Throughout this week, the North Carolina Utilities Commission is meeting under bright fluorescent lights to discuss Duke Energy Carolinas’ latest rate case. The hearing, which is being live-streamed on YouTube, went on for nearly seven hours on Monday, as Duke Energy’s representative defended the utility’s 11.6% residential rate hike proposal, among other things.
The hearing is also expected to address the large load tariff proposal Duke Energy submitted late last month — a special rate for big energy users, similar to those popping up all over the country, amidst concerns that data centers’ taxing power needs will have negative repercussions for everyone else. (Just yesterday, the Oregon Public Utility Commission approved a new Portland General Electric tariff that would hike rates for data center customers by approximately 29%.)
Starting in 2027, Duke’s tariff would apply to customers with loads of 50 megawatts or more with expected load factors of 80%, or any customer seeking 100 MW or more, regardless of load factor. The terms of the tariff would require customers to sign 10 or 15 year contracts, depending on load size, and pay for 75% of the demand they ask for, even if they don’t use it. Pulling out of the deal early would incur penalties equivalent to 25% of remaining bills, and customers would have to post some sort of security, such as cash or a letter of credit, as a guarantee upfront for those minimum bills.
This is intended to introduce some transparency in contracts with large load users, which, at the moment, are executed behind closed doors. And, as Canary Media highlighted in its article about Duke’s proposed tariff, it’s an interesting change of heart for a company that, over the past few months, mostly claimed a dedicated tariff wasn’t necessary.
As my colleague Nick Zenkin, Latitude Media’s energy industry analyst who is closely tracking large load tariffs points out, Duke’s tariff could have large repercussions, given the utility’s 2035 forecast of electricity demand from large customers in the Carolinas is 8 GW.
Duke’s proposed requirements are “pretty light-touch” compared to other large load tariffs that are being implemented across the country, Zenkin said. Dominion’s, in neighboring Virginia, is the current gold standard, he added. Where Duke asks for a 75% minimum on contract demand, Dominion, in neighboring Virginia, requires 85% of transmission and distribution demand and 60% of generation, a key structural difference from Duke’s. Dominion also locks customers into 14-year contracts and makes them post $1.5 million per MW in collateral.
And most importantly, it’s not isolating data centers as unique customers.
“Duke is not making its own rate class; it’s adding onto its old large load one, layering elements on top of it,” Zenkin explained. “This doesn’t isolate the costs of data centers from the ones of other large load users, and keeps things fuzzy in the long-run.”
Because the tariff bands together every large, steady load above 50 MW, regardless of whether it’s a data center or not, it denies residents a clear view of how much data centers specifically are driving grid costs.
And it lumps conventional large users, such as manufacturers with data centers, leaving the former exposed to rate pressure driven by the latter. That would make Duke’s customers the latest example of a pattern emerging nationally, as manufacturers start to feel the negative consequences of the data center boom.
Earlier this week, for example, Reuters reported that manufacturing electricity bills are rising faster than for many homes and other businesses, due in large part to data centers’ power demand, with one company in Ohio reporting a surge of 90% last year. And last month, a long-planned First Ammonia plant in ERCOT was ordered to post collateral, which it says it can’t raise. That was under Texas’s new Batch Zero ruling, introduced to regulate new large loads because of high data center demand in the area.
The Public Staff of the North Carolina Utilities Commission is asking roughly $200 million of the $247 million for grid upgrades recently requested by Duke to be assigned directly to large load customers. That’s regardless of whether those are data centers or not, even though most of the 8 GW of large load forecast for 2035 is expected to come from data centers. It would have large manufacturing loads footing the bill for upgrades that exist mostly because of the data centers.
Duke, however, is resistant to assigning those costs directly to large loads. Under the proposal as it stands, the $247 million would be spread across every category, and everyone, including residential customers, would foot the bill for the grid upgrades. That’s because, Duke argues, everyone will benefit from these upgrades.
The issue is that it’s currently hard to know how much of the cost is coming from data centers. A dedicated data-center class, Zenkin argues, would make that easier to trace: you’d apply for the class, and the costs you cause would be yours. “If a data center causes you to build $10 million in grid infrastructure upgrades, the data center is then on the hook, because it’s literally by itself,” he said. “It’s very easy to separate the cost that the data centers are putting on, because they’re in their own bucket.”
A version of this story was published in the AI-Energy Nexus newsletter on July 8. Subscribe to get pieces like this — plus expert analysis, original reporting, and curated resources — in your inbox every Wednesday.
