The question is not whether data centers will be built, but whether AI can produce enough annual revenue to support the infrastructure investment now being projected. A Construction Dive interview with Bain partner Peter Hanbury, published Oct. 8, 2026, put the implied hurdle at an AI market approaching $6 trillion in annual revenue by 2031. That is a reported projection—not a settled forecast—and it depends on both economic growth and the ability to deliver power, equipment, labor and permits at scale.
What does the $6 trillion figure mean?
It is an estimate of the annual AI-market revenue that could be needed to support the projected buildout, not the construction budget. Construction Dive reported Bain estimates of $5 trillion to $6.5 trillion in cumulative data center spending through 2030 and about $1.5 trillion in annual AI infrastructure spending by 2031. In the interview’s framing, if capital expenditure is roughly 25% of industry revenue, an AI market approaching $6 trillion in annual revenue by 2031 would be needed to sustain that level of spending. Construction Dive’s Oct. 8, 2026 interview attributes these figures to Bain; the underlying methodology for these specific estimates is not established in the interview.
As an Amazon Associate I earn from qualifying purchases.
The interview also cites a Bain estimate of roughly $780 billion in 2026 capital expenditures by Microsoft, Google, Amazon, Meta and Oracle. That amount is not all data center spending, so it should not be read as a direct measure of construction investment. These are distinct estimates with different scopes and time periods:
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →| Reported estimate | What it describes |
|---|---|
| $5 trillion–$6.5 trillion through 2030 | Cumulative data center spending attributed to Bain |
| About $1.5 trillion by 2031 | Annual AI infrastructure spending attributed to Bain |
| Nearly $6 trillion by 2031 | Annual AI-market revenue implied by the interview’s roughly 25% capex-to-revenue assumption |
| Roughly $780 billion in 2026 | Capital expenditures by five named technology companies; not all is data center capex |
The figures are Bain-attributed estimates reported in the 2026 interview, not independently verified outcomes. A separate Bain analysis of data center energy needs says meeting global demand could require more than $2 trillion in new energy generation resources. That is broader energy context, not an interchangeable measure of the interview’s specific data center or AI projections.
#1 Best Overall
What must go right for the investment case to work?
Hanbury’s argument rests on more than continued efficiency gains from AI. To justify infrastructure spending at this scale, the market would need to develop new revenue and sources of economic value, while the physical and financial systems supporting data centers expand. The interview identifies four conditions:
- AI creates new value: Beyond productivity improvements, applications such as autonomous systems, physical AI, new consumer experiences, and AI-enabled products or industries would need to generate meaningful economic activity. These are possibilities, not guaranteed growth.
- Physical constraints ease: More power generation, faster grid interconnection, behind-the-meter capacity, storage and better coordination would help address electricity bottlenecks.
- More forms of financing and risk-sharing emerge: Infrastructure investors, utilities, sovereigns and governments could broaden the pool of participants bearing the cost and risk.
- Project selection becomes more disciplined: A large pipeline is not enough; projects need credible demand and viable delivery plans.
Construction Dive also reported a Bain-attributed count of at least 75 projects worth $130 billion blocked or delayed in the first quarter of 2026, nearly matching the impact in all of 2025. The interview does not establish the underlying methodology for that estimate, so it is best understood as a reported indicator of project friction rather than a complete measure of all delayed construction.
Why power can set the construction schedule
For a large data center campus, a site can be ready on paper while the electricity needed to operate it remains unavailable. Hanbury estimated that adding major grid capacity can take four years or more. Substations, transmission upgrades and interconnection therefore belong on the project’s critical path alongside the building itself; some programs may also coordinate onsite generation and storage.
Free tools Windows power users keep installed
One-click scans. No signup required.
As Hanbury put it: “At this scale, the slowest constrained input sets the schedule for the entire program.” Power availability is one such input, but not the only one: equipment, specialist labor, approvals and stable design can each become the limiting factor.
Which construction skills are most exposed to shortages?
The interview identifies specialized electrical labor as the sharpest likely workforce constraint, particularly high-voltage, substation and mission-critical electrical skills. It also flags mechanical trades, pipefitters, controls specialists and commissioning roles, with demand pressure likely to grow as liquid cooling scales.
These needs are interconnected. Data center designs increasingly require coordination from chips and racks through networking and cooling to electrical architecture, the building and its power source. The interview calls this “chip-to-grid codesign”: choices made for compute hardware can affect cooling and electrical requirements, which can reshape the facility and its energy plan. A faster, more varied silicon roadmap makes early coordination across those disciplines more important.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How should contractors evaluate a data center pipeline?
Hanbury’s four questions offer a practical screen for whether a project is sufficiently real and mature to warrant committing resources:
Recommended Free Tools
- Is the power real? Confirm the power source, capacity and interconnection path, including the schedule for grid work or any onsite generation and storage.
- Are the customer and financing commitments real? Look for credible demand and a financing structure that can carry the project through delivery.
- Is permission to build real? Check approvals and community support, including concerns about electricity use, water, noise, emissions and other local effects.
- Is the design stable enough to build? Confirm that compute, cooling, electrical systems and construction plans are coordinated enough to support procurement and execution.
Hanbury summarized the screen this way: “I would ask four questions: Is the power real? Is the customer and financing commitment real? Is the permission to build real? And is the design stable enough to build?” Contractors can extend that screen with practical delivery checks: access to the required electrical and mechanical trades, supplier coordination, and a credible sequence for procurement and commissioning.
Best Value
What changes when data centers are delivered at industrial scale?
The interview’s proposed response is to treat the boom less as a string of isolated building projects and more as coordinated infrastructure programs. That changes how owners, contractors and suppliers plan work:
- Integrate site and power planning: Coordinate the campus with substations, transmission, interconnection and, where applicable, onsite energy and storage.
- Procure earlier: Long-lead systems and equipment need to be accounted for before they become schedule-critical.
- Use prefabrication and modular designs where they fit: Moving appropriate work into controlled settings can support repeatable delivery, although it does not remove the need for site power, approvals or skilled installation.
- Coordinate contractor and supplier portfolios: Compute, cooling, electrical work and construction must align as one delivery sequence rather than proceed as disconnected packages.
The commercial logic is straightforward: a construction pipeline only converts into operating capacity when its dependencies advance together. A project with a customer but no credible power path, or power but an unstable design, can still stall.
How certain is the $6 trillion projection?
The projection should be treated as a scenario-dependent hurdle, not a prediction that the market will necessarily reach that scale. The principal source is Construction Dive’s Oct. 8, 2026 interview with Hanbury, which attributes the figures to Bain. The specific estimates for cumulative data center spending, annual AI infrastructure spending, implied AI-market revenue and delayed-project value are not accompanied there by Bain’s underlying methodology. The interview is useful for understanding the scale of the claim and the constraints it identifies, but it does not independently validate the projected totals.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
