Amazon is constructing a natural gas power plant in Pecos County, Texas, that would become the single largest source of carbon pollution in the United States. The facility, designed to power a new AI data center, would generate up to 7.65 gigawatts of electricity using 35 natural gas turbines and release 33 million tons of carbon dioxide annually, according to project permits reviewed by the New York Times.

This development presents a stark contradiction. Amazon co-founded The Climate Pledge in 2019, a voluntary coalition of over 700 companies committed to achieving net-zero carbon emissions by 2040. Yet the company's emissions have risen each year for several years, driven largely by the explosive growth of data centers needed to support artificial intelligence applications.

Why Is AI Driving Such Massive Energy Demands?

The surge in AI infrastructure is reshaping global electricity consumption at an unprecedented pace. Global electricity use by data centers rose 17 percent in 2025, but consumption by AI-focused data centers surged approximately 50 percent in the same period. The International Energy Agency projects that total data-center electricity consumption could reach roughly 945 terawatt-hours by 2030, more than double its recent level.

Data centers arrive as concentrated loads requiring large blocks of power and rapid interconnection to the grid. This creates a timing problem: setting up new power infrastructure to connect power-hungry data centers to existing electrical grids takes years. To bypass these delays, hyperscalers and data center operators are increasingly turning to on-site power generation, a strategy known as "behind-the-meter" power.

Natural gas has become the preferred fuel for these dedicated power plants because it offers scalability and speed of availability compared to renewable energy sources. The Trump administration has further accelerated this trend by promoting oil, natural gas, and coal over renewable alternatives for data center projects.

What Are the Physical Limits of AI's Power Hunger?

Beyond the carbon question, AI infrastructure is colliding with hard physical constraints. During the June 2026 European heat wave, French nuclear power output fell by approximately 4.1 gigawatts as several reactors reduced production in response to high river temperatures and environmental restrictions. Further reductions occurred during the July heat wave.

This reveals a brutal paradox: the hotter the air becomes, the more electricity people need for cooling. Yet the same heat warms rivers, lowers water levels, and impairs the heat-rejection systems that nuclear and other thermal power plants depend on. Nuclear plants must reject waste heat through cooling water, and when that water becomes too warm, plant efficiency declines and environmental protections kick in, forcing operators to reduce output precisely when demand peaks.

Electricity planning has traditionally treated heat waves mainly as demand events. They must now be understood as simultaneous demand-and-supply crises. Extreme heat raises air-conditioning loads while affecting thermal generation, hydropower, transmission equipment, and sometimes wind and solar performance. The International Energy Agency reports that air conditioning and data centers are among the structural forces accelerating electricity demand, yet grid investment has lagged generation investment in many systems.

How Are Governments and Companies Responding to Grid Stress?

The strain on electricity infrastructure has forced grid operators to take extraordinary measures. In the United States, emergency orders have authorized grid operators to call on backup generation at data centers and other large facilities when the public system approaches its most serious emergency conditions.

Meanwhile, some technology companies are pursuing alternative power sources. Microsoft has secured a deal to reopen the Three Mile Island nuclear facility, and Meta has agreed to secure up to 1 gigawatt of orbital solar energy capacity. However, these initiatives remain exceptions rather than the norm.

Amazon acknowledged the tension between its AI ambitions and climate commitments. "The world looks different now than when we co-founded the climate pledge," stated Margaret Callahan, an Amazon spokeswoman. Still, "our commitment hasn't changed," she said, echoing similar statements by Microsoft, which is also struggling to fulfill its 2030 sustainability promise amid carbon-heavy AI expansion.

Amazon

Steps to Understanding AI's Energy Challenge

• Recognize the scale mismatch: AI data centers require gigawatts of continuous power, but grid infrastructure expansion takes years, forcing companies to build dedicated power plants on-site.
• Understand the carbon trade-off: Natural gas plants offer speed and scalability but generate massive carbon emissions, creating a direct conflict with net-zero pledges made by major tech companies.
• Consider climate feedback loops: Extreme heat increases cooling demand while simultaneously reducing the efficiency of thermal power plants, creating a vicious cycle during heat waves.
• Evaluate alternative solutions: Nuclear power and renewable energy offer lower-carbon options but face their own constraints, including cooling limitations and slower deployment timelines.

The Amazon Texas project illustrates a fundamental tension in the AI era. Technology companies have made ambitious climate commitments, yet the infrastructure required to power artificial intelligence at scale often conflicts with those goals. As AI continues to expand, the decisions made about how to power these systems will shape global emissions trajectories for decades to come.