We’re leading an all-out national mobilization to defeat the climate crisis.

Join our work today to help us build a thriving and just clean energy future. 

Donate

We’re leading an all-out national mobilization to defeat the climate crisis.

Join our work today to help us build a thriving and just clean energy future. 

Four Ways States Can Meet AI Energy Demand With Clean Energy

With key policy interventions, we can address data center demand and hit our climate goals

Graphic: a solar panel connected to states and those states connected to a data center.

© 2024 Andrew Hartnett/Evergreen Action. States: GISGeography. Data center: AI

What Is the Impact of Data Centers on Energy Demand and Climate?  

Data center construction is booming, due to the rise of artificial intelligence (AI) and its increasing power demands. This electricity load growth threatens to blow a hole in states’ clean energy and climate targets and utility ratepayers’ pockets—unless it is met with rapid clean energy buildout and a proactive policy response. 

Data centers are projected to increase their consumption of annual electricity demand from 4 percent in 2023 to 9 percent in 2030. It’s clear that data center load growth is already threatening state climate plans: Utilities across the country are attempting to justify new gas plants and delay coal retirement plans in response. It’s time to change course before this manageable problem derails climate progress.

The question ahead for lawmakers and advocates is how to manage this potentially useful technology while ensuring the companies behind it cannot push their costs and their pollution back into communities.

We have already seen unintended consequences in communities where data centers have been built. From harmful non-disclosure agreements making it difficult for communities to understand development impacts to noise pollution and strain on water resources to lower-than-expected job creation, states’ overly industry-serving approach to data center development is having real negative consequences.

 

Projections of Potential Electricity Consumption By U.S. Data Centers

Line chart: U.S. data centers will increase electricity consumption by 3.7-15% between 2023 to 2030.

Data centers are projected to increase their consumption of annual electricity demand from 4 percent in 2023 to 9 percent in 2030. Graph: 2024 Electric Power Research Institute (PDF)

What Is the Role of Policymakers, Advocates, and Tech Companies?

We've seen tech booms and busts before. Let's not let this one revive fossil power and leave states with pollution and grid problems. Policymakers need to balance the hype with sensible laws to protect communities. We need binding laws and regulations, not just corporate goodwill, to ensure clean energy meets growing data center demands. 

Without policy, increased electricity needs will lead to more fossil fuel use and grid pollution. While some tech companies are building new renewables, others are buying power from existing projects or simply buying power from the grid. If new demand from these data centers is not met with new renewable energy, this demand will be met by running existing fossil plants more often or by building new ones.

Without policy intervention now, states will struggle to hit their climate targets, and utility monopolies and tech companies could pass the cost of generation and grid needs onto everyday customers. We can meet this challenge with clean energy, and it’s important we get this right.

 

How Can States Address AI Energy Demand and Hit Their Clean Energy Targets?

States are almost always in the vanguard when confronting new policy puzzles. Absent a federal clean electricity standard, states can pursue these four solutions, right now, without Congress:

 

1. Requiring additionality and the “three pillars”

First, we should make sure that data centers don’t just claim existing clean energy for themselves, while the wider grid ramps up fossil fuel plants to meet higher demand. Tech firms must be responsible for funding the new clean energy needed to meet their needs. To do that, we can borrow from a proven approach Evergreen and other advocates are calling for in the buildout of truly clean hydrogen, referred to as the “three pillars.” 

To ensure that data center load doesn’t increase grid pollution, its electricity must come from clean energy that has three key attributes: additionality, regionality, and time matching. These three pillars, like those needed to ensure clean hydrogen is truly clean, will ensure that renewable energy serving data centers is newly built and is generated in the same location and time that it is needed. Without these three attributes, data centers will still be increasing pollution on the grid, even if they are claiming “clean” energy on paper.

Some tech companies are already pursuing additionality and hourly-matched renewables to serve their data centers. But others are not: Amazon is claiming existing nuclear power for one data center in Pennsylvania. By claiming existing clean power that otherwise would have gone to other users and not building any new clean energy to compensate, this project will increase pollution on the grid. Other companies are making no renewable energy purchases at all—Goldman Sachs has projected that 60 percent of data center load is set to be served by gas plants through 2030.

State policymakers need to step up. In the absence of a federal clean energy standard, or federal three pillar requirements, states can require additionality and the other pillars on their own. In states with tax credits or incentives for data centers, lawmakers should condition that support on additionality and hourly matching requirements.

States without these incentives could require all new data centers (or those above a certain size) to provide additional and time-matched renewables to meet their power needs. Similar requirements could be set for cryptocurrency mining or other activities with extreme power demands and relatively small economic benefits. Without these safeguards, state climate targets are at risk.

2. Protecting customers from higher bills

States can also create safeguards for consumer costs to prevent some of the world’s wealthiest companies from passing their costs onto everyday customers. When massive new hyperscale data centers are built, grid costs can skyrocket. These facilities need both new power generation and accompanying grid infrastructure, potentially including new or upgraded transmission and distribution lines and substation upgrades.

The cost of these upgrades is likely to be socialized and passed on to other customer classes less able to afford them, including residential ratepayers, small businesses, and low- and moderate-income households.

States, either through legislation or action by the public utility commission, should require utilities to create specialized tariffs and rate classes for data centers. These actions would prevent the cross-subsidization by other ratepayers of this new inflexible load and would ensure data centers are paying their fair share.

Specialized green tariffs (voluntary utility programs that allow customers to buy renewable energy at a special rate) could incentivize tech companies to pay more and support the growth of new clean energy sources. Green tariffs could include new or advanced nuclear plants or other clean technologies that, despite higher construction costs, can provide consistent, around-the-clock power, better meeting the specific needs of data centers.

Duke Energy’s new Accelerating Clean Energy tariffs allow tech companies to (voluntarily) bear the cost and risk of bringing new nuclear and long-duration storage online. They allow companies to achieve their own clean energy goals earlier without pushing costs onto other customers. These tariffs should be designed to encourage flexibility in energy demand and to reflect the full cost of new data center energy demand. Since data centers often run at high capacity around the clock, they put extra strain on the grid compared to other electricity users, and the huge profits that AI creates allow these companies to afford higher rates that fully pay for needed upgrades.

 

3. Creating interconnection fast tracks

The long timeline needed to build new power generation is a critical bottleneck to powering AI and data center growth with clean energy. Interconnection fast tracks could speed up clean energy deployment and allow new data centers to come online faster. Utilities and grid operators can offer expedited interconnection tracks and faster interconnection studies to data center projects under certain conditions.

State lawmakers or utility regulators can require the creation of such fast tracks. To receive such carrots, tech companies could be required to sponsor or assist in interconnection studies, agree to pay for transmission upgrades, or provide certain amounts of renewable energy or energy storage behind the meter. By providing additional resources to unclog the interconnection backlog, data centers can be built faster and all customers will benefit from new, low-cost clean energy coming online sooner.

Drone shot of a Google data center in Iowa

States could require data centers, like the one pictured above, to co-locate with solar or battery storage to reduce their overall energy need and speed up interconnection. Google Data Center in Iowa. © 2024 Chad Davis/Flickr CC BY 2.0

4. “BYO clean energy” and efficiency requirements

States could also require new data centers to bring their own clean energy to the table. These requirements can either be set in legislation or as a condition for tax credits or other incentives. Behind-the-meter requirements would ensure that data centers are co-locating with solar or battery storage, reducing their overall energy needs, providing backup generation, and increasing their demand flexibility.

PJM has recently released guidance on how to site new load behind the meter of existing generation. This would speed interconnection by co-locating load and generation and reducing the need for grid upgrades.

Many companies have already seen the benefit of behind-the-meter generation. But some of these data centers are relying on dirty diesel or gas generators for on-site power. These are some of the dirtiest technologies around. By requiring behind-the-meter clean generation and storage, states can reduce fossil pollution in communities, along with the overall load growth that data centers cause.

It’s important to add that states can use their air permitting tools to require truly clean backup power rather than allowing more combustion generators. This is because air permit laws generally require the best technologies to be used, and those technologies, increasingly, are zero-emission battery storage options.

Energy efficiency requirements are also a key tool in keeping load growth manageable. Efficiency improvements reduce costs for tech companies and minimize the amount of new power they need to find to serve their project. ENERGY STAR and the U.S. Department of Energy both provide assistance on data center energy efficiency. States should set requirements for or condition their incentives on energy efficiency.

 

Why Do We Need More Transmission Lines? 

It makes sense to require tech companies to solve the load growth problems they are creating by procuring their own renewables. But without transmission buildout and faster interconnection, these clean energy projects will take too long to come online. Speeding overall transmission deployment is the best way to prevent AI from blowing a hole in the clean energy transition.

If three pillar requirements don’t go into force everywhere, we can still help get more clean power online by proactively building more transmission and clean energy around the country. That will help states to stay ahead of load growth and make sure there is enough new clean energy to meet all this demand and reduce pollution from existing fossil fuel power plants. 

States have a vital role to play here. Governors and utility regulators can pressure or require grid operators and utilities to plan transmission lines and speed up interconnection queues. Legislators can also reform renewable energy and transmission siting, as states like Michigan and New York have done. Instead of waiting years to build new transmission infrastructure, state utility regulators or legislators can also require utilities to study or deploy grid-enhancing technologies and reconductoring to get more capacity out of the existing grid.

Legislators can also create shared savings mechanisms for these technologies, so that their deployment is in utilities’ financial interest. States with existing clean energy standards and annual renewable energy credits (RECs) can also build momentum toward 24/7 clean energy by pioneering hourly matching requirements and time-matched RECs. There’s no transition without transmission.

 

Will AI Accelerate or Hinder the Clean Energy Transition? 

The rise of AI is creating challenges for the energy transition, as some fossil fuel plants are being extended to meet data center demand. But this isn’t some natural, inevitable process. It’s a choice by tech companies and utilities, one that states can reshape by insisting that tech companies take real responsibility consistent with their own green corporate pledges.

There are two paths ahead. In one scenario, policymakers do nothing, allowing utilities to justify a boom in new gas funded by their customers. In this world, our clean energy targets are increasingly out of reach and even well-meaning states fail to achieve their climate deadlines. 

In the other world, policymakers require tech companies to live up to their green promises. Here, large tech companies embrace the opportunity to unlock new clean energy to solve the problem their growth has created, allowing states and the country to pursue the path to 100 percent clean energy and lower energy bills. In this scenario, we get to enjoy the benefits of AI innovation while still living in a safer climate future. Which route we take is up to us.

 


 

Headshot of Charles Harper

Author - Charles Harper

Charles is the power sector senior policy lead for Evergreen. He leads Evergreen’s campaign to fully decarbonize the electric sector by building a thriving clean energy economy.

Headshot of Medhini Kumar

Editor - Medhini Kumar

Medhini is the writing/editing digital lead for Evergreen. Through powerful storytelling, she hopes to help move the needle on climate policy and contribute to our collective fight for a livable planet.