As 2025 winds to a close, there are at least 65 large load tariffs either pending or in place across 34 states across the country. No less than 46 of those tariffs were new this year, according to data collected by the Smart Electric Power Alliance.
You don’t have to look far to understand the significant increase in tariff activity. Utilities have long used such regulatory mechanisms to manage the costs that accompany new large loads joining the grid, like aluminum smelters or paper mills. But the data center boom has brought a distinct type of customer set: one that’s larger, operates at a higher load factor, and demands faster energization.
Though the level of activity around large load tariffs started to pick up in 2024, 2025 saw utilities scrambling to both attract load and protect themselves and their existing customers, explained Ryan Hledik, a principal at Brattle Group focused on regulatory economics and rates.
Of course, it’s not the case that a utility couldn’t previously serve a data center without a specific tariff. Rather, the existing mechanisms weren’t designed for the scale and impact of the hyperscale data centers that the AI industry has in the works.
“You don’t have to have large load tariffs; utilities have rates in place to recover costs from large customers,” Hledik said. “The reason it’s happening now is because utilities are seeing that these customers, in a variety of ways, are different from the rest of the customers they’ve been serving historically.”
Despite the massive uptick in new large load tariffs this year, the landscape is still relatively fractured. It would be overly generous to say that there are any clear “trends,” Hledik added, but themes are certainly starting to emerge. These include: protecting against the risk of stranded assets (in the event that a planned data center never comes to fruition or doesn’t end up needing as much power as planned), protecting ratepayers from increased costs, and making regions attractive for large loads and the economic benefits they could bring with them.
Trending mechanisms
On the first goal — protection against stranded assets — Hledik points to AEP Ohio, whose data center tariff was approved in July after a contentious proceeding at the state’s public utility commission that pitted hyperscalers against the utility. Data centers larger than 25 megawatts now must pay for 85% of their contract capacity each month, regardless of how much power they consume — a clear effort to circumvent the risks of an AI bubble or construction delays.
AEP Ohio was among the first to introduce minimum monthly billing requirements, but in the wake of that tariff’s introduction, many others are following suit, Hledik said.
And while it’s still early, AEP already appears to be feeling the impact of its tariff. In its third quarter earnings presentation the utility described a significant drop in its large load queue. Pre-tariff, that queue consisted of approximately 30 GW of anticipated data center load, leading AEP to temporarily pause new service agreements for data centers in 2023. In the months post-tariff, the queue has winnowed down to just 36 formal study requests, totalling 13 GW.
Lauren Shwisberg, a principal at RMI whose research focuses on grid planning and investment, said that when it comes to mechanisms for protecting ratepayers, 2025 has been an extremely active year.
“When we started the year, there were a lot of ideas but not a lot of convergence yet on what structures could look like,” Shwisberg explained. Twelve months later, there’s at least some consensus, if not a clear framework: “We are seeing a lot of convergence around minimum contract lengths,” she said, nodding to the more than a dozen tariffs this year that include 15 year term lengths. That’s up from the three-year contract length that was common under prior generations of tariffs.
Shwisberg also pointed to the rise of “backstop provisions,” including collateral requirements, exit fees if a large load leaves early, and capacity reassignment provisions. AEP Ohio’s new tariff, for example, requires customers to pay 100% of construction costs if they cancel or delay by more than 12 months. “A lot of these provisions are strengthening some of the structures that we’ve seen in historic tariffs,” she explained. “Collateral requirements have always been part of a contracting process for a large load customer…but what we’re seeing is a real strengthening, and making these terms a lot more explicit so that customers know what they’re getting into.”
For example, Hledik pointed to a tariff approved in November for Evergy Kansas, which featured another emerging pathway for protecting ratepayers from unexpected costs: a 10% “adder.” In Evergy territory, data centers will now pay a higher rate for power than standard industrial averages — around 10% higher — in order to account for the marginal costs of new generation, and, in theory, ensure other ratepayers aren’t subsidizing the massive infrastructure upgrades a data center may require. Evergy also requires customers to pay two years of minimum monthly bills up front as collateral.
As to the last theme Hledik identified — attracting data center customers — utilities are having to get creative. “Offering them a rate discount is pretty much off the table at this point,” he explained. Instead, some utilities are appealing to corporate decarbonization goals, introducing tariffs that would allow data centers to pay a premium to ensure the energy they’re getting is coming from carbon-free sources, he added.
Another potential pathway includes bring-your-own-capacity provisions, that could allow new load to jump the queue by bringing co-located or adjacent capacity.
The flexibility quotient
Notably absent from most approved tariffs are flexibility provisions. Throughout the year, the idea of flexible data centers has had the AI and energy industries abuzz, prompting the emergence of new startups and coalitions, hyperscaler-backed frameworks, and flexibility demonstrations.
But when it comes to demonstrated flexibility at scale, “the examples are still a little thin,” Shwisberg said. While there are one-off success stories — she pointed to the recent deal Aligned Data Centers struck with a utility in the Pacific Northwest to fund a 31 MW battery and get online faster — these remain bespoke contracts, rather than replicable tariff features.
That said, flexibility is likely to be a factor for tariff design in 2026. “The initial focus of a lot of utilities over the past year has just been ‘how do we get a tariff in place that recognizes that [data centers] are a different class of customer and get guardrails in place?’” Hledik said. “We are now very quickly moving into a situation where…flexibility is becoming more important.”
Shwisberg pointed to the potential for tariffs that create pathways for investment in onsite DERs or offsite flexibility, such as procuring capacity via VPPs. That approach is going to “require some more experimentation in 2026,” she added, but standardization is likely to begin next year.
For more on the mechanics of data center flexibility, listen to this recent episode of the Catalyst podcast:
The specifics will be defined in part by the outcome of an ongoing proceeding on interconnection at the Federal Energy Regulatory Commission. Back in October, the Department of Energy directed the commission to consider a framework for large load interconnection, and to consider expediting the study process for flexible loads, potentially limiting it to just 60 days.
DOE’s end goal for the rulemaking isn’t necessarily at odds with the goals of current tariffs, but they aren’t exactly in line with them either; the primary objective of the government’s proposal is speeding up deployment, which doesn’t always align with efforts to limit risk. The proposal also asks FERC to go further on operational flexibility than utilities have so far been comfortable with; they’ve worked to standardize how data centers pay for grid access via the wave of new tariffs, but have stopped far short of standardizing how they interact with the grid.
One of the key issues being debated in the overflowing large load docket is whether FERC has jurisdiction to weigh in on the interconnection process at all. State public utility commissions in particular are opposed to the proceeding, arguing against what they consider to be federal overreach. In its initial order, DOE gave the commission an admittedly ambitious deadline for final action by April 2026.
That impending ruling will land amid a patchwork of existing, more defensive tariffs, Shwisberg acknowledged, but will also give stakeholders more certainty to move forward on writing new tariffs.
