Amazon's Texas Data Center Will Draw Power from America's Worst-Polluting Gas Plant

What happened

Amazon has secured a permit to draw power for its new GW Ranch data center in Pecos County, Texas from a natural-gas-fired power plant that Cleanview tracks as the worst-polluting facility of its kind in the United States. The company received approval to release up to 33 million tons of CO2 through this arrangement.

The gas-burning plant at GW Ranch contains 35 natural-gas turbines capable of delivering 7.65 gigawatts of electricity to Amazon's data center operations. According to Cleanview, which monitors data centers and their associated power projects, the emission permit for this facility exceeds even the largest coal plant in the country.

Amazon confirmed it purchased the site and plans to purchase power from GW Ranch. This development comes as the Trump administration has sought to lift restrictions on polluting power plants amid surging demand for electricity from AI data centers.

The broader trend shows that major tech companies including Meta and Google have begun building their own power plants, turning increasingly to gas, coal, and other non-renewable energy sources to meet their insatiable computing demands. Amazon's data centers alone consumed 2.5 billion gallons of water last year, highlighting the environmental footprint of AI infrastructure expansion.

Terrence O'Brien, The Verge's weekend editor who has covered the tech industry for over 18 years, reported on this story, bringing attention to how the AI boom is reshaping America's energy landscape in ways that may contradict climate goals.

Why it matters

This development signals a troubling trend: as artificial intelligence demands explode, tech giants are increasingly turning to fossil fuel infrastructure rather than renewable alternatives. The GW Ranch facility represents a significant commitment to dirty energy at a time when the industry claims to pursue sustainability.

The scale is staggering—33 million tons of CO2 emissions from a single data center's power supply. For context, that exceeds even the largest coal plant in America. This suggests that AI infrastructure growth may be outpacing the transition to cleaner energy sources, creating a paradox where "green" tech companies rely on some of the dirtiest power generation in the country.

The Trump administration's policy shift to relax restrictions on dirty power plants while AI data center construction accelerates creates a feedback loop: more computing demand leads to more fossil fuel infrastructure, which then enables even more computing expansion. This cycle may lock in decades of carbon-intensive energy use for the very technologies marketed as solutions to climate change.

The water usage figure—2.5 billion gallons annually from Amazon's data centers—adds another layer of environmental concern. Combined with heavy reliance on fossil fuels, this paints a picture of AI infrastructure that is both thirsty and dirty, requiring substantial resources while contributing significantly to greenhouse gas emissions.

What to watch

Several developments will determine whether this represents an isolated case or the beginning of a broader pattern:

• Other tech companies' power sourcing: Will Meta, Google, and others follow Amazon's lead in securing permits for polluting plants, or will they pursue cleaner alternatives?

• Policy responses: Can regulators reinstate restrictions on dirty power plants as AI infrastructure expands, or will the Trump administration's deregulatory approach continue?

• Renewable energy deployment: At what pace can renewable generation scale to meet data center demand without relying on fossil fuel infrastructure?

• Corporate sustainability commitments: How will Amazon reconcile this development with its stated climate goals and net-zero pledges?

• Water usage trends: Will water consumption from data centers accelerate alongside power demand, creating a dual-resource crisis?

The GW Ranch permit represents a critical inflection point. If major tech companies continue to secure permits for polluting plants as AI infrastructure scales, the environmental cost of the AI revolution may be far greater than currently acknowledged. Conversely, if this proves to be an anomaly that other companies avoid, it could signal that the industry is still finding ways to balance growth with sustainability.

The coming years will reveal whether America's data center boom can coexist with climate goals—or whether the two are fundamentally incompatible under current policy and technology trajectories.

By the numbers

• CO2 emission permit: 33 million tons
• Turbines: 35 natural-gas turbines
• Electricity capacity: 7.65 gigawatts
• Water usage (Amazon data centers): 2.5 billion gallons last year
• Publication date: August 8, 2026, 1:53 PM EDT
• Editor tenure: over 18 years

Source snapshot

Sources: - https://www.theverge.com/ai-artificial-intelligence/977124/amazon-data-center-worst-polluting-power-plant - https://www.theverge.com/ai-artificial-intelligence (homepage)