Why Data Center Power Deals Need Plan B
FERC’s rejection of much of PJM’s proposed large-load framework is a reminder that a promising power construct is not the same thing as a bankable customer commitment. If you sell, develop, or market data center capacity, build a Plan B into every power story before a prospect asks for one.
What changed at PJM?
PJM had proposed a framework for connecting data centers and other large loads, including interim service and a “bring your own new capacity” approach. FERC largely rejected it, finding that key parts could intrude on state jurisdiction or be discriminatory (Utility Dive).
The practical lesson is bigger than one filing. The market is trying to invent ways to connect very large new loads without shifting unreasonable costs or reliability risk onto everyone else. That is a legitimate problem. But the rules, responsibilities, and approvals are still being negotiated in public.
For an operator, the mistake is presenting a proposed construct as if it were settled commercial infrastructure. Sales teams understandably want a simple answer to “Can you get me 100 MW?” Yet “the market is considering a path” does not answer the questions a serious buyer, lender, or utility will ask next:
- Who has authority to approve the service arrangement?
- What exactly must the customer fund?
- What happens if transmission work slips?
- Is service firm, conditional, interruptible, or simply under study?
- What alternative operating plan exists if the preferred arrangement does not clear?
Those questions belong in the first substantive deal conversation, not the diligence scramble after a letter of intent.
Why does this matter beyond PJM?
Because large-load power has become a multi-party transaction. The data center customer may be willing to fund generation, transmission, or both. The utility has to plan a reliable system. The grid operator has its processes. State regulators may have a separate view of rate design and cost allocation. And the developer needs a delivery date it can put in front of its board.
Each party can be acting rationally while the overall deal remains uncommitted.
Look at the scale of concepts entering the process. PJM filed a study agreement for SoftBank/SB Energy’s proposed Ohio Flagship AI campus, which covers roughly 10 GW of data-center load, about 10.2 GW of gas generation, and three planned 765-kV transmission lines. Crucially, the study does not grant service or interconnection rights (Utility Dive).
That distinction matters even for a regional colocation provider nowhere near that scale. A study can validate that a concept deserves analysis. It cannot be marketed as delivered capacity. Buyers know the difference, and sophisticated channel partners will test whether your team knows it too.
What should a credible Plan B contain?
A Plan B is not a vague assurance that you have “multiple power options.” It is a documented operating and commercial path if the preferred source, tariff, or transmission milestone changes.
Start with a scenario table that sales, development, finance, and operations all recognize. Keep it plain enough to use in a customer meeting. For each scenario, state the available IT load, expected service conditions, dependencies, customer costs, and decision gates.
Your scenarios might include existing utility service, phased energization, customer-funded network upgrades, on-site or behind-the-meter supply, and a later expansion phase. Do not imply that every option is equivalent. The point is to make the trade-offs visible.
Then give each scenario an owner and an evidence trail. “Utility service” should connect to the relevant utility correspondence, executed agreement, or published process. “On-site generation” should identify fuel arrangements, permits, equipment lead times, operating limits, and who carries the risk. “Expansion after transmission completion” should clearly say what has been approved versus what remains subject to study or construction.
This work can feel conservative when demand is strong. Honestly, it is a selling advantage. A buyer with an urgent deployment does not need you to predict every regulatory outcome. They need you to explain what you know, what you do not know, and how their deployment proceeds in either case.
Can behind-the-meter supply be the fallback?
Sometimes. But it should be treated as a real power program, not an escape hatch on a slide.
Distributed supply is attracting meaningful commitments. Mainspring Energy announced an agreement to supply AEP with up to 1.6 GW of linear generators through 2034, with an initial 190-MW tranche scheduled to begin delivery in 2027 (Mainspring Energy). That is evidence that utilities are considering more than the conventional “wait for the next transmission project” playbook.
But a facility-level fallback has to withstand operational review. Can it run continuously or only during specified conditions? How is it fueled? What emissions, noise, and local permitting constraints apply? Does its output support the critical load you are selling, or does it reduce grid draw while the customer still needs utility supply? What happens during maintenance?
The same rigor applies to customer-funded generation. NiSource’s agreement with Microsoft for a northwest Indiana campus is expected to add roughly 1,100 MW to the utility’s resource pool; Microsoft is set to begin service in 2028 and fund generation and transmission needed for its load (NiSource). That model may be viable for a customer with scale, capital, and patience. It is not automatically a template for a mid-market buyer seeking a near-term, low-complexity deployment.
Match the power structure to the buyer’s actual constraints. A retailer migrating a disaster-recovery environment, an AI startup expanding a training cluster, and a hyperscaler reserving a future campus do not buy risk in the same way.
How should sales teams talk about uncertainty?
Use precise language. It earns more trust than a heroic answer ever will.
Replace “we have 50 MW available” with language such as: “We can support the first phase from existing service under these conditions. The next phase depends on these named upgrades, and here is the customer-funded alternative we are evaluating.” That sentence creates a useful next step: a technical and commercial review, rather than a later argument about what “available” meant.
Also separate four terms that are too often collapsed into one claim:
- Site capacity: physical space and design potential.
- Utility capacity: service the utility has agreed to provide.
- Deliverable capacity: power that can reach the customer under the stated timeline and conditions.
- Contracted capacity: power reserved under an executed customer agreement.
Your website, broker materials, and seller talk tracks should use the same definitions. If marketing says “future-ready” while the power team says “under evaluation,” prospects will notice the gap.
A good sales leader will also create an escalation rule. Once a prospect requests a load above the first firm tranche, or asks for a delivery date tied to a pending regulatory or grid milestone, bring in power, legal, and development before sending a proposal. That may slow an early call by a day. It can save months of rework at redlines.
Common questions
Does FERC’s decision mean large data center loads cannot connect in PJM?
No. It means PJM’s proposed framework was not accepted in the form presented. Individual projects and other arrangements can still move through applicable utility, grid, and state processes; do not treat the decision as either a blanket ban or a green light.
Should we stop using customer-funded power structures?
No, but qualify them carefully. Customer funding can align incentives and accelerate work, yet it does not erase approval, construction, operational, or jurisdictional risk.
What is the first document we should build?
Build a one-page power evidence sheet for each sellable site and phase. It should show the source of supply, current status, dependencies, dates only where substantiated, and the fallback path if a dependency moves.
Will buyers accept uncertainty in a power plan?
They will accept uncertainty better than concealment. What they usually reject is an ambiguous capacity claim that turns into a materially different deal once their engineers begin diligence.
Where this leaves you
The market will keep testing new ways to serve very large loads, and some will become durable models. Until then, sell what is firm, label what is conditional, and make the contingency plan part of the offer rather than a private internal document. GridReach helps data center and energy companies turn expertise like this into qualified pipeline.