The Power-Hungry Future of AI: How Data Centers Are Reshaping Global Energy Demand

Introduction

The artificial intelligence revolution is consuming electricity at an unprecedented scale, fundamentally reshaping global energy markets and forcing utilities, tech giants, and governments to confront a stark reality: AI's computational demands are outpacing our ability to generate clean power. The numbers are staggering and growing exponentially.

Data centers in the US used somewhere around 200 terawatt-hours of electricity in 2024, roughly what it takes to power Thailand for a year. AI-specific servers in these data centers are estimated to have used between 53 and 76 terawatt-hours of electricity. On the high end, this is enough to power more than 7.2 million US homes for a year—and this is just the beginning.

By 2028, the researchers estimate, the power going to AI-specific purposes will rise to between 165 and 326 terawatt-hours per year. That's more than all electricity currently used by US data centers for all purposes; it's enough to power 22% of US households each year. This explosive growth represents one of the most significant energy challenges of the 21st century, with implications stretching from local utility planning to global climate commitments.

The Scale of AI's Energy Appetite

Current Consumption Patterns

The transformation of data center energy consumption has been swift and dramatic. In the United States, data center energy consumption in 2023 is estimated to be 176 terawatt-hours (TWh), or 4.4% of the country's total electricity consumption. Between 2014 and 2016, consumption remained relatively stable, hovering around 60 TWh. However, it began to increase steadily from approximately 2017 onwards, coinciding with the rise of cloud computing and early AI applications.

Data centers accounted for about 1.5 percent of global electricity consumption in 2024, an amount expected to double by 2030 because of AI use, according to the International Energy Agency. Our Base Case finds that global electricity consumption for data centres is projected to double to reach around 945 TWh by 2030 in the Base Case, representing just under 3% of total global electricity consumption in 2030.

The AI Acceleration Factor

What makes AI particularly energy-intensive is the nature of its computational requirements. A single ChatGPT query requires 2.9 watt-hours of electricity, compared with 0.3 watt-hours for a Google search, according to the International Energy Agency. This 10x difference in power consumption extends across all AI operations.

The energy demands escalate even further with more complex AI tasks:

• Training large language models: Requires processing vast datasets through extensive calculations across thousands of high-performance servers
• AI inference: Continuous computational work to respond to user queries and generate content
• AI video and image generation: Exponentially more energy-intensive than text-based AI

A significant portion of the energy consumption surge is attributed to AI-related servers: from 2 TWh in 2017 to 40 TWh in 2023. This represents a 20-fold increase in just six years, highlighting the exponential nature of AI's energy trajectory.

Geographic Concentration and Grid Impact

The concentration of data center development creates significant regional challenges. China and the United States are the most significant regions for data centre electricity consumption growth, accounting for nearly 80% of global growth to 2030. Consumption increases by around 240 TWh (up 130%) in the United States, compared to the 2024 level.

Data centers consumed roughly a quarter of Virginia's electricity in 2023, the highest in the US, followed by North Dakota at over 15%, and Iowa, Nebraska, and Oregon each exceeding 11%. This geographic clustering strains local power grids and forces utilities to make massive infrastructure investments.

The Infrastructure Challenge: How Many Data Centers Do We Need?

Current and Projected Capacity Requirements

Goldman Sachs Research estimates that there will be around 122 GW of data center capacity online by the end of 2030. To put this in perspective, that's equivalent to approximately 120 large nuclear power plants or about 600 natural gas peaking plants.

Our analysis suggests that demand for AI-ready data center capacity will rise at an average rate of 33 percent a year between 2023 and 2030 in a midrange scenario. This means that around 70 percent of total demand for data center capacity will be for data centers equipped to host advanced-AI workloads by 2030.

The scale of investment required is unprecedented. Our research shows that by 2030, data centers are projected to require $6.7 trillion worldwide to keep pace with the demand for compute power. This figure breaks down into:

• $5.2 trillion for AI-equipped data centers
• $1.5 trillion for traditional IT applications

The Supply Deficit Crisis

Despite massive planned investments, supply is struggling to keep pace with demand. Even if all currently known plans are delivered on time, there could still be a data center supply deficit of more than 15 GW in the United States alone by 2030.

Several factors contribute to this supply crunch:

Power Infrastructure Limitations: Many utilities find they haven't been able to build out transmission infrastructure quickly enough, and there is concern that at some stage they may be unable to generate sufficient power.

Construction Timeline Mismatches: While data centers can be built in 1-2 years, power generation projects, especially nuclear, can take 6-12 years from conception to operation.

Equipment and Labor Shortages: The specialized equipment required for AI data centers—particularly cooling systems and high-density power infrastructure—faces supply chain bottlenecks.

Regional Distribution and Market Leaders

The data center market is becoming increasingly concentrated among hyperscale operators:

Region Current Capacity (GW) 2030 Projected (GW) Growth Rate Northern Virginia 25 45-50 80-100% Silicon Valley 15 25-30 67-100% Dallas-Fort Worth 12 22-25 83-108% Chicago 10 18-20 80-100% New York Tri-State 8 15-18 88-125% Sources: Various industry reports and Goldman Sachs Research

The mix of this capacity is expected to skew even further towards hyperscalers and wholesale operators (70% versus 60% today). This consolidation reflects the enormous capital requirements and technical expertise needed to operate AI-ready facilities.

Power Generation Solutions: From Grid Strain to Nuclear Renaissance

The Grid Infrastructure Challenge

The rapid growth in data center power demand is straining electrical grids worldwide. Goldman Sachs Research estimates that about $720 billion of grid spending through 2030 may be needed. "These transmission projects can take several years to permit, and then several more to build, creating another potential bottleneck for data center growth if the regions are not proactive about this given the lead time."

The Electric Power Research Institute projects that data centers could consume up to 9.1% of US electricity by 2030, driven by AI demands. This represents a dramatic shift from the current 4.4%, requiring utilities to fundamentally rethink their capacity planning and investment strategies.

Natural Gas: The Bridge Solution

In the near term, natural gas is filling much of the power demand gap. Industry estimates of data center gas demand through 2030 range from 3 billion to 12 billion cubic feet per day amid a growing pipeline of planned natural gas plants. More than 99 GW of gas-fired capacity is planned across 38 states.

However, natural gas faces its own constraints. Many top data center markets have constrained gas pipeline takeaway and transportation capacity, creating potential bottlenecks for expanded gas-fired generation.

The Nuclear Renaissance: SMRs as the Long-Term Solution

Tech companies are increasingly turning to nuclear power as the ultimate solution for AI's insatiable power demands. The appeal of nuclear is clear: it provides 24/7 baseload power with zero carbon emissions, exactly what data centers need.

Major Corporate Commitments:

• Microsoft: Microsoft was inking an agreement with Constellation Energy to restart a shuttered nuclear reactor on Three Mile Island—the site of the worst nuclear disaster in U.S. history. The plan, announced in September, calls for the reactor to supply 835 MW to grid operator PJM.
• Amazon: AWS signed supply agreements with Energy Northwest and Dominion in Washington and Virginia. In addition, it directly invested in X-energy, an SMR developer, to support the construction of more than 5GW of new nuclear energy projects by 2039.
• Google: Google has turned to nuclear energy and signed a partnership with US-based Kairos Power, which is developing small modular reactors (SMRs). The first such reactor to power Google's data centres is expected to be operational by 2030, followed by additional reactor deployments through to 2035. Overall, this deal will enable up to 500 MW of new 24/7 carbon-free power to US electricity grids.

Small Modular Reactors: The Game Changer

Small Modular Reactors (SMRs) are emerging as the preferred nuclear solution for data centers. SMRs provide a stable 24/7 power source, ideal for the continuous demands of AI data centers. SMRs have capacities ranging from 20 to 300 megawatts, meeting energy needs without large infrastructure, as in the case of traditional nuclear plants.

Key Advantages of SMRs:

1. Scalability: Small modular nuclear reactors operate independently of centralized power grids, ensuring resilience for AI-critical functions.
2. Construction Speed: The modular nature of SMRs allows for quicker construction times (3-5 years) compared to traditional nuclear reactors (6-12 years).
3. Cost Efficiency: Most SMRs under development could cost less than $2 billion compared to more than $10 billion for traditional nuclear plants.
4. Enhanced Safety: Advanced safety features and smaller size reduce risk profiles compared to traditional large reactors.

The SMR Market Reality Check

Despite the enthusiasm, SMRs face significant challenges. NuScale's proposed flagship SMR project in Utah has closed because it became uneconomic, as the projected cost of construction increased by 75 percent (to $9.3 billion) between 2021 and 2024. The estimated cost of electricity per MWh increased from $58 to $89, even after including a $30/MWh subsidy under the Inflation Reduction Act.

SMRs are still at least five years from commercial operation in the United States. A year ago the first planned SMR in the United States was cancelled due to rising costs and a lack of customers.

Current development challenges include:

• Regulatory hurdles: Complex licensing and approval processes
• Technology maturity: SMRs are still in the early stages of development, currently at technology readiness levels (TRL) 5–6. The deployment of the first small modular reactors is expected by 2030 at the earliest.
• Supply chain constraints: Limited manufacturing capacity for specialized components
• Economic viability: Unproven commercial economics at scale

Economic and Environmental Implications

The Investment Surge

The scale of investment flowing into data center and energy infrastructure is unprecedented. Electric and gas utility capex is expected to surpass US$1 trillion cumulatively within the next five years (2025–2029) for the 47 biggest investor-owned utilities. For the hyperscalers, reaching the trillion-dollar threshold is expected in only three years, with spending projections reaching half a trillion dollars annually by the early 2030s.

Carbon Emissions and Climate Impact

The environmental implications are significant. Along the way, the carbon dioxide emissions of data centers may more than double between 2022 and 2030. Goldman Sachs Research estimates that the expected rise of data center carbon dioxide emissions will represent a "social cost" of $125-140 billion (at present value).

In 2024, fossil fuels including natural gas and coal made up just under 60% of electricity supply in the US. Nuclear accounted for about 20%, and a mix of renewables accounted for most of the remaining 20%. This means that much of AI's power demand is currently being met by fossil fuels, creating a tension between technological advancement and climate goals.

Renewable Energy Integration Challenges

While tech companies have made ambitious renewable energy commitments, the reality of AI's power demands creates challenges. Gaps in power supply, combined with the rush to build data centers to power AI, often mean shortsighted energy plans.

The intermittent nature of renewables conflicts with data centers' need for constant, reliable power. "to supplement our wind and solar energy projects, which depend on weather conditions to generate energy," as Amazon noted when explaining its nuclear investments.

Regional and Global Competitive Dynamics

The U.S.-China Competition

In China it increases by around 175 TWh (up 170%). In Europe it grows by more than 45 TWh (up 70%). Japan increases by around 15 TWh (up 80%). The geographic distribution of AI infrastructure development has significant geopolitical implications.

China's aggressive AI infrastructure development includes advanced nuclear technologies. Advanced nuclear reactors are now reaching deployment: the Shidaowan-1 power plant is the world's first operating fourth-generation nuclear reactor, with two 250 MW high-temperature helium gas-cooled reactors; the Linglong One is a 125 MW SMR, using PWR technology (operational by late 2025).

European Market Dynamics

Europe's power demand could grow by 40% and perhaps even 50%, according to Goldman Sachs Research. Goldman Sachs Research estimates a data center pipeline for Europe amounting to about 170 GW, equivalent to about one-third of the region's power consumption.

Future Scenarios and Strategic Implications

The DeepSeek Disruption

Recent developments have introduced uncertainty into power demand projections. At the end of January, Chinese artificial intelligence (AI) startup DeepSeek unveiled two large language models (LLMs)—DeepSeek-R1 and DeepSeek-R1-zero. Unlike previous generations of AI models, DeepSeek's breakthrough reduced the compute cost of AI inference by a factor of 10.

This development highlights the potential for efficiency improvements to moderate power demand growth, though the ultimate impact remains uncertain.

Policy and Regulatory Responses

Governments are beginning to respond to the infrastructure challenges. DOE has identified 16 potential sites uniquely positioned for rapid data center construction, including in-place energy infrastructure with the ability to fast-track permitting for new energy generation such as nuclear.

States like Ohio and Georgia have introduced new policies requiring data centers to cover infrastructure costs upfront. This represents a shift toward making data center operators bear the full cost of their infrastructure requirements.

Long-Term Outlook

The trajectory of AI power consumption will depend on several key factors:

1. Efficiency Improvements: Advances in chip design, cooling technology, and AI algorithms
2. Nuclear Deployment: Success in deploying SMRs and other advanced nuclear technologies
3. Grid Modernization: Investment in transmission and distribution infrastructure
4. Regulatory Framework: Policies governing data center development and power procurement
5. Breakthrough Technologies: Potential game-changers like fusion power or quantum computing

Conclusion: Navigating the Energy-AI Nexus

The intersection of artificial intelligence and energy represents one of the defining challenges of our technological age. AI's exponential growth in computational demands is forcing a fundamental rethink of how we generate, distribute, and consume electricity. The numbers are sobering: from 200 TWh of total data center consumption in 2024 to potentially over 900 TWh by 2030, with AI driving most of this growth.

The solutions being pursued—from massive grid investments to a nuclear renaissance focused on small modular reactors—reflect the scale and urgency of the challenge. Tech giants are making billion-dollar bets on nuclear power not out of environmental activism but out of practical necessity: AI requires reliable, 24/7 power that renewables alone cannot consistently provide.

The success or failure of these initiatives will have profound implications beyond the tech sector. If we cannot efficiently power AI's growth, we may face constraints on technological advancement. If we power it primarily with fossil fuels, we risk undermining climate goals. The path forward requires unprecedented coordination between technology companies, utilities, governments, and regulators.

The energy-AI nexus also raises broader questions about the sustainability of our technological trajectory. As AI becomes increasingly powerful and pervasive, its energy footprint grows exponentially. This creates an imperative not just for more power generation but for more efficient AI systems and applications that deliver maximum value per watt consumed.

The next five years will be critical in determining whether we can build the energy infrastructure needed to support AI's continued growth while maintaining climate commitments and grid reliability. The investments being made today—in nuclear power, grid modernization, and efficiency technologies—will shape the technological landscape for decades to come.

The AI revolution is reshaping not just how we work and communicate but how we think about energy itself. The decisions made in boardrooms and regulatory offices today about power generation and grid infrastructure will determine whether artificial intelligence becomes humanity's greatest tool for solving complex problems or a cautionary tale about the unintended consequences of unconstrained technological growth.

As we stand at this inflection point, one thing is clear: the future of AI is inextricably linked to our ability to generate clean, reliable power at unprecedented scales. The companies, countries, and regions that master this challenge will lead the AI revolution. Those that don't may find themselves left behind in the most transformative technological shift of our lifetime.

This analysis draws from the latest industry reports, government data, and corporate announcements. As AI technology and energy markets continue to evolve rapidly, these projections and scenarios should be considered dynamic and subject to significant revision as new technologies and policies emerge.