The Return of Fossil Fuels: A Decade of Shift in US Energy Infrastructure

The US energy mix is undergoing a significant shift. A newly released industry report shows that the scale of fossil fuel infrastructure under construction in the United States has reached its highest level in nearly a decade, while the growth rate of renewable energy has slowed markedly. This change is not an isolated phenomenon but is tightly intertwined with accelerating electricity demand—data centers, factories, and electric vehicles are connecting to the grid at the fastest pace in more than a decade.

Demand-Side Pressure: Who Is Driving Up Electricity Consumption

The electricity demand growth curve is becoming steeper. For more than a decade, US electricity consumption remained flat or even stagnant, as efficiency gains and industrial offshoring offset the incremental demand from economic growth. But that balance is being disrupted.

Data centers are the most closely watched variable. The deployment of large-scale AI training and inference clusters has pushed the power consumption of individual facilities to several times that of traditional data centers. At the same time, factory construction driven by manufacturing reshoring policies and rising EV penetration together form a triple push on the demand side. The report explicitly states that electricity demand is growing at the fastest rate in more than a decade.

This demand-side shift directly changes how utilities and policymakers assess supply reliability. As grid margins narrow, the urgency of building new dispatchable generation outweighs considerations about the pace of low-carbon transition.

Dispatchable generation and baseload are key power system concepts for understanding this round of fossil fuel resurgence. Power systems require real-time balance between generation and consumption, and generation sources can be divided into two categories by output characteristics: weather-dependent wind and solar are "variable sources" whose output cannot be controlled; gas, coal, nuclear, and hydro can increase or decrease output according to dispatch instructions and are called "dispatchable sources." The portion of generation that meets the all-hours minimum electricity demand is called "baseload," typically handled by units with low operating costs and stable output. Energy storage (especially long-duration storage) could in theory "convert" variable sources into quasi-dispatchable resources, but its large-scale deployment will take time. Therefore, at a stage when the storage gap is significant, natural gas generation—thanks to flexible start-stop capability and relatively low unit cost—has become the most realistic option for filling baseload and peaking gaps. This is precisely the technical logic behind the infrastructure rebound described in the report.

Supply-Side Choices: Why Fossil Fuels Are Back in Focus

Demand pressure is only half the story. The rebound in fossil fuel infrastructure construction reflects the power system's repricing of "stability."

The cost advantages of wind and solar remain, but their output depends on weather conditions and requires supporting storage or backup generation to meet baseload demand. At a stage when storage scale-up is not yet fully in place, natural gas generation is seen as the most realistic option for filling the gap. The number of pipeline, LNG terminal, and gas turbine projects approved and breaking ground has therefore risen noticeably.

Another signal revealed by the report is the slowdown in renewable energy growth. This slowdown may stem from multiple factors: excessively long interconnection queue periods, rising supply chain costs, and adjustments in policy support in some states. When the pace of renewable project deployment cannot keep up with demand growth, the gap is naturally filled by fossil fuels.

It is worth distinguishing that the rebound in infrastructure construction and the long-term trajectory of carbon emissions are not the same thing. New gas plants often have design lifetimes of decades, meaning today's investment decisions will lock in the emissions structure for a considerable period ahead.

Interconnection queue is one of the core bottlenecks behind the slowdown in renewable energy growth. In the United States, connecting a new generation project to the transmission grid requires processes including interconnection application, system impact studies, and agreement signing. In recent years, due to surging application volumes, tight grid capacity, and limited approval resources, queue periods in many regions have lengthened to several years, creating a "projects waiting for the grid" backlog. At the same time, slow transmission line expansion has further exacerbated congestion. The longer the queue period, the greater the interest rate and supply chain cost risks developers face, and some projects have withdrawn as a result. This mechanism explains why, even though wind and solar have a cost advantage, their actual deployment pace may still fail to keep up with electricity demand growth, thereby ceding the gap to gas projects with shorter construction cycles and more flexible siting.

The Role of Data Centers: The Energy Bill of the AI Boom

Among the many demand sources, data centers best explain the suddenness of this round of change. The pace of AI compute expansion has exceeded the expectations of most grid planners, and interconnection applications in some regions are already queuing up.

This creates a structural contradiction: tech companies publicly commit to clean energy goals on the one hand, while on the other hand the grids their compute expansion depends on must rely more on fossil fuels in the short term to ensure supply. This tension is reflected in the report's data as the opposite trajectories of fossil fuel infrastructure and renewable growth.

On a longer timescale, uncertainty remains over whether data center electricity demand will continue growing at the current pace. Improvements in chip energy efficiency, model efficiency optimization, and the trend toward self-built generation could all change the eventual magnitude of electricity demand. But within the current investment cycle, the power system is making decisions based on the assumption that demand will continue to grow rapidly.

Behind-the-meter generation is an emerging strategy for tech companies to cope with interconnection bottlenecks, referring to users building their own generation facilities behind the meter to supply data centers directly, thereby bypassing the public grid's interconnection queue. Common forms include self-built gas turbines, fuel cells, and long-term direct supply agreements with nuclear or renewable projects. This trend has dual implications for the report's conclusions: on the one hand, it may cause some new electricity consumption to be excluded from public grid load statistics, thereby underestimating the actual scale of fossil fuel expansion; on the other hand, behind-the-meter generation also gives tech companies a more direct channel to procure clean electricity. Therefore, the prevalence of behind-the-meter generation is an important variable in determining which generation sources ultimately meet data center electricity demand.

What This Means for the Energy Transition

Interpreting this round of change simply as a "transition setback" is not accurate, but it does reveal the pacing problem of the energy transition. Decarbonization targets set the long-term direction, while the grid must meet present reliability requirements, and the time gap between the two is being filled by fossil fuel projects.

For observers, several indicators are worth tracking: the installed capacity and commissioning timelines of new gas plants, changes in renewable energy interconnection queue periods, and whether falling storage costs can reopen the economic gap with gas peaking. These variables will determine whether the current round of fossil fuel construction is a short-term stopgap or forms a new round of path dependence.

Infrastructure decisions in energy systems have long-tail effects. Projects breaking ground today will continue to affect electricity prices, emissions, and grid structure for the next decade. The trend presented in the report is essentially a question about "how to rank priorities when demand accelerates," and the answer is being written by specific projects one by one.

One-line Summary

US fossil fuel infrastructure hits a decade high as AI and electrification drive up electricity demand, while renewable growth slows.

Paragraph Summary

A significant shift is occurring in the US energy mix: fossil fuel infrastructure construction has reached its highest level in nearly a decade, while renewable energy growth has slowed in tandem. This change is closely tied to accelerating electricity demand, with data centers (especially AI compute clusters), factory construction from manufacturing reshoring, and rising EV penetration forming a triple push on the demand side. Because wind and solar output depends on weather and storage scale-up is not yet in place, natural gas generation has become the most realistic option for filling baseload and peaking gaps. The report also flags a risk: new gas plants have design lifetimes of decades, so today's investment decisions may lock in future emissions structure and create path dependence. The direction of the energy transition remains unchanged, but its pace is being reordered by real-world reliability needs.