What DOE's Load Growth Playbook Means for Site Selection
U.S. data centers used about 176 TWh of electricity in 2023 and could use 325 to 580 TWh by 2028, according to a DOE-commissioned report from Lawrence Berkeley National Laboratory. The U.S. Department of Energy argues that this surge can be met with clean generation, storage, grid upgrades, efficiency, and flexible demand. For developers, that turns the siting question into a power question: the best sites will be the ones where grid headroom, clean supply, and buildable land overlap.
For two decades, U.S. electricity demand was essentially flat. That era is over. In its resource hub, Clean Energy Resources to Meet Data Center Electricity Demand, DOE notes that total load growth demand could grow 15% to 20% over the next decade, driven by AI, data center expansion, new domestic manufacturing, and electrification. Data centers are the sharpest edge of that growth, projected to reach up to 9% of U.S. generation by 2030, up from about 4% in 2023.
How fast is data center electricity demand growing?
The most cited numbers come from the 2024 United States Data Center Energy Usage Report, which Lawrence Berkeley National Laboratory prepared for DOE under a congressional mandate. Its findings, summarized in DOE's release:
Data centers used 58 TWh in 2014 and 176 TWh in 2023, roughly a tripling in under a decade.
That 2023 figure equals about 4.4% of total U.S. electricity.
By 2028, consumption is projected at 325 to 580 TWh, or roughly 6.7% to 12% of national electricity use.
The width of that 2028 range is the story. A 255 TWh spread between the low and high cases reflects real uncertainty about chip efficiency, cooling, and how many announced campuses actually secure power. In practice, the binding constraint is not demand. It is how quickly new load can be interconnected at specific substations.
What does DOE's clean energy playbook actually include?
DOE's page reads like a catalog of federal levers, but grouped by theme it points to five ways the agency expects data center load to be met. Each has a siting implication.
What DOE emphasizes | Example programs on the DOE page | What it means for site selection |
Grid-scale clean generation | Title 17 loans (1703, 1706), hydropower incentives, SMR pathway, i2X | New supply clusters near retired plants, dams, and existing interconnection points |
Grid expansion and enhancement | GRIP ($10.5B), Transmission Facilitation Program ($2.5B), reconductoring tool, NIETC corridors | Watch upgrade-funded substations and corridors where headroom may open up |
Data center efficiency | LBNL Center of Expertise, DCEP training, Better Plants, 179D | Lower PUE stretches every MW of secured capacity further |
Demand flexibility | Connected Communities 2.0 ($65M), Distributed Energy Systems Demos ($50M) | Flexible load can win faster interconnection in constrained markets |
Onsite and behind-the-meter power | Onsite Energy Program and TAPs, CiFER storage pilot | Land suited to solar plus storage beside the load gains value |
1. New clean generation, often on old energy sites
DOE highlights loan programs for innovative clean energy and for repowering or repurposing retired energy infrastructure, plus support for hydropower, small modular reactors, and existing nuclear. A recurring theme is reuse: retired coal plants and non-powered dams already sit on interconnection points and transmission. For data center developers, those sites and the land around them deserve a closer look, especially where existing power infrastructure can carry new generation and new load together.
2. More capacity from the grid that already exists
Grid programs on the page include GRIP grants, transmission financing, National Interest Electric Transmission Corridors, and a reconductoring analysis tool. Reconductoring, which replaces existing lines with higher-capacity conductors, is one of the fastest ways to raise transfer limits without new rights-of-way. Developers who track transmission data and available transfer capacity can spot where those upgrades are likely to open headroom before the market prices it in.
3. Efficiency that stretches every megawatt
DOE points to national lab facilities running at a power usage effectiveness (PUE) of 1.03 and to an Energy Efficiency Scaling initiative targeting a 1,000x gain in microelectronics efficiency over two decades. Efficiency will not erase demand growth, but it changes the math on a constrained site: a lower PUE means more compute per megawatt of secured offtake capacity.
4. Flexible demand and onsite power
Demand flexibility and onsite generation are where clean energy and data center siting converge most directly. A campus that can shift load, run onsite storage, or draw from co-located solar is easier for a utility to connect and less exposed to interconnection delays. That makes nearby land with strong solar resource and room for battery storage a real asset in data center site selection, not just an ESG add-on.
5. Planning and permitting support for states and communities
DOE also lists technical assistance for public utility commissions, state energy offices, and local governments, including renewable siting support for communities. Where states and utilities are actively planning for large loads, timelines tend to be more predictable. Where they are not, community opposition and tariff uncertainty can stall a project even when power is technically available.
Has the federal approach changed since this page was published?
Yes, in emphasis if not in diagnosis. Many programs on DOE's page were funded through legislation passed between 2021 and 2022, and their status has continued to evolve. In 2025 DOE issued its Speed to Power request for information, which seeks large-scale generation, transmission, and grid projects that can support between 3 GW and 20 GW of new load for AI, data centers, and manufacturing. The RFI states plainly that projected demand from large users is outpacing the existing grid, and that these loads are highly location-specific and depend on rapid interconnection.
The common thread across both is location. Whether the resource is solar plus storage, reconductored transmission, a repowered coal site, or a new nuclear unit, the value only materializes where the grid can deliver it to a buildable parcel.
What this means for data center and renewable developers
DOE's framing reinforces a shift LandGate sees across the market: power availability, not land price or tax incentives, now decides which sites move forward. A few practical takeaways:
Screen for power first. Start with substation offtake and available transfer capacity, then look for land. LandGate's data center offtake insights show where large loads can realistically connect.
Treat renewable land as a power source. Parcels suited to solar and storage near a strong substation can support behind-the-meter supply and faster interconnection.
Follow the upgrades. Transmission and reconductoring projects can change a market's headroom within a few years. Track them alongside queue data.
Verify federal support before you underwrite it. Program funding and priorities move. Confirm current status directly with DOE before a grant or loan becomes part of the capital stack.
Renewable developers should read the same signals from the other side of the meter. Hyperscale and colocation buyers are increasingly willing to contract directly for clean supply near their campuses, which makes data centers a new class of offtake counterparty for solar, wind, and storage projects.
Frequently asked questions
How much electricity do U.S. data centers use?
According to the 2024 Lawrence Berkeley National Laboratory report commissioned by DOE, U.S. data centers consumed about 176 terawatt-hours (TWh) in 2023, roughly 4.4% of total U.S. electricity. That is about three times the 58 TWh they used in 2014.
How much will data center electricity demand grow by 2028?
LBNL projects U.S. data center consumption of 325 to 580 TWh by 2028, or about 6.7% to 12% of national electricity use. The range depends on AI chip adoption, cooling technology, and how quickly new capacity actually comes online. DOE's broader framing on its clean energy resources page puts data centers at up to 9% of U.S. generation by 2030.
What is DOE's position on data center load growth?
DOE frames data center growth as an opportunity to accelerate clean energy buildout, improve demand flexibility, and modernize the grid. Its resource page groups federal support into five areas: grid-scale clean generation, grid infrastructure expansion, data center efficiency, demand flexibility, and technical assistance for states and communities.
Can clean energy alone power new data centers?
Not on its own in most markets. Data centers need around-the-clock power, so solar and wind are usually paired with battery storage, firm resources like nuclear, hydro, or gas, and grid supply. The practical question for developers is which combination is available at a specific site, which depends on substation offtake capacity, transmission headroom, and land suited to onsite generation.
Why do retired coal plant sites matter for data centers?
Retired coal sites often have existing high-voltage interconnection, transmission access, industrial zoning, and water rights. DOE's Title 17 Energy Infrastructure Reinvestment (1706) program and its coal plant redevelopment visualization tool both target these sites, which makes them strong candidates for co-located generation and large loads.
What is demand flexibility and why does it help data centers?
Demand flexibility means a large load can reduce or shift consumption during grid stress, for example by curtailing noncritical compute or running onsite storage. Utilities and grid operators can often connect flexible loads faster because they do not require the grid to be built for the absolute peak. DOE funds grid-edge demonstrations through programs like Connected Communities 2.0.
Has federal data center energy policy changed since the DOE page was published?
Priorities have shifted. In 2025 DOE issued its Speed to Power request for information, seeking large generation, transmission, and grid projects able to support at least 3 GW and up to 20 GW of new load for AI, data centers, and manufacturing. Developers should confirm the current status and funding of any specific program before building it into a project plan.
How can developers find sites where clean power and grid capacity line up?
Start with power, not parcels. Screen available transfer capacity and substation offtake, map nearby transmission and power infrastructure, then evaluate surrounding land for solar and battery storage potential. LandGate's data center platform combines these layers so teams can compare sites on the same basis.
Find where clean power and grid capacity line up
LandGate maps offtake capacity, ATC, transmission, and land across the U.S. so developers can compare sites on power first. Explore LandGate's data center site selection tools, pull the same layers into your own models through the LandGate API and data exports, or book a demo to see how it works on your target markets.
