AI Revolution Sparked Growth in Data Centers

In 2024, the backbone of America’s digital transformation rested on thousands of data centers providing the computational power for AI, cloud services, and high-speed internet connectivity.

The number of new data center projects announced in the first half of the year have tripled, driven by the explosive demand for AI processing and cloud services. Estimates based upon a study from Lawrence Berkeley National Laboratory suggest that U.S. data centers consumed more than 70 billion kilowatt-hours (kWh), or roughly 2% of total U.S. electricity consumption.

Yet, as their energy demands soared this year, data centers faced ignited debates over energy sustainability and their impact on local communities.

Major tech companies began exploring renewable and alternative energy sources to power their operations sustainably. While solar and wind energy remain popular, nuclear energy has emerged as a serious contender for meeting the immense energy requirements driven by AI processing.

In September, Energy Department Secretary Jennifer Granholm urged artificial intelligence companies to provide clean energy for their new data center projects, warning that spiked demand on the electricity grid risks backlash from local communities. 

“If that extra power cost is put on everyday citizens, then there will be no desire to see these data centers permitted and placed in the United States,” Granholm warned at the AI+ Energy Summit in Washington. With an eye toward the future Granholm said, “Many of the data center companies have said ‘we are willing to do this.’”

AI-fueled growth in 2024

Artificial intelligence has emerged as a driving force behind data center expansion, fundamentally transforming the scale and nature of demand on these facilities. 

Unlike streaming or data retrieval applications that defined earlier decades of growth, AI systems require extraordinary computational power, pushing energy consumption at data centers to unprecedented levels.

The energy demands of AI-driven workloads are staggering. At 2.9 watt-hours per ChatGPT request, AI queries are estimated to require 10x the electricity of traditional Google queries, which use about 0.3 watt-hours each, according to the Electric Power Research Institute. 

Emerging applications, such as real-time video generation and AI-based image processing, push these requirements even higher, with no precedent for their computational intensity.

With the shift to cloud computing and AI, new data centers are growing in size and capacity. It is not unusual to see new centers being built with capacities from 100 to 1,000 megawatts of power — ten times that of traditional operations – and roughly equivalent to the electricity needs of 80,000 to 800,000 homes.

The shift to AI and cloud computing has introduced a constant and unpredictable demand for processing power, placing added pressure on power grids and infrastructure.

“Data centers are not new. But the way we’re using them is,” said National Telecommunications and Information Administration Administrator Alan Davidson in September, speaking at a Center for Strategic and International Studies event on data center growth. “Even before artificial intelligence supercharged demand, data centers had become a linchpin of our digital economy.”

Engines of growth and energy guzzlers

Nowhere is this strain more evident than in Northern Virginia, known as “Data Center Alley,” where data centers reportedly consume 24% of Dominion Energy’s power. While estimates vary about how much of U.S. electricity is consumed by data centers, some projections have the figures rising to between 4 and 6% by 2026, and even as much as 12% by 2028. 

While national growth is striking, the industry’s geographic concentration amplifies local challenges. Just 15 states account for 80% of the national data center load, with hubs like Northern Virginia; Prineville, Oregon; and Dallas-Fort Worth, Texas, leading the way.

This dramatic growth has reshaped local economies and sparked both opportunities and challenges. These regions have benefited from job creation and infrastructure investments, but the strain on energy resources and local environments has raised alarms.

States are responding with ambitious strategies to attract investment. For example, New Jersey launched a $500 million tax credit initiative in 2024, aiming to position itself as a hub for AI companies and cutting-edge data center technology. Yet, such incentives come with increasing concerns about sustainability and energy reliability, as communities and policymakers question how to balance economic growth with environmental impact.

Becky Coleman of Gate City in Scott County captured these tensions in a piece published in Broadband Breakfast: “The growth of data centers and the availability of energy have got to pace each other. Otherwise, there will be a rush to build data centers with a possibly thoughtless approach to powering them, particularly in rural areas.”

Growing opposition and NIMBYism

The rapid pace at which data centers are being proposed and built has heightened public fears about unchecked development. In some communities, residents feel that their voices are not adequately heard in the planning process, leading to frustration and distrust of both the companies behind the data centers and the local governments that approve these projects.

The outer suburbs of northern Virginia have long fought against the proliferation of data centers in their area. Public outcry against further development came as Republican Gov. Glenn Youngkin embraced data centers in an announcement to invest $35 billion into data centers by 2040. Loudoun County residents called for a moratorium on all new data center construction in March.

The perception that data centers bring limited local benefits — such as relatively few jobs compared to their environmental and land use impacts — further fuels opposition, as communities question whether the trade-offs are worth it.

The push for greener data centers

Governments at various levels have taken up the call to drive towards more sustainable data center operations. Federal programs like the Department of Energy’s Better Buildings Initiative encourage data centers to adopt energy-efficient technologies by offering technical assistance and recognition for achieving significant energy savings.

State and local governments play a pivotal role. In states like Virginia, the government introduced incentives such as sales and use tax exemptions for data centers that meet certain energy efficiency standards. These incentives are designed to attract data center investment while ensuring that new facilities are built with sustainability in mind.

States like California, with its aggressive renewable energy targets, have pushed data centers to align with broader state goals, requiring higher percentages of energy to come from renewable sources and setting standards for energy efficiency.

Nuclear energy emerged as a viable and increasingly discussed option for powering data centers. Companies like Amazon, Microsoft, and Google have taken steps to explore nuclear power’s potential to meet the immense and growing energy demands of artificial intelligence processing and cloud computing.

Microsoft, for instance, has entered into an agreement with Constellation Energy to restart energy production at the Three Mile Island plant in Pennsylvania by 2028, while Google is planning to deploy small modular nuclear reactors by 2030 through a partnership with Kairos Power. These reactors, which are compact and designed for localized power needs, could provide data centers with stable, low-emission energy, reducing reliance on fossil fuels.

However, nuclear energy for data centers is not without its challenges. In November, the Federal Energy Regulatory Commission rejected Amazon’s bid to power its data center in Salem Township, Pennsylvania, with energy from the nearby Susquehanna nuclear plant. Regulators expressed concerns over grid reliability and consumer costs, highlighting the complexities of integrating nuclear power into data center operations.

Granholm, a staunch advocate for innovative green energy solutions, posed a challenge to the industry at the 2024 ARPA-E Energy Innovation Summit in May: “Let’s try building modular data centers for AI computing that slash the power needed for cooling servers.”

Federal agencies address data center challenges

In September, NTIA took a proactive step by seeking feedback from data center operators on the obstacles hindering growth and how policymakers might mitigate them.

Access to necessary power and water were key concerns, but the White House’s top telecom advisor also asked for feedback on workforce availability, data security challenges, permitting challenges and land availability, and the resilience of supply chains.