Skip to main content
Marsh logo

Article

The power equation: What the digital infrastructure boom means for utilities and IPPs

Power has become a key constraint on digital infrastructure growth. The explosive build-out of artificial intelligence infrastructure has turned reliable electricity supply into the single most consequential factor determining how the next generation of data centres will be built.

13/07/2026

Power has become a key constraint on digital infrastructure growth. The explosive build-out of artificial intelligence infrastructure has turned reliable electricity supply into the single most consequential factor determining how the next generation of data centres will be built. And the pace at which build-outs are taking place has placed utilities and independent power producers at the centre of one of the largest capital deployment cycles the energy sector has ever seen.

The opportunity is structural, not cyclical. Data centre electricity consumption in the United States stood at roughly 192 terawatt-hours (TWh) in 2024, or about 4.7% of total national demand. By 2030, projections from the Lawrence Berkeley National Laboratory suggest that figure could reach anywhere from 521 to 843 TWh, as much as 15.3% of the country’s total electricity use. Those are not incremental numbers. They represent a fundamental realignment of who the grid serves and how it needs to be built.

But the risk profile embedded in this opportunity is unlike anything the sector has previously navigated. The digital infrastructure boom will create value across the energy sector. But that value will not accrue evenly. The energy providers best positioned to succeed will likely be those that can connect commercial ambition with risk discipline — and build the ecosystem required to deliver power reliably and profitably.

Across Marsh, we see a market where the upside potential is real, but the path to capturing it is lined with financial, operational, regulatory, and reputational risks. These demand a level of integrated analysis the industry is still building. To help our clients, we have brought together a team of specialists across all asset classes and insurance needs to look at this sector in an integrated fashion — across individual project life cycles and investor balance sheets, and through the lens of the capital markets that will fund and ultimately bear much of this risk.

A different kind of customer

The demand driving this cycle is not coming from broad-based economic growth. It is concentrated among a relatively small number of hyperscalers, such as Microsoft, Amazon, Google, Meta, that are racing to build AI-capable infrastructure at a scale that would have seemed implausible five years ago. Individual hyperscale campuses now routinely require 500 MW to 1 GW or more of dedicated power. They expect “five nines” reliability (99.999% uptime), operate at 90% to 100% load profiles around the clock, and demand energisation timelines of 18 to 36 months. These are not typical utility customers. The combination of scale, speed, and reliability expectations creates a demand profile that is fundamentally different from anything the sector has historically served. 

The dual mandate

Utilities and IPPs are not simply being asked to serve more load. They are being asked to do two things simultaneously: build new power generation and upgrade an aging grid.

On the generation side, that means adding dispatchable, firm capacity (such as gas, long-duration storage, fuel cells, and, in some cases, nuclear and small modular reactors) capable of sustaining the 24/7 load profiles hyperscalers require. It also means structuring complex long-term contracts, build-own-operate arrangements, and power purchase agreement (PPA) terms are part of serving customers that demand near-absolute uptime, which agreements collectively carry significant credit and financing implications.

On the grid side, the challenge is equally pressing. Approximately 70% of US transmission lines are over 25 years old. Power transformer lead times now stretch to two to three years, and the US faces an estimated 30% shortfall in transformer availability. The country needs roughly 5,000 miles of new high-capacity transmission per year through 2050, but only 888 miles were built in 2024. The nation’s largest utilities — Duke Energy, Exelon, AEP, and Edison — are projecting more than $30 billion in combined annual transmission and distribution CapEx through the end of the decade. Whether those commitments can be delivered on the projected timelines is an open question.

Both elements must be executed in parallel, under constrained supply chains, and against timelines that don’t align with how the sector has historically planned and built infrastructure.

The risk landscape

Will contracted demand produce durable, financeable cash flows without creating balance sheet strain?

  • Off-taker and demand risk: Load may not materialise at the scale, timing, or duration anticipated. Single-counterparty concentration, volume shortfalls driven by AI investment cycles, or default can erode project economics and trigger covenant breaches.
  • Credit-metric, rating, and liquidity risk: Rapid, large capital deployments financed with debt can compress funds from operations-to-debt ratios, reduce headroom above investment-grade thresholds, raise financing costs, and strain liquidity. 
  • Residual value and stranded asset risk: Long-life generation assets (30 to 40 years) backed by shorter PPAs (10 to 15 years) create post-contract exposure. If the load does not renew or fully ramp, significant capital may be unrecoverable. Previous utility investment cycles have shown that large capital commitments do not automatically translate into shareholder value.

Can the system be built, interconnected, and operated to deliver the timing, capacity, and quality required?

  • Interconnection and delivery risk: Multi-year regional transmission organisation and independent system operator (RTO/ISO) queue delays, Electric Reliability Council of Texas (ERCOT) on-ramp rules, transformer shortages, and permitting bottlenecks can delay energisation and defer revenue well beyond projected timelines. Between 30% and 50% of US data centres planned for 2026 are already expected to face power constraints and grid equipment shortages.
  • Reliability and performance risk: Hyperscalers demand near-absolute uptime and tight dispatch. Underperformance can often carry severe consequences: penalty provisions, curtailment orders, and contract disputes that can rapidly diminish value.

Can utilities and IPPs maintain regulatory and stakeholder support when expectations and costs diverge?

  • Regulatory change risk: Commission rulings on cost recovery, prudence, tariff design, and jurisdictional treatment operate on six- to 12-month cycles, not aligned with 15- to 30-year capital commitments. The divergence between outcomes in Louisiana and Virginia illustrates how quickly project economics can shift.
  • Reputational risk: Perceptions that ratepayers and communities are subsidising hyperscaler growth can trigger rapid backlash, litigation, and erosion of stakeholder legitimacy. An estimated $60 billion in projects across 24 states have been delayed or cancelled due to public resistance.
  • States are moving to restrict or condition data centre development: Texas's SB 6 passed in 2025 imposes interconnection costs and emergency curtailment requirements on loads above 75 MW. Florida's SB 484, which is now in effect, requires large data centres to bear their full cost of service and preserves local authority to reject projects, and Maine's legislature passed the nation's first statewide data centre moratorium before the bill was vetoed. At the federal level, the proposed Artificial Intelligence Data Center Moratorium Act (S. 4214) would halt new AI data centre construction until national safeguards are established.

Behind-the-meter power and the competitive threat

Adding to the complexity, hyperscalers are increasingly pursuing behind-the-meter (BTM) power solutions as an alternative to relying on grid-delivered supply. Faced with multi-year interconnection queues, developers are turning to on-site generation gas turbines, hybrid renewable-plus-storage stacks, microgrids, and longer-term solutions, like small modular reactors. By 2030, 38% of data centre facilities are expected to use on-site generation for primary power, and 27% expect to be fully powered by on-site generation.

For utilities, this is not a niche development. Customers who bypass the grid reduce volumetric throughput, increase the risk of stranded T&D investment, and can erode the regulated rate base that underpins credit quality. Managing the boundary between enabling BTM solutions and protecting the utility’s own financial position is one of the defining strategic tensions of this cycle.

A business model under pressure

Taken together, these dynamics are forcing a re-thinking of what it means to be a utility in the current economy. The traditional model, passive demand forecasting, reactive infrastructure buildout, and a captive customer base, is giving way to an increasingly risk-exposed business. Utilities that succeed in this environment will be those that proactively shape demand, build strategic ecosystems that attract hyperscaler commitment, and differentiate on speed, capability, and coordination. IPPs face a related but distinct set of pressures. While their business models are largely unchanged by digital infrastructure, their exposure to offtaker concentration, PPA structure risk, and capital markets has increased substantially.

The Marsh perspective

This is an environment that demands more than capital and ambition. It requires integrated expertise across risk quantification, transaction structuring, regulatory strategy, workforce transformation, and balance-sheet management, applied simultaneously, not sequentially.

Marsh’s global risk advisory and insurance capabilities address the physical and financial exposures that accompany large-scale infrastructure development. Marsh Risk brings dedicated capabilities in quantifying and transferring the complex liability profiles that emerge when long-duration capital commitments meet fast-moving market conditions. Oliver Wyman assists with the strategic and analytical foundation — on business model transformation, demand uncertainty, and counterparty risk, assessing whether an opportunity is genuinely value-accretive or simply capital-intensive. Guy Carpenter brings reinsurance market access and risk transfer structuring for portfolios growing rapidly in both scale and complexity. And Mercer supports the workforce and organisational transformation that utilities must undertake to compete for the next generation of energy demand.

Digital infrastructure opportunities will define the trajectory of the energy sector for the next decade. But the value it creates for shareholders, ratepayers, and the broader economy will depend entirely on how well energy providers understand the risks they are taking on, and how effectively they manage them.

Interested in learning more? Fill in the form below to speak with a Marsh representative.