Infrastructure Outperformance Will Be Earned by Managing System Bottlenecks

Rapid growth in electricity demand has shifted US energy systems towards resilience-led operating models that prioritize security, reliability and affordability. This transition reflects the fragmentation of globalization priorities (e.g., US re-industrialization, sovereign supply-chain resilience) and a more pragmatic decarbonization pathway, accelerated by power-intensive economic activity led by artificial intelligence (AI) workloads and electrification.

These forces have broadened-out opportunities across energy-related infrastructure, but investment returns cannot depend on electricity demand growth as a secular tailwind alone. Outperformance will be earned by those operators able to overcome system-level constraints – from power availability, to grid interconnection, land access, permitting timelines, labor capacity, and access to finance.

Collectively, these execution bottlenecks determine whether capital converts into operational infrastructure assets – including energy, data centers, storage facilities – on economically viable timelines. Favorable policies can advance project build-out and shape execution mechanics, but fundamentally returns are driven by value levers: demand, location, pricing, speed-to-market, sector-specific constraints/tailwinds, and cost of capital.

Energy models pivot to delivery speed

US electricity demand is projected to grow by nearly 2% annually through 2030, adding more than 420 TWh, according to Energy Information Administration (EIA) data. Data centers account for around half of this growth, with total data center electricity usage forecast to surge more than three-fold – from 176 TWh in 2023 to up to 580 TWh in 2028. Additional demand is attributed to electric vehicles, residential and commercial building air conditioning and heating, advanced manufacturing reshoring (i.e., semiconductor and battery manufacturing plants), and critical minerals processing.

This accelerated electricity dependency has repositioned U.S. energy policy around expanded generation across all energy types – renewables, natural gas, nuclear, thermal and coal. This “all-of-the-above” model means grid reliability, transmission, capacity and affordability are all investable variables.

The rapid scalability of renewables – capable of clearing permitting and construction in around 12-18 months – makes them the primary solution for immediate grid bottlenecks, in addition to their function as a decarbonization play. Intermittent energy outputs further create adjacent demand for investment in storage capacity and infrastructure.

Natural gas delivers rapid near-term capacity additions, meeting around 40% of incremental power demand through 2030, according to EIA, as a bridge to Small Modular Reactors (SMRs) and scaled renewable-plus-storage. The US nearly tripled its gas-fired capacity in development in 2025, totaling almost 252 gigawatts (GW), according to Global Energy Monitor data. But natural gas is more expensive and turbine supply constraints are a limiting factor.

Meanwhile, nuclear power – long plagued by scale, capital intensity, long construction lead times, complexity, and risk liability – has re-entered discussions with federal backstops, supporting stable baseloads for energy-intensive industries. For example, Meta’s recently announced January 2026 nuclear projects with three partners target 6.6 GW of new and existing clean energy by 2035. These deals are de-risked by tax credits and cost-sharing incentives. However, licensing delays, cost overruns, workforce and supply constraints, and pre-construction funding gaps persist.

The rapid scalability of renewables – capable of clearing permitting and construction in around 12-18 months – makes them the primary solution for immediate grid bottlenecks, in addition to their function as a decarbonization play. Intermittent energy outputs further create adjacent demand for investment in storage capacity and infrastructure.

Data centers: where bottlenecks become investable

Data centers are the most visible stress-test of this new US energy regime. The asset class concentrates demand, exposes grid and permitting bottlenecks, geographic cost dispersion and social impact risks, while also attracting capital flows into both power assets and digital infrastructure. McKinsey estimates approximately $5.2 trillion will be required for data centers, power generation and grid infrastructure by 2030, plus another $1.5 trillion for traditional IT—around $7 trillion in total global infrastructure investment. The five largest hyperscalers (i.e., Meta, Google, Microsoft, Amazon and Oracle) are guiding predominantly AI-related capex investment up to $680 billion in 2026, taking the four-year spend to $1.5 trillion. The bulk of hyperscaler capex will be funneled into compute hardware, data centers, related power assets and other digital infrastructure, but the binding constraints are deployment: power, interconnection and delivery capacity. Increasingly, flexibility itself is priced as a service. Data center operators and hyperscaler occupiers are negotiating grid access partly on their ability to adjust demand at peak times, co-locate storage, or accept staged power delivery. This creates new revenue opportunities linked to reliability, timing and system stability.

Resolving these mismatches between demand and delivery constraints is where investment outperformance is earned.

Resolving these mismatches between demand and delivery constraints is where investment outperformance is earned. The US grid itself is a case in point. The grid was built for one-way power flows from generation to consumption, and unsurprisingly ill-suited for sustained, high-intensity load growth from data centers and bidirectional flows created by distributed generation and storage. In practice, what matters is delivering power to specific sites and assets within a workable timetable. In hotspots like Northern Virginia, Phoenix and Dallas-Fort Worth, connection queues can stretch up to seven years. Where hyperscalers are competing for constrained grid capacity, capital alone cannot clear these queues. In response, some are trading flexibility for time-to-power and security – agreeing to reduce usage at peak times, or pairing sites with on‑site generation and storage– to secure earlier, but power-limited, connection dates. This shows that large power users are becoming active grid participants, rather than buying power at the prevailing market price.

There is also regional cost dispersion and social impact to consider. Hyperscalers are already shifting to remote markets in response to community pushback over new developments in proximity in residential neighborhoods, in a reminder that demand must accommodate community risk alongside physical constraints.

Bottlenecks as outperformance catalysts

Surging electricity demand has driven a pragmatic shift in the US energy transition in support of industrial competitiveness, energy reliability, capacity growth and speed to delivery. For developers across energy, data centers and industrial infrastructure, the binding variable remains “time-to-power”.

The Trump Administration has recategorized AI data centers and related power assets as “critical infrastructure” priorities, paving the way for permitting reform and land access to be operationalized as industrial policy levers. On 23 July 2025, an executive order directed agencies to expedite qualifying data center and supporting energy projects through federal permitting pathways, including authorizing development on certain federal lands.

With US electricity demand outpacing new supply additions through 2030, bottlenecks will shift from generation to grid integration and transmission. But these bottlenecks also create opportunities for operators able to create value through strategies that overcome them. Developer-operators that clear transmission and grid integration hurdles improve asset underwriting clarity and resolve system constraints, enabling economic expansion rather than acting as passive beneficiaries of demand.

Constraint-resolution translates to several investable pathways, including:

Time-to-power de-risking
Securing grid connections and shortening the path for assets to become operational – interconnection approvals, utility-side substations, high-voltage switchgear, and transmission/distribution upgrades. This includes powered land, where permitting, interconnection rights and utility commitments secure electricity before construction begins—and which carries inherent execution and queue risk.
Power source flexibility
Distributed generation, storage and microgrids that reduce reliance on constrained interconnection pathways and improve resilience during outages and periods of peak pricing.
Delivery-bottleneck monetization
Scarce equipment and specialist capacity that protects schedules – transformers and switchgear, packaged substations, specialist electrical contractors, electrician labor, and prefabricated/modular components.

At scale, these could materially influence the pace of US industrial and digital growth. At scale, these could materially influence the pace of US industrial and digital growth. Ultimately, what matters most is risk-adjusted returns with limited technology risk, with the appropriate solution mix dependent on regional value drivers as well as the durability and security of contracted revenues. Operators who compress multi-year timelines into operational assets capture the premium constrained markets are willing to pay. Infrastructure returns in this AI-accelerated cycle belong to those who overcome system constraints and monetize bottlenecks, rather than those who ride on the back of demand growth tailwinds and wait in line.