What is the data centre contractor concentration problem?
The contractor concentration problem refers to the fact that the global hyperscale and AI data centre build-out, which runs at hundreds of billions of dollars per year, depends on a critically small number of specialist construction firms capable of delivering mission-critical facilities at scale. Industry insiders consistently identify fewer than 10 firms globally with the full-spectrum capability to lead a large hyperscale project from groundbreaking to live operations.
Those same firms appear repeatedly on the project lists of Microsoft, Google, Amazon, and Meta across multiple continents. Several are operating at or near capacity, stretched across simultaneous campuses in the United States, the Middle East, Southeast Asia, and Europe. The risk is straightforward: if one of those firms suffers a major financial or operational failure, the consequences would cascade through a supply chain with very little slack in it.
How large is the AI data centre investment wave?
The scale of demand is unprecedented. The five largest hyperscalers (Amazon, Microsoft, Alphabet, Meta, and Oracle) had combined capital expenditure growing at an average annual rate of 72% between mid-2023 and the end of 2025, nearing half a trillion dollars in 2025 alone. In 2026, Google, Amazon, Microsoft, and Meta are collectively expected to spend approximately $725 billion on capital expenditure, up 77% from the prior year’s record.
That investment is translating directly into construction activity. In 2024, the amount of data centre supply under construction in North America jumped 69% year-on-year. In the UK, early-stage data centre projects increased from 2 GW in 2025 to 8 GW in 2026. Globally, hyperscaler-operated data centres now account for 48% of worldwide capacity and are on course to represent 67% by 2031, according to Synergy Research Group.
The digital economy is being built at a pace that has no historical precedent, and it is being built by the same small group of contractors.
Why can’t more contractors simply enter the market?
Data centre construction is fundamentally different from building offices or logistics parks. The convergence of civil engineering, complex mechanical and electrical (MEP) systems, ultra-low tolerance commissioning, and critical power infrastructure requires a skills base that has taken decades to develop. It cannot be replicated quickly.
The specialist skills shortage is severe and well-documented:
- Over half of data centre operators struggle to attract and retain qualified staff, according to the Uptime Institute, which forecasts a looming shortage of hundreds of thousands of candidates globally.
- 63% of data centre executives cited a shortage of data centre-related skilled labour as their number one obstacle to securing talent, according to Deloitte’s 2025 AI Infrastructure Survey.
- Up to half of all data centre engineers may retire in the next three years, while global demand for engineers is expected to rise by approximately 300,000 over the same period.
- Data centre job postings surged 64% between 2023 and 2025, far outpacing the 4% growth in equivalent postings across the broader economy (Deloitte, 2025).
There is also a demographic problem compounding the skills gap. In the UK, 19% of data centre professionals are over the age of 55. In the US, the figure is 16%. The pipeline of replacement talent is not keeping pace.
This means that even where clients want to diversify to new contractors, the specialist workforce needed to staff those contractors does not yet exist at scale.
Who are the dominant data centre construction contractors?
A small set of firms dominates the global market for large hyperscale project delivery. In the US market, AECOM, Turner Construction, DPR Construction, Holder Construction, and Skanska USA are the major players.
Turner reported a record $44.3 billion project backlog as of late 2025, with approximately 37–40% of that tied to data centre construction. AECOM has delivered over 11 GW of data centre capacity across 45 countries over 25 years. DPR Construction has delivered more than 1.5 GW of critical data centre load capacity since 1997, with mission-critical infrastructure accounting for around 31% of its revenue in 2024.
In Europe, the picture is similar. Leading firms include Skanska, Mace, Laing O’Rourke, Mercury Engineering, Bouygues Energies & Services, and Jacobs Solutions – the same names recurring across Ireland, the UK, and continental Europe.
The market is moderately concentrated, with Turner, DPR, and AECOM typically locking in design-build contracts 24 months before groundbreaking, compressing procurement windows by up to nine months. That early lock-in further entrenches the position of established contractors and makes it harder for new entrants to compete.
What would a major contractor failure actually look like?
The construction industry has two recent and instructive precedents.
Carillion (January 2018) When Carillion went into compulsory liquidation, around 30,000 suppliers and subcontractors were caught in the collapse, collectively owed approximately £2 billion. Carillion’s liquidation triggered a 20% spike in UK building firm insolvencies; 780 companies fell into insolvency in the first quarter of 2018 alone, with small and specialist subcontractors particularly hard hit. Some projects (including PFI hospital contracts in Liverpool and Birmingham were ultimately delayed by up to five years.
ISG (September 2024) ISG’s collapse was described as the largest failure in UK construction since Carillion. The sixth-largest UK contractor by turnover, ISG had over £2.5 billion of work on-site and a further £1.7 billion in awarded contracts that had not yet started. The collapse left 33 awarded contracts, 57 in-progress projects, and three near-completion jobs in immediate uncertainty, including the Slough Data Centre Campus Phase 2 in Berkshire, a £200 million contract. ISG had also been dropped from three further data centre projects in the weeks before its administration was announced.
The aftermath was precisely what the article’s scenario describes. Clients were immediately told to find replacement contractors or face months of delay. One firm responding to the situation noted: “You don’t know what liabilities you’re going to find”. The CEO of the Building Engineering Services Association said, “the lessons of Carillion have not been learned”, and called on the government to help put a more robust industry framework in place.
The ISG case is particularly significant because it demonstrates that the risk is not hypothetical. A major UK contractor with active data centre contracts failed suddenly, leaving projects in limbo.
How have hyperscaler procurement practices made the problem worse?
The procurement strategies of the largest cloud and AI companies have, if anything, concentrated the risk further. The preference for framework agreements and preferred supplier relationships (rational for individual buyers seeking certainty of delivery) has accelerated the consolidation of work into fewer hands.
When hyperscalers lock in design-build contracts 24 months ahead of groundbreaking, smaller regional contractors with genuine capability find it structurally difficult to break in. Approved vendor lists are set years in advance and change slowly. Meanwhile, the same handful of tier-one contractors are simultaneously winning work on multi-billion-dollar campuses across multiple continents and time zones.
The result is a positive feedback loop: more concentration of work leads to greater operational strain on the surviving contractors, which increases the probability that one of them eventually suffers a failure. At which point, there is almost no one else to turn to.
Is there any industry-wide response to contractor concentration risk?
No. This is the most striking finding. Individual hyperscalers conduct their own supply chain assessments, but these are internal exercises, not shared intelligence. There is no equivalent of the financial sector’s stress-testing frameworks, no mechanism by which the industry collectively examines its exposure to contractor concentration and develops contingency strategies.
Some operators are beginning to act unilaterally. A small number of large-scale developers have quietly begun investing in in-house construction capability or taking stakes in specialist subcontractors, precisely to reduce dependence on the tier-one market. Others are working to develop second-tier contractors into first-tier capability through training programmes and joint ventures.
These are sensible responses. But they are piecemeal, and the window to diversify the contractor base without disrupting live delivery is narrowing as AI investment accelerates timescales rather than extending them.
The data centre industry has spent considerable effort stress-testing its operational resilience. It has spent far less effort stress-testing the resilience of the process by which those facilities are built in the first place.
Key statistics: datac entre contractor concentration at a glance
| Metric | Figure | Source |
| Combined hyperscaler capex, 2026 | $725 billion | Financial Times / Q1 2026 earnings |
| Growth in N. American DC construction pipeline, 2024 | +69% YoY | CBRE / industry data |
| Share of hyperscalers in global DC capacity by 2031 | 67% | Synergy Research Group |
| DC operators struggling to recruit qualified staff | >50% | Uptime Institute |
| DC executives citing labour shortage as #1 barrier | 63% | Deloitte 2025 AI Infrastructure Survey |
| DC job posting growth, 2023–2025 | +64% | Deloitte |
| ISG contracts left in limbo on collapse | £4.2 billion+ | Glenigan / Construction Index |
| Increase in UK construction insolvencies post Carillion | +20% (Q1 2018) | Moore Stephens |
