When a new data center needs hundreds of megawatts, the toughest question is often not “can the grid serve it?” The toughest question is who pays for grid upgrades for data centers. The answer depends on where the upgrades occur, how the project connects, and which regulatory bucket the costs land in.
This guide explains the major cost pathways in plain English: interconnection costs, distribution upgrades, regional transmission planning, and rate cases. It also explains why public backlash is rising in high-growth regions, and what both developers and communities can ask to make cost responsibility clearer.
Key takeaways
- There is no single rule. “Who pays” depends on the utility, ISO/RTO tariff rules, and state regulation.
- Interconnection is only one cost bucket. A project can pay interconnection facilities and still drive broader upgrades that flow into rates.
- Transmission planning can socialize costs. Some upgrades enter regional plans and are allocated across many customers, which is where ratepayer debates intensify.
- Rate cases decide the retail outcomes. Even when federal rules apply to transmission, state rate cases often determine how costs are recovered from customers.
- Forecast risk matters. If load projections do not materialize, customers may still pay for infrastructure built “in anticipation.”
Why “who pays” became a major data center issue
Data centers are scaling faster than many parts of the grid can. Upgrades take years, while development timelines are measured in quarters. That mismatch pushes utilities to build ahead, pushes grid operators to redesign queue rules, and pushes regulators to revisit rate design and special tariffs.
In PJM, public discussion has escalated because large load forecasts and market outcomes are colliding with affordability concerns, including debate over how projected loads are influencing costs before facilities are online. (See: Utility Dive coverage of PJM cost discussions.)
If you want the foundational grid context first, these are helpful companions:
- How data centers interact with the power grid
- How utilities plan for data center growth
- Data centers and grid reliability explained
The four buckets of “grid upgrade costs”
When people say “grid upgrades,” they often lump together costs that are governed by different rules. Separating these buckets is the fastest way to understand who pays.
1) Interconnection facilities
These are the direct facilities needed to connect a specific site, such as a dedicated line from a substation to the facility, breakers, relays, metering, and certain substation work required only for that customer. These costs are often directly assigned to the project, depending on the tariff and the utility’s rules.
2) Network upgrades triggered by interconnection
These are broader upgrades required because the new load changes flows, reliability performance, or contingency results. For generation interconnection, federal rules and ISO/RTO tariffs are central. For large load interconnection, frameworks vary by region and may run through retail utility processes. Either way, this is where projects can face major, unexpected costs.
3) Distribution system upgrades
Distribution upgrades include feeder capacity, local substation expansion, transformers at lower voltages, and protection changes. These costs are typically regulated at the state level through utility commissions and recovered through retail rates or special tariffs, depending on how the utility is allowed to treat large loads.
4) Regional transmission planning upgrades
Transmission upgrades can enter a regional plan when needed for reliability, public policy, or economic reasons (terms vary). Once an upgrade is in a regional plan, costs may be allocated across a broad set of customers using formula rates and tariff-based allocation methods. This is the bucket that most often creates “ratepayer subsidy” headlines.
How cost allocation works in practice
Cost allocation is ultimately about two questions:
- Causation: Did the data center cause the need for an upgrade?
- Benefits: Does the upgrade benefit others besides the data center?
The argument for assigning more costs to data centers is straightforward: large new loads should pay for incremental infrastructure built primarily to serve them, so existing customers are protected.
The argument against full assignment is also straightforward: the grid is shared, upgrades can benefit reliability and future customers, and isolating costs to one customer can discourage investment and slow economic development.
In reality, most systems land somewhere in the middle. That “middle” is often decided in rate cases, tariff filings, and regional planning processes, not in a single negotiation.
The rate case: where “who pays” becomes a retail bill question
State utility commissions decide how utilities recover costs from customers. This is where large-load tariffs, special contracts, or new rate classes can be created to address data center growth. Michigan is a good example of this dynamic, where the Michigan Public Service Commission has signaled that cost allocation and rate design for large loads should be addressed in general rate cases and related filings rather than being fully resolved through standalone service approvals.
Rate-case mechanisms that often show up in “who pays” debates include:
- Large load tariffs: specific rates for high-usage customers, sometimes with minimum bills or demand charges
- Special contracts: negotiated agreements approved by regulators for a specific project
- Line extension policies: rules for how much a new customer must pay for distribution extensions
- Cost-of-service studies: analysis of whether a class of customers is paying its share
For communities, the key point is that “the data center pays for its connection” does not automatically mean “ratepayers are protected.” Protection depends on how the utility and commission treat upgrade costs, demand risk, and long-term recovery.
Transmission planning: why costs can become socialized
Transmission is planned and paid for differently from distribution in many regions. When upgrades enter a regional transmission plan, they can be spread across many customers. That can be justified when upgrades improve broad reliability. It becomes controversial when the primary driver appears to be a small number of very large new loads.
In PJM and other high-growth regions, this is where the public policy tension spikes. If planners approve large upgrades based on projected demand that is not yet online, customers may worry they are paying “up front” for growth that may not arrive on schedule.
Why colocation and “net load” complicate cost fairness
Another emerging complexity is colocation, where a data center sits next to generation or behind-the-meter resources and draws some of its energy without fully relying on the grid at all times. Depending on the structure, this can reduce the data center’s billed load for certain charges even though the facility may still rely on the grid during certain conditions.
Regulators and researchers have flagged this as a potential fairness issue because grid costs are recovered from customers through specific billing determinants. If large loads reduce their contribution to those determinants while still depending on the grid as backup, costs can shift to other customers unless rules evolve.

What data center developers should do before committing to a site
If you want to reduce unpleasant surprises, treat cost allocation like a diligence workstream, not an afterthought.
Due diligence questions to ask utilities and regulators
- What upgrades are required for the first phase, and which are required for full buildout?
- Which costs are directly assigned to the project, and which may flow into rates?
- Are there existing or proposed large-load tariffs, minimum bills, or special contract options?
- What is the utility’s process for cost-of-service validation for large loads?
- What happens if the project ramps slower than expected or does not reach projected load?
Project structure tactics that can reduce risk
- Phased energization: start with a smaller tranche while upgrades for later phases are built
- Contract alignment: avoid power supply contracts that assume energization before upgrades are complete
- Transparent load profile: provide realistic ramp schedules and operational assumptions
If you have not read it yet, this interconnection explainer pairs directly with the “who pays” question:
What communities and local governments should ask
If a new data center is proposed near you, the most important questions are not only about noise and traffic. They are also about power system costs and transparency.
- Upgrade scope: What substations, feeders, or transmission upgrades are required, and where will they be built?
- Cost pathway: Which costs are assigned to the project, and which costs are expected to be recovered from customers?
- Rate protections: Are there minimum bills, demand guarantees, or special tariffs designed to prevent cost shifting?
- Forecast risk: What happens if the project does not reach projected load?
- Public filings: Which docket numbers, tariff filings, or commission cases contain the relevant cost decisions?
For the broader community-facing view, these are relevant:
Frequently asked questions
Do data centers always pay for their grid upgrades?
No. Data centers often pay some direct interconnection and local upgrade costs, but broader transmission planning upgrades can be allocated across many customers depending on tariff and regulatory treatment.
Why would ratepayers pay for upgrades driven by data centers?
Because the grid is planned as a shared system and utilities recover many costs through rates. If an upgrade is treated as system-wide reliability infrastructure, it may be spread across customers, even if a small number of large loads are major drivers.
How do regulators prevent cost shifting?
Common tools include large-load tariffs, minimum bills, special contracts, cost-of-service studies, and performance conditions that require the large load to meet ramp commitments or pay certain costs if it does not.
Does colocation let data centers avoid paying for the grid?
It depends on the structure. Some arrangements reduce billed load for certain charges, which can raise fairness questions if the facility still depends on the grid for backup or intermittency support. Regulators are actively debating how to treat these cases.
Bottom line on who pays for grid upgrades for data centers
Who pays for grid upgrades for data centers is not one rule. It is a set of decisions made across interconnection processes, distribution upgrades, regional transmission planning, and state rate cases. The fastest path to clarity is to separate cost buckets, trace which regulator controls each bucket, and demand written answers early, before land, construction, and contracts lock in assumptions.
Sources and further reading
- FERC: Explainer on the Interconnection Final Rule (Order No. 2023)
- FERC: Explainer on Order No. 2023-A
- Utility Dive: PJM cost concerns and data center-related planning debate
- Michigan Public Service Commission: Consumers Energy data center service terms and rate case direction
- Johns Hopkins Energy Institute: regulator questions on transmission costs and colocation
- Union of Concerned Scientists: PJM data center transmission cost discussion (PDF)


