The conversation about powering AI data centers usually begins with what needs to be built: new generation, new transmission lines and new substations. Those investments are necessary, but they overlook a different question. How much usable energy is already moving through the infrastructure we have?

Consider the natural gas pipeline network. It was designed to transport fuel, not supply electricity. Yet at certain points along that network, gas arrives at a higher pressure than downstream users require. Operators reduce that pressure every day to deliver gas safely. In conventional pressure-regulating equipment, that pressure energy is dissipated without producing useful power.

As data center developers search for power that can be delivered sooner, that overlooked resource deserves a closer look.

The Power Hidden in Pressure Reduction

Natural gas moves through long-distance pipelines at high pressure. Before it enters lower-pressure distribution systems or serves industrial customers, that pressure must be reduced to the required operating level. Traditionally, regulators or control valves accomplish this by throttling the gas.

Pressure energy recovery offers another option. A turboexpander can use a portion of the pressure drop to drive a generator, producing electricity while the gas continues downstream at the required pressure. Unlike a gas-fired generator, the process produces electricity without additional fuel input or combustion.

That distinction matters, but so do the limitations. Recoverable output depends on gas flow, pressure differential and operating hours. The gas also cools as it expands, creating a low-temperature stream that may require thermal management or, where site conditions allow, can be used for cooling. Any assessment must account for auxiliary energy use, equipment efficiency, pressure control and pipeline operating constraints.

Pressure energy recovery is not a standalone answer to the power requirements of a hyperscale data center. At suitable locations, however, it can provide local electricity from energy that would otherwise go untapped. Where the low-temperature stream can be integrated into the cooling system, the same process can also reduce mechanical cooling demand. The value of each benefit depends on site scale, gas conditions and the ability to use the electrical and thermal output.

From Fuel Corridor to Energy Corridor

The broader opportunity extends beyond a single technology. Gas pipelines already connect production regions, industrial centers and many areas where new data center capacity is being considered. Existing compressor stations, pressure-reduction facilities and pipeline interconnections may offer starting points for evaluating distributed energy projects.

A pipeline-adjacent site could combine multiple resources: electricity recovered from pressure reduction and, where site conditions allow, usable cooling, gas-fueled generation where appropriate, battery storage and a conventional utility connection. Each performs a different role. Generation supplies substantial dispatchable capacity; storage helps manage short-term fluctuations; recovered energy improves use of an existing process; and the grid remains essential for large-scale service and long-term growth.

This is not a case for putting data centers beside every pipeline. Some locations will lack the necessary land, fiber connections, cooling resources or electrical infrastructure. Others will have insufficient gas throughput or limited opportunities to use or export recovered electricity. Pipeline proximity is useful only when the entire site can be engineered and permitted to function as an energy system.

The point is to evaluate these assets together, rather than treating the pipeline solely as a fuel-delivery route.

A Practical Addition to the Power Strategy

The timing makes that evaluation increasingly important. The International Energy Agency has warned that grid constraints could delay around one-fifth of planned data center capacity worldwide by 2030. Major transmission projects, transformers and generating equipment also face long development or delivery timelines.

Distributed energy cannot eliminate those bottlenecks. Installing a pressure-recovery system does not automatically create a direct electrical connection to a nearby data center. The project still needs power conditioning, protection equipment, commercial arrangements and, where applicable, utility approvals. Infrastructure may be close geographically but far away in development terms.

That is why technical screening should begin early. Developers and midstream operators need to examine hourly and seasonal flow profiles, pressure requirements, available generation output, potential cooling value and potential customers. They should assess whether electricity can be consumed behind the meter, supplied through an approved local arrangement or exported to the grid, and whether the low-temperature output can be used on site. They also need to account for maintenance, safety, redundancy and the operating priority of delivering gas reliably.

The strongest projects will be those where energy recovery fits naturally into infrastructure that is already performing its primary job, not those that force gas operations to accommodate an attractive power forecast.

Rethinking Existing Infrastructure

Meeting AI's electricity needs will require new generation and expanded grids. It will also require more productive use of existing assets. Pressure recovery will not deliver hundreds of megawatts from a single station, and it should not be presented as a shortcut around essential grid investments.

But the data center power crunch is too large to solve with one technology or one development model. Smaller, site-specific resources can complement larger investments, particularly when they can be identified early and integrated into a phased power strategy.

For decades, the energy industry has measured pipeline infrastructure by how efficiently it transports natural gas. The next question is whether some of those same assets can also contribute electricity and, in suitable applications, cooling where they are needed. The fuel network may have more to offer than fuel alone.