Behind the Meter: How Energy Companies Win the Data Center Power Race

Data centers need power that utilities can't deliver fast enough. Behind-the-meter generation gives energy companies a direct path to premium offtake contracts.

Quick Reference

Hyperscale data centers are adding gigawatts of power demand faster than the grid can supply it. Behind-the-meter natural gas generation creates a direct bilateral relationship between energy producer and power consumer — bypassing the utility queue entirely.

The most valuable infrastructure contract available to a natural gas company right now is not a pipeline service agreement or a gas supply deal with a utility. It is a long-term power purchase agreement with a hyperscale data center operator. The reason is simple: data center operators need firm, dispatchable power in quantities that the utility interconnection queue cannot deliver on their timelines. Behind-the-meter generation — power produced and consumed on the customer’s side of the utility meter — bypasses that queue entirely.

For energy companies with gas supply, land positions, and generation expertise, the behind-the-meter market is one of the most compelling opportunities in the sector.

The Data Center Power Demand Surge

The numbers behind data center power demand have moved from surprising to staggering in the space of three years. Training a single large language model can consume as much energy as 100 US homes use in a year. A data center running continuous inference workloads at scale consumes power that would have supplied a small city a decade ago. And the build rate is accelerating.

The International Energy Agency projects that data centers will consume approximately 1,000 TWh of electricity annually by 2026 — up from roughly 200 TWh in 2022. That is a five-fold increase in four years. No other sector in the US economy is growing its energy consumption at anything close to that rate.

The hyperscalers are not being shy about their plans. Microsoft has committed to spending over $80 billion on data center infrastructure in fiscal year 2025 alone. Google, Amazon, and Meta have made comparable commitments. The capital is committed. The buildings are being built. The compute racks are being ordered. What is not keeping pace is the power supply.

Why the Utility Queue Is Broken

Getting power to a large new load through the utility interconnection process has always taken time. It now takes enough time to be incompatible with hyperscaler development timelines. The PJM Interconnection, which covers 13 states in the Northeast and Mid-Atlantic — precisely the geography where data center demand is most intense — has over 300 GW of pending interconnection requests in its queue. The median wait time for projects entering the PJM queue since 2018 exceeds four years.

ERCOT, the Texas grid, has a queue exceeding 226 GW of pending projects. California’s CAISO and the Midwest’s MISO are similarly constrained. The interconnection queue problem is not a regional anomaly — it is a national structural failure.

The core issue is that utility interconnection was designed for a world where new large loads connected infrequently, at measured pace, with years of lead time. A hyperscaler making a multi-billion dollar capital commitment to a data center campus needs a different kind of answer. “Your power will be available in 2029, pending interconnection studies” is not a business plan. It is a reason to find an alternative.

Behind-the-meter generation is that alternative.

BTM Generation Mechanics

Behind-the-meter generation means the generator is physically located on the customer’s side of the utility meter. Power flows from the generator directly to the data center’s load without crossing the utility grid. In most configurations, this means no interconnection agreement is required for the generator itself — it is simply a private energy system serving a private load.

The natural gas arrives via pipeline — a well-established, permitted, and reliable delivery mechanism. The generator converts that gas to electricity on-site. The electricity serves the data center’s compute, cooling, and facilities load directly. The utility connection, if one exists, serves as backup rather than primary supply. In some configurations, there is no utility connection at all for the primary power supply.

The data center operator and the energy producer enter into a long-term power purchase agreement that specifies price, volume, term, and reliability standards. Because this is a bilateral commercial contract — not a utility tariff — the terms can be negotiated to reflect the specific needs of both parties. The energy producer receives a long-duration, creditworthy offtake contract. The data center operator receives firm, dispatchable power without a utility queue timeline.

This structure has precedent in industrial applications — refineries, chemical plants, and steel mills have operated BTM generation for decades. The innovation is applying it to data centers at a scale and pace that was not contemplated in prior energy planning frameworks.

Risk Structure and Contract Terms

The quality of a BTM power investment is almost entirely determined by the quality of the offtake contract. The underlying generator is relatively standardized technology; the pipeline gas supply is a commodity market. The contract is the asset.

Key terms to scrutinize in any BTM power purchase agreement:

Contract duration. Minimum 10-15 years is the institutional standard for financing a BTM generation investment. Below that threshold, the asset’s useful life materially exceeds the contracted revenue certainty, which complicates debt financing and increases equity return requirements.

Pricing structure. Indexed pricing — where the power price adjusts against a benchmark such as CPI or a regional power price index — provides inflation protection for the energy producer and price certainty for the data center operator. A fixed nominal price over 15 years is disadvantageous for the producer in an inflationary environment.

Capacity payments. The contract should include both a fixed capacity payment (available whether or not the data center is using power, covering the fixed cost of the generator and fuel supply) and a variable energy payment (covering fuel costs on actual generation). This mirrors the capacity-plus-energy structure of utility power markets and provides revenue floor certainty.

Fuel cost pass-through. Gas price risk should be passed through to the offtaker, not borne by the generator owner. A contract that includes a fuel cost pass-through provision insulates the energy producer from commodity price swings while preserving the data center operator’s ability to hedge their own fuel costs in the gas market.

O&M responsibility. Long-term generator maintenance — major overhauls, turbine hot section replacements — should be addressed explicitly. Contracts that place all O&M risk on the generator owner create operating cost uncertainty that is difficult to underwrite over a 15-year horizon.

Curtailment rights. The data center operator will want the right to curtail generation during periods of low compute demand or planned maintenance. Curtailment provisions should include minimum capacity payments that prevent revenue from going to zero during curtailment periods.

How Sourcerock Approaches BTM Partnerships

The BTM opportunity requires integrating capabilities that most companies have only part of. A pipeline company has gas supply and delivery expertise, but not site control or generation knowledge. A real estate company has land, but not gas supply or power expertise. A generation developer has engineering capabilities, but not upstream gas relationships.

Sourcerock approaches BTM as an integrated package: gas supply from owned mineral rights or pipeline-connected supply, a land program that identifies sites appropriate for BTM development (proximity to existing gas transmission, adequate power load density in adjacent development areas, favorable permitting jurisdictions), and generation partnerships with established turbine operators.

Data center operators bring the compute and the offtake commitment. What they need from an energy partner is reliability, speed, and integration — a single counterparty who can move from term sheet to power delivery without the fragmented coordination challenges that plague multi-party project development.

The timeline advantage of BTM over utility interconnection is, in many markets, three to five years. In a sector where hyperscalers are competing on compute availability, that timeline difference is the entire business case.


Read next: GPU Pods and Distributed Power · Why Natural Gas Infrastructure Wins · Our BTM Power Program

Exploring a BTM power partnership? Contact Ryan