible supply, backup generation and distribution designed to fail safely, and the contractors capable of delivering it are a small subset of the general construction industry.

Commissioning is the part clients care about most and outsiders understand least. A data centre must be proven under simulated load and failure conditions before it can accept live workloads, and getting that wrong delays revenue for a client whose computing capacity is already committed.

Why does schedule matter more than price?

Because the client’s cost of delay dwarfs the construction cost difference. A technology company that cannot deploy computing capacity on schedule loses revenue, competitive position and, in the current environment, the ability to serve demand it has already sold.

That inverts the normal contracting dynamic. In conventional construction the client selects on price and the contractor absorbs schedule risk; in data centres the client selects on demonstrated ability to deliver on time and pays a premium for confidence.

For contractors this is the most favourable market condition available. It rewards track record, capacity and technical capability rather than the willingness to bid lowest, which is the mechanism by which construction margins are normally competed away.

Where is the constraint?

Skilled electrical and mechanical labour, grid connection, and equipment lead times. Every large market building data centres faces shortages of the specialist trades required, and wages for those skills have risen sharply.

Grid connection is frequently the binding constraint on the client rather than the contractor. A facility cannot operate without power, connection queues in most markets run to years, and the availability of grid capacity increasingly determines where data centres are located — which is why locations with spare generation and transmission capacity have become valuable.

That connects directly to the energy sector analysis elsewhere in this hub. Spain’s combination of cheap renewable electricity, available industrial land and Atlantic connectivity has made it a target market, constrained by exactly the grid capacity issues described in the Iberdrola case study.

What is the Spanish angle?

Two distinct opportunities. Spanish contractors are building data centres abroad, principally in the United States through ACS’s Turner and CIMIC operations, which is where the earnings impact appears today.

Separately, Spain itself is attracting data centre investment because of low-cost renewable power, mild climate in parts of the country, available land at former industrial sites and subsea cable landings. Those projects require construction, grid connection and long-term power contracts.

The domestic opportunity is smaller and strategically more valuable. Building data centres in Texas earns a margin; hosting them in Aragón or Extremadura brings investment, employment and electricity demand that improves the economics of the renewable fleet described in the analysis of Spanish power prices.

How long does this last?

Longer than a typical construction cycle and not indefinitely. Computing demand growth is genuine and the infrastructure deficit is real, which supports several years of elevated construction activity even under conservative assumptions.

The risk is that capacity is being built against demand forecasts rather than contracted demand, which is the classic pattern preceding overbuilding in any infrastructure category. Telecommunications did exactly this around 2000, and the fibre built then took a decade to fill.

For contractors the practical implication is to take the margin while it exists and avoid becoming structurally dependent on it. The groups that emerge best from this cycle will be those that used the earnings to build capability and balance sheet strength rather than those that expanded capacity to match peak demand.

How do contractors manage the risk?

By contracting structure above all. Data centre work is frequently delivered under construction management or cost-plus arrangements rather than fixed price, which transfers cost inflation risk to the client in exchange for transparency and speed.

That structure is available precisely because clients prioritise schedule. A client who needs capacity by a fixed date accepts an open-book arrangement that lets work begin before design is complete, which no cost-focused client would tolerate.

It also caps the upside. Cost-plus work earns a fee rather than a margin on savings, which is why the specialist electrical and mechanical scope matters — that is where genuine margin exists rather than a percentage fee on someone else’s costs.

What does a hyperscale project involve?

A campus rather than a building. Large facilities comprise multiple data halls constructed in phases, dedicated substations, extensive cooling infrastructure, backup generation capable of running the site independently, and security and network connectivity throughout.

Construction runs in parallel with design on compressed schedules, which requires close coordination between the client’s technical team and the contractor and generates very large numbers of change instructions. Managing that process is the core competence.

Phasing matters commercially. Clients want the first data halls energised and revenue-generating while later phases are still under construction, which means commissioning and live operation occur on the same site simultaneously — an operational complexity that few construction categories involve.

What happens when the cycle turns?

Contractors with specialist capability retain an advantage; those that added generalist capacity to chase volume do not. The electrical, mechanical and commissioning skills built during this cycle transfer to pharmaceutical, semiconductor, energy and industrial work.

The historical parallel is worth studying. Telecommunications infrastructure construction boomed around 2000 on demand forecasts that proved premature, and contractors that had geared up specifically for it suffered badly while those with transferable capability moved on.

For clients and investors, the useful signal is contracted versus speculative capacity. Data centres built against signed long-term offtake are financed differently and carry different risk from those built on the expectation that demand will arrive.

How does this affect Spanish contractors at home?

By absorbing the capability they would otherwise deploy domestically. Skilled electrical and mechanical engineers working on American data centres are not available for Spanish infrastructure or housing, which tightens an already constrained domestic labour market.

The financial flow runs the other way. Profits earned abroad support group balance sheets, fund dividends to Spanish shareholders and finance investment including in Spanish assets, so the domestic economy captures value even where the work occurs elsewhere.

For Spain the strategic question is whether it can attract the facilities rather than only build them. Hosting data centres brings investment, electricity demand and long-term employment; building them abroad brings a construction margin and nothing else.

What did the first half of 2026 show about the cycle?

That it is still accelerating. Turner booked $12.1 billion of new orders in the first quarter of 2026, up 48% on the year, and its backlog reached $48.9 billion, up 34%, with data centres the largest single driver and ten projects awarded in the year to date each valued above $1 billion.

Work-in-place revenue of $7.7 billion in the quarter, up 25%, means the backlog is converting into activity rather than accumulating, which matters because a contractor’s earnings follow execution, not awards. ACS reported group net profit of €232 million for the first quarter, up 21.5%, after €950 million for the full year 2025, and its guidance for 2026 rests explicitly on Turner’s North American digital pipeline.

The number of billion-dollar projects is the tell. A decade ago a contractor might see one such award in a good year; Turner exceeded its entire 2025 count of them within a few months of 2026. Hyperscale campuses now run to multiple gigawatts of power and several billion dollars each, and the clients commissioning them are the same handful of technology companies whose capital plans are set out in this analysis as the sector’s concentration risk.

The domestic Spanish leg has also moved. Aragón has attracted announced data centre investment running to tens of billions of euros from Amazon, Microsoft and Blackstone-backed developers, and Extremadura and Castilla-La Mancha are competing for the same projects on the basis of grid capacity and renewable supply. The gap between announcement and construction remains the constraint, and it is measured in grid connection queues rather than in contractor availability.

What would a downturn look like, and who would be exposed?

A sequence of deferred awards rather than a crash. Hyperscale clients revise capital plans quarterly, and the first sign of a turn would be projects slipping from one year’s backlog to the next, followed by margin compression as contractors compete for a shrinking pool of the schedule-critical work that pays.

The exposure varies by contractor structure. Turner and its ACS parent have a diversified American business in healthcare, education and commercial construction that predates the data centre wave, which cushions a slowdown. Specialist electrical and mechanical firms, including the Dornan business Turner acquired, are more exposed because their growth has been almost entirely digital, and their skilled workforce is expensive to keep idle.

The financial mechanism is contract structure. Cost-plus and construction management arrangements protect the contractor from cost inflation during the boom and leave it with a fee rather than an asset when the boom ends; the clients hold the risk of building capacity they may not fill. That is the reverse of the 2000 telecoms cycle, when the network builders carried the debt and the equipment vendors the receivables.

For ACS shareholders the question is what the group does with the earnings. The 2025 profit was used to reduce debt and fund a higher dividend, and the case set out in the analysis of ACS and Turner’s data centre business is that a contractor that treats this cycle as a windfall rather than a new normal will emerge from it stronger, whichever year the turn arrives.

Frequently Asked Questions