Artificial intelligence is rapidly transforming from a software breakthrough into one of the most powerful infrastructure forces in the modern American economy. What began as a race to build smarter algorithms has evolved into a nationwide expansion of physical infrastructure involving massive data centers, power generation projects, transmission networks, semiconductor facilities, cooling systems, and utility upgrades.
Across the United States, technology companies are investing hundreds of billions of dollars into AI infrastructure. Microsoft, Amazon, Google, Meta, OpenAI partners, Oracle, and other hyperscalers are building increasingly larger data center campuses to support AI model training and inference workloads. The scale of electricity required for these operations is so large that utilities, grid operators, regulators, and policymakers are now revising long-term forecasts that had remained relatively stable for decades.
Several major research institutions now estimate that AI-driven data centers could account for roughly 30–40% of all new US electricity demand growth through 2030. Some forecasts suggest the number could become even higher under accelerated AI adoption scenarios.
The United States power sector is therefore entering a historic transition where electricity availability may become one of the defining constraints of the digital economy.
America’s Electricity Demand Is Rising Again
For much of the past two decades, US electricity demand growth remained relatively flat despite economic expansion. Energy efficiency improvements in appliances, industrial systems, and buildings helped offset rising consumption from population growth and digital technologies.
AI is changing that equation.
The rapid deployment of hyperscale data centers is now creating one of the strongest electricity demand surges seen in decades. Utilities that once planned for slow, predictable growth are suddenly preparing for unprecedented load increases concentrated in specific regional markets.
One of the most important baseline studies comes from the Lawrence Berkeley National Laboratory (LBNL). In the “2024 United States Data Center Energy Usage Report,” authored by Arman Shehabi and other researchers, LBNL estimated that US data centers consumed approximately 176 terawatt-hours (TWh) of electricity in 2023. That represented roughly 4.4% of total US electricity consumption.
The report projected that by 2028, data center electricity demand could rise to between 325 and 580 TWh annually, equivalent to around 6.7% to 12% of all US electricity use. These figures have become some of the most widely cited government-backed estimates in the AI infrastructure debate.
The implications are enormous. At the upper end of the forecast range, US data centers would consume more electricity than many developed countries.
The IEA Says Data Centers Already Drive Half of Demand Growth
The International Energy Agency (IEA) has also identified AI infrastructure as a major contributor to global electricity growth.
In its April 2025 “Energy and AI” report, followed by updates in the 2026 “Global Energy Review,” the IEA stated that data centers accounted for roughly 50% of US electricity demand growth in 2025. AI-focused data centers represented the fastest-growing segment.
The agency further projected that US data center electricity consumption could increase approximately 130% by 2030 compared with 2024 levels, adding nearly 240 TWh of additional demand over the decade.
These findings are especially important because the IEA traditionally focused on industrial production, transportation, and residential energy trends. The growing emphasis on AI infrastructure signals that digital computing is becoming a central pillar of global energy planning.
The IEA also emphasized that the United States represents one of the largest contributors to global AI-related electricity demand growth due to its dominant role in hyperscale cloud infrastructure and advanced AI development.
Goldman Sachs and Wall Street See a Power Boom Ahead
Wall Street analysts increasingly view electricity infrastructure as one of the biggest secondary beneficiaries of the AI revolution.
Goldman Sachs Research, in its report “Generational Growth: AI, Data Centers and the Coming US Power Demand Surge,” projected that AI-driven data centers could account for approximately 40% of total US electricity demand growth in coming years.
The investment bank also estimated that data center expansion may contribute around 1.2 to 1.5 percentage points to annual US power demand growth through 2030.
Globally, Goldman Sachs forecasts that data center electricity demand could rise between 165% and 220% by 2030 relative to 2023 levels. The United States is expected to remain the primary engine behind this expansion because of its concentration of AI firms, semiconductor infrastructure, and cloud computing giants.
Financial markets are increasingly recognizing that AI is not simply a software investment cycle. It is an infrastructure cycle involving utilities, power producers, engineering firms, construction companies, cooling specialists, and transmission developers.
In many ways, electricity is becoming the hidden foundation of the AI economy.
Why AI Workloads Consume So Much Electricity
Traditional cloud computing already required substantial electricity, but generative AI systems operate at an entirely different scale.
Training advanced large language models involves thousands or even tens of thousands of GPUs operating simultaneously for extended periods. These systems perform massive parallel computations that consume enormous quantities of electricity.
Inference workloads — the process of generating responses for users in real time — are also growing rapidly as AI becomes integrated into search engines, office software, coding tools, healthcare platforms, and enterprise systems.
Unlike older computing workloads, AI applications often require higher-density server racks with far greater energy consumption per square foot.
Cooling has become another major challenge. Advanced AI chips generate immense heat, forcing operators to deploy sophisticated liquid cooling and thermal management systems. Cooling infrastructure alone can account for a significant portion of total data center electricity use.
As AI adoption expands globally, these workloads are expected to operate continuously around the clock, increasing baseline electricity demand.
EPRI Warns Data Centers Could Consume 17% of US Electricity
Among the most striking projections comes from the Electric Power Research Institute (EPRI).
In its February 2026 technical report, “Powering Intelligence 2026: Updated Scenarios of U.S. Data Center Electricity Use and Power Strategies,” EPRI estimated that data centers could consume between 9% and 17% of all US electricity by 2030.
That represents a dramatic increase from today’s approximate 4–5% share.
EPRI also noted that its updated 2026 scenarios were roughly 60% higher than earlier 2024 projections due to the accelerating pace of AI adoption and hyperscale infrastructure construction.
These revised estimates illustrate how quickly expectations are changing within the energy industry itself. Even conservative utility planners are now acknowledging that AI infrastructure could reshape long-term electricity demand patterns across the country.
Utilities Are Experiencing a “Generational Load-Growth Phenomenon”
Utility executives are increasingly describing the AI boom in historic terms.
Bill Fehrman, Chairman, President, and CEO of American Electric Power (AEP), stated during the company’s Q1 2026 earnings call:
“We are in the midst of a generational load-growth phenomenon... AEP is executing on our strategic plan at an exceptionally high level during a time of unprecedented opportunity for our industry.”
AEP subsequently increased its capital investment plan to approximately $78 billion. The company also disclosed that roughly 90% of its expected incremental contracted load growth through 2030 — approximately 63 gigawatts — is associated with data centers and hyperscale customers.
That scale is extraordinary. Sixty-three gigawatts is comparable to the electricity demand of multiple US states combined.
Dominion Energy, one of the utilities most exposed to data center growth because of its Northern Virginia footprint, has reported similar trends.
Robert Blue, Chairman, President, and CEO of Dominion Energy, stated in late 2025 and early 2026 discussions:
“We continue to see robust demand from data centers... We’re seeing continued appetite for additional data center capacity in our service territory. Developers want to go fast… I don’t see any reason why that’s going to change.”
Dominion reported approximately 47–48.5 GW of data center capacity moving through various contracting stages by late 2025 and early 2026.
Northern Virginia already hosts the world’s largest concentration of data centers, and AI expansion is intensifying pressure on the region’s electricity infrastructure.
Texas and Regional Grid Operators Face Mounting Pressure
The Electric Reliability Council of Texas (ERCOT) has become another focal point in the AI electricity debate.
In its April 15, 2026 preliminary long-term load forecast for 2026–2032, ERCOT projected extremely large demand growth scenarios tied heavily to data centers and other large industrial loads.
Some scenarios suggested total electricity demand could approach approximately 278 GW by 2029. ERCOT also projected summer 2026 peak demand between 90.5 GW and 98 GW.
Texas is attractive for AI infrastructure because of its competitive electricity market, abundant land availability, relatively business-friendly regulations, and large renewable energy base. However, the scale of projected load growth is creating concerns about grid reliability and reserve margins.
Utilities and grid operators across the country are now confronting a difficult challenge: AI infrastructure is expanding faster than transmission and generation systems can be upgraded.
Transmission Bottlenecks Could Slow Expansion
One of the biggest constraints on AI growth may ultimately be the electric grid itself.
Building transmission infrastructure in the United States is notoriously slow. New high-voltage transmission projects often require years of environmental review, regulatory approval, permitting negotiations, and local stakeholder engagement.
At the same time, hyperscale data center developers want rapid deployment timelines.
This mismatch between digital infrastructure speed and physical infrastructure timelines is becoming a major industry concern.
A 2025 Deloitte survey involving approximately 120 power company and data center executives identified grid stress and infrastructure timing mismatches as some of the most important challenges facing AI expansion.
In some regions, utilities are already delaying interconnection approvals because existing infrastructure lacks sufficient capacity to support proposed data center campuses.
The situation is especially difficult because AI facilities demand extremely high reliability. Even brief power interruptions can disrupt operations and create enormous financial consequences.
The Search for Reliable Power Sources
The AI boom is also reshaping America’s energy mix discussions.
Technology companies continue investing heavily in renewable energy procurement through long-term power purchase agreements for solar and wind projects. Many hyperscalers maintain ambitious carbon reduction goals and seek clean energy to offset rising electricity consumption.
However, renewable energy intermittency creates challenges for continuous AI workloads that operate 24 hours a day.
As a result, utilities and technology companies are increasingly exploring complementary power sources including battery storage, natural gas, geothermal energy, hydrogen, and nuclear energy.
Small modular reactors (SMRs) have gained renewed attention as potential long-term solutions for providing reliable carbon-free power to hyperscale data centers.
Natural gas generation is also expected to remain critical during the transition period because of its dispatchable reliability.
Some analysts now believe the AI boom could delay the retirement of certain fossil fuel plants because utilities require stable generation capacity while renewable and transmission infrastructure continue expanding.
Water, Sustainability, and Community Concerns
Electricity is not the only infrastructure challenge tied to AI growth.
Water consumption has emerged as another major issue. Advanced cooling systems often require significant water resources, particularly in warmer climates.
This creates tension in drought-prone states already facing water scarcity concerns.
Communities near large data center developments are also raising questions about land use, noise pollution, environmental impacts, and rising electricity costs.
While data centers generate construction jobs and tax revenue, critics argue that the facilities sometimes create relatively limited permanent employment compared with their infrastructure footprint.
Policymakers increasingly face the challenge of balancing economic competitiveness with environmental sustainability and local community priorities.
➡️ Read the related Post: AI Data Centers and the Global Electricity Surge: Why Power Is Becoming the New Bottleneck of Digital Infrastructure
Efficiency Gains Could Moderate Some Growth
Despite aggressive demand forecasts, uncertainty remains about the ultimate scale of AI-related electricity consumption.
Technology companies are investing heavily in more efficient semiconductors, advanced cooling systems, and optimized computing architectures.
Future AI chips may deliver significantly higher computational performance per watt than current systems. Operators are also improving server utilization and workload optimization techniques.
Some experts believe these efficiency gains could partially offset the rapid growth in AI demand.
However, historical experience suggests that improved efficiency often leads to broader adoption rather than lower total consumption. As AI systems become cheaper and more accessible, businesses and consumers may simply use them more extensively.
This “rebound effect” has appeared repeatedly throughout previous industrial and technological revolutions.
Even under more moderate scenarios, most major institutions still project substantial increases in data center electricity demand over the remainder of the decade.
AI Is Becoming a National Infrastructure Strategy
The rapid expansion of AI infrastructure is increasingly viewed as a matter of economic and geopolitical strategy.
The United States considers leadership in artificial intelligence essential for future competitiveness, national security, military applications, and technological influence.
As a result, electricity infrastructure is becoming deeply intertwined with industrial policy.
Federal and state governments are accelerating discussions around transmission reform, energy permitting, domestic semiconductor manufacturing, and grid modernization.
The AI race is therefore no longer limited to software innovation alone. It increasingly depends on whether countries can build enough physical infrastructure — especially electricity infrastructure — to support advanced computing systems.
Access to abundant, reliable, and affordable power may become one of the defining competitive advantages of the next decade.
The Emerging Reality of the AI Economy
The growth of AI infrastructure is reshaping the relationship between technology and energy in ways few anticipated only a few years ago.
Data centers are no longer niche facilities supporting internet traffic. They are becoming some of the largest industrial electricity consumers in the modern economy.
Research from LBNL, EPRI, the IEA, Goldman Sachs, ERCOT, and major utility operators all point toward the same conclusion: AI is creating a historic surge in electricity demand that could redefine US energy planning.
Whether AI infrastructure ultimately accounts for 30%, 40%, or an even larger share of new electricity demand growth, the broader direction is increasingly clear.
The digital economy now depends on physical power infrastructure at an unprecedented scale.
Utilities are expanding investment plans. Grid operators are revising long-term forecasts. Policymakers are debating energy permitting reforms. Technology companies are pursuing dedicated power strategies. Investors are increasingly treating electricity infrastructure as a core AI opportunity.
The next phase of the AI revolution may therefore be determined not only by algorithms and chips, but by transformers, substations, transmission lines, and power plants.
In the years ahead, electricity could become the defining bottleneck — and one of the greatest investment opportunities — of the artificial intelligence era.
Explore Core Insights Review
Core Insights Review contributors publish research-based analysis and editorial insights on commercial real estate, PropTech, smart infrastructure, sustainable construction, industrial real estate, and emerging technologies shaping the future of the built environment.
Post a Comment
0Comments