Gridlocked intelligence: Why AI’s construction boom is really an infrastructure risk story

Artificial intelligence (AI) is driving one of the fastest investment cycles in decades. Yet the biggest obstacles to delivery are grounded in physical systems: power, water, grid access, permitting and long-lead equipment that look viable on paper are encountering friction well before a construction project begins.

Across markets, these enabling systems are struggling to keep pace.

This is the central message of “Gridlocked intelligence” in Zurich’s Beyond 2030: The Future of Construction report. Its core argument is that feasibility itself is becoming the key risk factor. Projects with capital, approvals and strong strategic backing can still stall when grid connections slip, cooling capacity falls short, water access tightens or critical components fail to arrive on time.

Delivery risk is moving earlier in the lifecycle and becoming harder to see.

Pressure spreads beyond data centers

Data centers sit at the leading edge of this shift. Their high power density, constant cooling needs and clustering effects make them especially exposed to infrastructure constraints.

However, the pattern is broader. Advanced manufacturing, logistics platforms and healthcare facilities are also becoming more energy-intensive, location-sensitive and time-critical. These sectors are encountering similar pressures around grid capacity, permitting timelines and equipment availability.

As a result, project outcomes are increasingly shaped before ground is broken.

A highly connected risk

The report’s wider analysis highlights a change in the structure of risk. Extreme weather and natural disasters are perceived to dominate in terms of severity over the next five years, followed by financial and labor pressures.

By contrast, disruption to critical infrastructure is seen as ranking lower on severity, yet playing a pivotal role in the system. It sits near the center of the network, shaped by other risks and capable of amplifying their effects across projects and portfolios.

This has important implications. Infrastructure disruption rarely stays contained within a single asset. A delayed substation upgrade, for example, can stall multiple projects across a region, compounding financial exposure and scheduling risk simultaneously. Risks that appear manageable in isolation can cascade through shared systems, supply chains and portfolios, turning local constraints into broader delivery challenges.

Climate stress tightens constraints

At the same time, climate pressure is reducing the resilience of the systems projects depend on.

Power and water infrastructure are operating under conditions they were not designed to handle. Heat, drought and flooding are narrowing operating margins for substations, transmission networks and cooling systems.

Zurich Resilience Solutions analysis highlights the scale of the challenge. In ASEAN, around three‑quarters of future renewable generation capacity is expected to face critical climate risk by 2030, with hydropower particularly exposed and vulnerability intensifying further towards mid‑century.

Water availability is also becoming a critical factor. A large data center can consume up to 19 million liters per day for cooling, roughly equivalent to the needs of a town of up to 50,000 people. At the same time, around one-quarter of existing facilities and nearly one-third under construction are located in regions expected to face greater water scarcity by 2050.

For risk managers, this places water alongside power as a core delivery, permitting and reputational consideration.

The grid as a gating factor

Grid access is emerging as the most immediate constraint.

In the U.S., interconnection queues now exceed total installed generation capacity, with median connection timelines extending beyond four years. Shortages of transformers and switchgear add further pressure, with lead times stretching into years and costs rising significantly.

These delays directly affect project timelines and viability. They shape siting decisions, influence financing risk and determine whether developments proceed as planned.

Procurement pulls risk forward

As a result, procurement strategies are evolving.

Developers are securing equipment earlier through advance commitments, prepayments and co-investment. This can improve schedule certainty but introduces new exposures. Working capital is tied up for longer periods. Storage, handling and degradation risks increase. Accountability shifts earlier in the project lifecycle.

Risk does not disappear. It changes form and location.

In some markets, developers are responding by building closer to power sources, investing in on-site generation and adopting more efficient cooling technologies.

Local bottlenecks define outcomes

Demand for AI infrastructure continues to grow rapidly, but the most important constraints are local.

Even with strong global capacity, pressure concentrates in specific regions. Around 85 percent of new data center capacity is expected in the U.S., China and the EU. In the U.S., these facilities could account for nearly half of electricity-demand growth this decade.

For risk managers, this reinforces a key point: Aggregate figures can obscure the real challenge. Delivery risk is driven by regional bottlenecks – grid congestion, water stress and permitting timelines – rather than global supply.

“Data center demand is no longer constrained by ambition or capital; it is increasingly constrained by infrastructure readiness at the local level,” says Patrick McBride, Zurich’s Head of International Construction. “In this environment, the projects that consistently outperform are those that secure power, water, permitting and critical equipment earliest, and treat these dependencies as core elements of feasibility, rather than risks to be resolved downstream.”

What this means in practice

Three priorities stand out.

1. Treat enabling systems as core dependencies
   Grid access, water availability, permitting and equipment supply need to be secured early and monitored continuously. Where delivery depends on external infrastructure, such as utility upgrades, project timelines must reflect realistic readiness.
2. Assess feasibility earlier
   Evaluating grid capacity, water-basin stress, climate exposure and supplier dependencies before major capital commitments can reduce late-stage disruption and support more robust decision-making.
3. Manage shared dependencies at portfolio level
   Projects often rely on the same substations, interconnection points or supplier networks. Failures in a shared component can affect multiple assets simultaneously, creating accumulation risk for both developers and insurers.

Constraint as a differentiator

These pressures also create a clear dividing line between projects.

Performance increasingly depends on how early constraints are identified and how effectively they are managed. Siting, redundancy planning, supplier strategy and utility engagement all become more critical.

Early involvement from insurers can also add value, providing insight into accumulation risk and system fragility before issues surface in financing or delivery performance.

The question every project must answer

AI is accelerating demand across the built environment. But delivery still depends on fundamentals:

• Where will the power come from?
• Where will the water come from?
• How long will the connection take?
• What happens when shared systems are under strain?
• These questions now shape feasibility as much as financing or design.

The projects that succeed will be those with secure access to the systems behind them. In the next phase of construction, competitive advantage will be defined by infrastructure.

Originally published in Commercial Risk on June 23, 2026.