Solmar Insights
Major US data center developers are moving to behind-the-meter natural gas generation to meet AI-driven power demand, shifting procurement from delivered electricity to direct fuel sourcing. With large campuses now consuming enough gas to rival utilities and LNG facilities, competition for pipeline access and firm supply is intensifying across the energy infrastructure sector.
Key figures
1 GW data centers can consume 140 MMcf/d of natural gas
Entergy’s Meta agreement: seven new combined-cycle gas plants, 5,200 MW
FERC launches targeted effort for large-load integration
Behind-the-meter generation demand
To avoid the delays of grid interconnection queues and keep up with the rapid buildout schedules required for AI infrastructure, hyperscale data center operators are deploying onsite natural gas power generation. This move allows developers to bypass slow utility processes and take direct control of their power supply chains, pivoting from buying electricity as a finished product to sourcing the fuel required for their own generators.
This strategic pivot places new pressures both on energy supply chains and grid operations. No longer reliant solely on utilities, data centers entering direct fuel markets must grapple with all aspects of energy logistics, including securing pipeline capacity, arranging storage, and managing backup supply. The transition fundamentally changes risk allocation: developers now shoulder responsibility for fuel procurement, dispatch coordination, and associated infrastructure investment.
The significance of this shift is underscored by the volumes involved. Enverus analysis indicates that a 1 GW data center can require 140 million cubic feet per day of natural gas, moving individual projects into the category of major industrial consumers. As AI and high-performance compute workloads continue to scale, so too does aggregate gas demand from the digital sector.
For the largest developers, onsite gas-fired generation represents the most commercially feasible path to obtaining flexible, dispatchable power at the scale and speed their operations require, even as it brings new complexity to project execution.
Fuel supply rivalry intensifies
As data centers transition into the fuel procurement business, they are now in direct competition with entrenched buyers such as electric utilities, industrial plants, and LNG exporters for pipeline space and firm natural gas supply. Where once digital infrastructure consumed electricity delivered over regional grids, operators now seek strategic positions adjacent to midstream networks and secure access to critical infrastructure.
The infrastructure burden is notable: at load levels resembling those of city-scale utility customers, hyperscale data centers face a need for reliable, high-volume pipeline transport, compression, and sometimes even storage assets. This new class of direct fuel buyer is redefining traditional roles, moving data centers upstream in the hierarchy of North American gas demand.
Securing firm transportation rights and negotiating long-term gas supply agreements has become a central feature of large-scale data center project development. In practice, this means competing with longstanding utility and industrial customers, as well as with the rising volume of LNG projects seeking access to constrained pipeline corridors.
Market participants must now consider not just delivered electricity pricing, but the dynamics of pipeline congestion, regional basis differentials, and upstream supply risk, all of which increasingly impact the true cost and reliability of digital infrastructure power.
Project models in the field
Governed by urgency and driven by the scale of AI demand, prominent new projects illustrate the new paradigm of data center-fuel integration. Entergy’s agreement to support Meta’s data center expansion in Louisiana involves building seven gas-fired combined-cycle power plants totaling over 5,200 MW. This involves not just generating capacity but also considerable new gas infrastructure and transmission assets tailored specifically to digital loads.
Similarly, the Socrates Power Solution Facilities in Ohio, developed by Williams, target the growing data center segment with 400 MW of new gas-fired power. In both instances, traditional boundaries between digital infrastructure, generation, and midstream assets are dissolving, as projects are planned around the simultaneous coordination of fuel availability, power plant construction, and campus siting.
These integrated approaches align the needs of hyperscale customers, fuel providers, and electric grid operators, but further entangle the stakeholders in energy and infrastructure sectors. The complexity of these arrangements requires participants to consider fuel logistics and pricing well in advance of construction, often reshaping where and how new digital campuses are developed.
By integrating planning for energy supply, generation, and digital capacity, developers seek to reduce timelines, de-risk capacity additions, and deliver on the enormous and often inflexible compute requirements of their clients.
Regulatory and grid response
The shift toward direct fuel sourcing by hyperscale data centers is prompting regulatory agencies to reconsider long-standing power market planning structures. The Federal Energy Regulatory Commission (FERC) has initiated targeted efforts to speed the integration of large loads, recognizing that traditional utility planning and interconnection tariffs are outpaced by emerging demand from the digital sector.
This new planning challenge for regulators stems from a market in which the largest new loads are no longer standard customers waiting on utility-delivered power, but sophisticated buyers developing projects around direct fuel access and onsite generation. For FERC and regional transmission operators, accommodating these loads requires adjustments to grid planning models, infrastructure investment strategies, and reliability assessments.
Transmission and pipeline planning are increasingly interdependent, as the energy requirements of data centers force a reevaluation of where new assets are most needed and who should shoulder the costs. With data centers dictating the pace and geography of new infrastructure, regulators face growing pressure to adapt both process and policy to the fast-moving demands of digital infrastructure.
Policy outcomes from these efforts will influence the economics and feasibility of large digital projects nationwide, affecting risk allocation, fuel diversity options, and the stability of both electricity and pipeline networks.
Strategic value of pipeline assets
The surging demand for direct gas generation by data centers is affecting how midstream infrastructure is valued by both developers and investors. Traditional metrics for pipeline assets focused on throughput and utilization rates, but proximity to strategic data center locations is adding a new premium.
Pipelines that connect near major digital campuses or can support large quantities of behind-the-meter generation are becoming more valuable, influencing both acquisition activity and the strategic planning of midstream companies. Asset owners must assess not only historical customer bases but also the prospect of material new digital loads.
This evolving demand is likely to reshape capital allocation for pipeline expansion and upgrades, especially in regions where both data center and LNG facility development are accelerating. For developers and investors, understanding the geographic and contractual dynamics of data center gas load is now a material factor in asset valuation and project finance.
Energy recovery technologies and improved fuel efficiency are also being explored by developers as ways to maximize output from existing gas infrastructure, potentially mitigating some incremental demand without increasing total system supply requirements.
What this means for buyers
Buyers and investors should expect mounting competition for firm gas supply and pipeline access, especially near major digital campuses. The importance of securing reliable upstream fuel arrangements now sits alongside power procurement in data center site selection and risk assessment. Midstream asset valuations are likely to reflect both legacy demand and the strategic proximity to AI infrastructure. Stakeholders planning digital infrastructure need to understand regulatory shifts and evolving supply chain complexity as hyperscale projects increasingly depend on direct fuel integration.
