Data Center Co-Location Explained: Who Pays for Grid?
Data center co-location explained starts with the question you are probably hearing in every reliability meeting right now: When a large AI data center ties in directly to a power plant, who pays for the grid services that still keep the system steady? You can move a lot of electrons “next door,” but the wider network is still the safety net when equipment trips, voltage sags, or the weather turns nasty. From where we sit at the Alliance for Competitive Power (ACP), the answer has to be simple and fair: Costs should follow use, and no one should be able to duck charges just because the wiring diagram looks clever.
Co-location is showing up fast in PJM, and you may be evaluating it as a regulator, utility, transmission owner, generator, market participant, or consumer advocate. Below, we walk you through what co-location means in plain English, why it is surging, what co-located load FERC action requires in PJM, and the practical questions you should be asking before any new arrangement gets blessed.
What Co-Location Means in Normal Terms
In the power world, co-location usually means a big electricity user, often a hyperscale or AI data center, is built at or beside a generating plant. The twist is where it connects. Instead of taking service like most customers through the shared transmission system, the data center connects on the generator’s side of the Point of Interconnection. In everyday language, it is “plugged into the plant” before the power hits the regional grid.
Federal regulators often call this a co-located load. If you want a crisp legal description, a Jones Day analysis lays out the concept as end-use load connected on the generator side of the interconnection point in PJM and why that distinction matters for tariffs and cost allocation. You can read that overview at Jones Day’s explanation of co-located load in PJM.
Why AI Developers Are Chasing Co-Location
If you are advising a project team, you already know the pain points: Long interconnection queues, uncertain network upgrade costs, and timelines that do not match the pace of data center construction. Co-location can look like a workaround. Pair the load with a generator, build a short physical tie, and you may avoid years of waiting for traditional transmission service arrangements.
PJM has become the center of the conversation because the region’s large-load growth is coming in hot, and the market signals are flashing. A widely shared industry breakdown notes how quickly data center demand has climbed alongside recent PJM reliability discussions and auction results. That summary is available at Introl’s PJM co-location guide.
Here is the part you should keep in view: Speed is not the same thing as system readiness. Co-location can be a smart siting choice, but it does not automatically erase the need for backup service, restoration support, or transmission planning. It just changes how those needs show up and who is on the hook for paying.
The “Free-Riding” Problem in Plain Sight
The argument is not really about whether co-location is “good” or “bad.” It is about whether a co-located data center can use the grid like an insurance policy without paying the premium.
Even when a data center takes most of its energy directly from the plant next door, it may still lean on the grid for things you cannot ignore:
Backup power when the generator trips or goes offline for maintenance.
Voltage and frequency support that comes from the synchronized network.
System restoration and resiliency after major disturbances.
Deliverability and operational flexibility when conditions change faster than contracts do.
If the tariff does not clearly assign responsibility for those services, the bill does not disappear. It lands somewhere else, usually on households and small businesses that do not have the leverage to negotiate bespoke arrangements. Competitive markets only work when charges line up with cost causation. Otherwise, you end up with quiet cost shifting and distorted investment decisions.
Evaluating Co-Location Realities
Transmission Cost Responsibility
Why it matters: If a large load avoids transmission charges but still relies on the network for backup and deliverability, other customers can pick up the tab.
Reliability and Operating Services
Why it matters: Frequency and voltage stability are shared benefits. If a co-located load benefits, it should contribute.
Interconnection and Deliverability Rules
Why it matters: How PJM studies exports and withdrawals affects queue timelines and the cost of upgrades for everyone.
Competitive Neutrality
Why it matters: Special carve-outs can tilt outcomes, favor certain business models, and undermine price signals.
Co-Located Load FERC Action: The PJM Reset
In December 2025, FERC stepped in and told PJM to write clearer, transparent tariff rules for co-location. The Commission’s message was basically, “innovation is fine, but the tariff has to be just and reasonable, and it has to spell out the rates, terms, and conditions.” You can review FERC’s announcement at FERC’s news release on directing PJM to create new co-location rules.
For you as a stakeholder, the practical takeaway is this: Co-location is not supposed to be a tariff blind spot. If a co-located load depends on grid services, the rules need to say how that dependence is measured and how charges are assigned. Transparency is not a nice-to-have here. It is the guardrail.
Matching Contract Options with Grid Dependence
One of the harder parts of the co-location puzzle is that co-located loads are not all the same. Some will rarely touch the grid. Others will lean on it often, especially during outages or ramping events. A “one size fits all” approach can either invite free-riding or discourage efficient investment.
FERC’s direction pushed PJM toward service options that better track a co-located facility’s net use of the network. A utility-focused legal write-up describes how co-located loads would still pay their share for transmission and reliability-related services, even if the grid is not their everyday supply source. That summary is available at Gravel2Gavel’s overview of FERC’s co-location order and implications.
Two concepts you will see discussed for transmission service are:
Firm Contract Demand: The load contracts for a defined level of service tied to expected net withdrawals from the grid, with firm delivery priority.
Non-Firm Contract Demand: Similar idea, but without firm priority. It may reduce cost, but it also reflects a lower level of guaranteed service.
Where you should press for detail is metering and settlement. If a data center uses the grid more than it said it would, the tariff needs a way to true-up charges. If it uses the grid less, the structure should not force it into paying for a level of service it is not taking. Getting that balance right is how you protect consumers and keep the door open for productive new investment.
Balancing Interconnection Rights and Queue Pressure
Co-location is also about interconnection rights and how much power a generator is planning to push onto the grid. If a plant has a large on-site customer, it may export less than its nameplate capability. That matters because interconnection studies, network upgrades, and timelines are driven by what the project says it will inject and withdraw.
If you are trying to reduce queue pressure without cutting corners on reliability, this is the nuance to watch. A framework that aligns requested interconnection rights with expected exports can help, but it only works if the rules also account for the moments when the co-located load swings back to the grid for supply.
Market Dynamics: Shifting Stakes and Customer Risk
You will hear very different views depending on where someone sits. Independent generators may see co-location as a path to long-term contracts that keep plants running. Transmission owners and utilities may worry about fixed grid costs being spread across fewer traditional customers if large loads migrate behind the interconnection point. Consumer advocates are right to ask whether any of this turns into a cost shift.
At ACP, we focus on the market guardrails that keep competitive outcomes intact. If you want a quick refresher on why we push for open, competitive electricity markets, you can start at ACP’s homepage. If you are working through how these issues show up in customer bills, you may also find it useful to review our explainer on how electricity rates are set in regulated versus competitive models.
Checklist to Ask Before You Approve Siting Plans
If you are reviewing a proposal, you do not need to be sold on the hype. You need a checklist that forces the project to be specific about reliability and cost responsibility. Here are the questions we think you should insist on answering up front:
How will net grid withdrawals be measured and audited? You need clear metering points, data access, and settlement rules that stand up to scrutiny.
Which grid services does the co-located load still rely on? Make the proposal name the services and the conditions when they are used.
What happens during generator outages and derates? Spell out the backup plan, including duration, magnitude, and any priority claims.
Where could cost shifting occur, and what stops it? Look for enforceable tariff provisions, not side letters and assumptions.
Does the structure preserve competitive neutrality? If one class of customer gets a better deal by exploiting tariff gaps, you will see it in market outcomes.
FAQ: Navigating Co-Location Tariffs
What does “data center co-location” mean in electricity markets? In this context, it means your data center connects on the generator’s side of the Point of Interconnection, so it can take power directly from a plant instead of relying solely on the shared transmission grid.
Why is co-located load FERC policy so focused on PJM? PJM is seeing rapid large-load growth, and FERC concluded PJM’s tariff needed clearer rates, terms, and conditions for co-located load to avoid unclear obligations and potential cost shifting.
Who pays for grid services when a load is co-located? The policy direction is that co-located loads should pay for the grid services they cause and benefit from, including transmission and reliability-related services, based on measurable net dependence on the network.
Does AI power plant co-location improve reliability? It can, especially if it supports investment and aligns new demand with available supply. It can also hurt reliability if the grid becomes an unpaid backstop. The difference comes down to metering, enforceable tariff terms, and fair cost allocation.
What details should you monitor as PJM finalizes its approach? Watch how PJM defines and measures net withdrawals, how contract demand service is structured, and how charges are applied during outages and other periods when the co-located facility leans on the grid.
Conclusion: Link Long-Term Assets to Measurable Network Use
Co-location can speed development and make smart use of existing generation sites. That is the upside. The risk is letting a new class of very large customers lean on the grid when they need it while everyone else pays for the readiness and reliability they benefit from.
Our view at ACP is consistent across these debates: Competitive markets deliver for consumers when prices and charges are transparent, when the rules do not hand out special carve-outs, and when cost responsibility matches actual grid use. If you want to track how this evolves and what it may mean for your state, you can follow our updates at ACP’s news page and dig into why cost shifting shows up so often under monopoly structures in our explainer on utility monopolies and consumer impacts.
