How to Review a Large Load Tariff for a Data Center Deal
The tariff now shapes underwriting as much as the lease does.
Utility Dive reports that 77 large load tariffs are pending or in place across 36 states, and 29 were approved in 2025. That is the clearest sign yet that data center developers are no longer negotiating only with landlords and tenants. They are negotiating with rate design, cost allocation and utility policy.
For institutional teams, this belongs in Data Center Due Diligence, Power Analysis for Data Centers and AI Site Selection. The tariff now affects whether a project is financeable, not just whether it is connectable.
1. Start with the load threshold.
The first question is simple: when does the tariff apply?
Thresholds vary by state and utility. Some trigger at 20 MW or 25 MW. Others use 50 MW, aggregate load across nearby facilities or a mix of individual and portfolio-level tests.
AI can pull the filing, extract the threshold language and compare it across jurisdictions in minutes. Human judgment decides whether the project should be structured to stay below the threshold, split across entities or reworked around a different service class.
2. Read the cost-causation language.
Large load tariffs exist to make heavy users pay for the grid they consume. The common terms are upfront engineering study costs, transmission and distribution upgrades, minimum contract terms, load ramp schedules, exit fees and financial security.
AI is good at spotting these clauses, comparing versions and flagging what changed between filings. It is also good at building a clean summary for the development team. It is not good at deciding whether the language is commercially survivable.
That part belongs to utility counsel, commercial leads and the people who know how much flexibility the tenant really has.
3. Model the deal as if the tariff will be enforced.
Tariff review is a financial exercise, not just a regulatory one.
If the tariff requires collateral, take-or-pay commitments or long minimum terms, it changes the project's IRR, DSCR and exit value. If it pushes more upgrade cost onto the developer, it can turn a good site into a marginal one.
Investment Committee Memo workflows should include tariff sensitivity just like capex, lease-up and power timing. AI can run the scenario math fast. Humans need to decide which assumptions are real and which are wishful.
4. Compare territory by territory.
The state picture is moving quickly.
Oregon, Minnesota, Indiana, Texas and Virginia have already passed laws governing large load customers, and Pennsylvania has developed a model large load tariff framework. That means the same project can have very different economics depending on where it lands.
The right question is no longer, 'Can we get power?'. It is, 'What does power cost, how long is the commitment and who pays for the upgrade path?'
5. Decide whether self-supply is cheaper than the tariff.
Some developers are bringing their own capacity or flexing demand during periods of grid stress to improve treatment. Others are building around behind-the-meter generation or storage.
AI can monitor which utilities are rewarding flexibility and which are punishing it. It can also track docket changes across multiple states without someone refreshing regulatory websites all day. But it cannot decide whether a self-supply strategy is cheaper than living inside the tariff.
That call depends on capital cost, schedule risk, tenant appetite and the local utility's actual posture.
6. Bake the tariff into the power schedule.
This is where CRE Automation pays off.
A good workflow takes docket language, rate tables and utility correspondence and turns them into a running deal memo. It tells the team when a clause changes, when a proceeding accelerates and when the project needs legal escalation.
If a site's power path relies on a clean tariff outcome, the tariff is part of site selection. If the team treats it as a late-stage admin task, the underwriting is already wrong.
FAQ
Q: What is a large load tariff?
A: It is a special rate and service framework for very large electricity users, usually data centers or industrial loads. The goal is to make the customer pay the true cost of serving the load instead of shifting that cost to other ratepayers.
Q: Why does it matter for data center underwriting?
A: Because it affects cost, timing and risk. A tariff can change the amount of collateral, the length of the contract, the upgrade bill and the exit economics of the deal.
Q: What can AI do in tariff review?
A: AI can track docket filings, extract threshold language, compare tariffs across states and summarize changes quickly. It can also flag which terms changed and what they mean for the project team.
Q: What still needs human judgment?
A: Utility interpretation, legal review and negotiation strategy. The model can surface the issue, but people decide whether the tariff is tolerable, avoidable or a reason to walk.