Amazon’s West Texas Data Center Is Tied to a Gas Plant That Could Rank Among America’s Dirtiest

Updated August 8, 2026 9:54 pm

Amazon has helped bankroll a new natural-gas power project in West Texas that has been permitted to emit as much as 33 million tons of carbon dioxide a year, a scale that could make it one of the most polluting single power facilities in the United States. The plant is intended to supply electricity to Amazon’s expanding data center campus in Pecos County, underscoring how the AI boom is pushing major tech companies deeper into fossil-fuel infrastructure.

The project, known as GW Ranch, reflects a broader shift in the technology industry as cloud and AI workloads drive electricity demand beyond what many existing grids can easily support. Amazon has confirmed it bought the site and plans to purchase power from the facility, which is being built around 35 natural-gas turbines and, at least initially, will not rely on the Texas grid to move electricity to the data center.

The arrangement is drawing attention because it collides with Amazon’s publicly stated climate goals. Founder Jeff Bezos has said the company aims to reach carbon neutrality by 2040, but Amazon’s emissions have risen in recent years, in part because of the energy requirements of artificial intelligence and the infrastructure that supports it.

What Amazon is building in Pecos County

Amazon’s latest infrastructure project in Texas pairs a large data center with a custom-built power plant rather than depending solely on public utility supply. The setup is designed to provide roughly 7.65 gigawatts of electricity, enough to support a massive computing campus and future AI-heavy operations.

According to reporting cited by The New York Times and data tracked by Cleanview, the GW Ranch site received permission from Texas regulators to emit up to 33 million tons of carbon dioxide. That figure is a permit cap rather than a guaranteed annual output, but it places the project in rare company among U.S. energy facilities.

Why the permit matters

The permit matters because it indicates the scale of pollution regulators are prepared to allow, and because it signals how aggressively large technology companies are securing energy for AI expansion. Even if the plant never reaches its maximum allowed emissions, the ceiling alone suggests a level of fossil-fuel use that would have been difficult to imagine in the early years of cloud computing.

Natural-gas generation is often presented as a cleaner alternative to coal, but a plant of this size would still create substantial greenhouse-gas emissions and local air pollution. For a company that has publicly framed itself as a climate leader, the optics are especially significant.

How the plant is supposed to work

The GW Ranch project is structured to serve the data center directly. Rather than feeding power first into the statewide grid, the plan calls for gas turbines on-site or nearby to produce electricity primarily for Amazon’s computing facilities. That design can improve reliability and speed to power for a hungry campus, but it also reduces the role of public-grid oversight and shifts more responsibility onto a privately financed energy system.

In practical terms, the project appears to be part power plant, part digital infrastructure strategy. It allows Amazon to lock in large volumes of electricity for data processing without waiting for grid upgrades or competing with residential and commercial users for capacity. That flexibility is increasingly important as AI training and inference consume more power than older forms of cloud computing.

Why Amazon’s emissions problem is getting harder to hide

Amazon has spent years promoting ambitious climate targets, but the rapid growth of AI has made those promises more difficult to square with its infrastructure choices. Data centers already require enormous amounts of electricity and cooling, and AI can significantly increase both demand and volatility.

Amazon’s own environmental reporting has shown rising emissions over multiple years. The company and its peers have been forced to confront a reality in which more computing capacity often means more energy use, more hardware, and more pressure to secure power wherever it can be found.

Amazon spokesperson Margaret Callahan told The New York Times that the company’s climate pledge was created for a different world than the one it operates in now, while emphasizing that the company says its overall commitment remains in place.

That response captures the core tension: the company is still publicly committed to long-term decarbonization, yet its current growth strategy increasingly depends on the same fossil fuels that climate plans are meant to phase out.

Why are Big Tech companies turning to their own power plants?

Big Tech companies are turning to their own power plants because AI has outgrown the pace of many existing energy systems. The quickest way to guarantee enough electricity for a new campus is often to build or finance dedicated generation, including gas-fired plants that can deliver large volumes of reliable power on a schedule that grid expansion cannot always match.

That strategy is not unique to Amazon. Meta and Google have both moved toward greater direct involvement in power infrastructure, including fossil-fuel projects, as data-center demand accelerates. The move has sparked concern among climate advocates and regulators who worry that the AI race is effectively locking in new emissions for decades.

• Data centers need continuous, high-density power.
• AI workloads can sharply increase electricity demand.
• Private generation can be faster to deploy than grid upgrades.
• Natural gas remains a common fallback because it is scalable and dispatchable.

How bad could the pollution be?

The short answer is that the facility could be extraordinarily dirty, even if it falls well below its legal maximum. A permit to emit 33 million tons of carbon dioxide a year is striking because it exceeds the scale of even the nation’s biggest coal facilities, which are already among the most carbon-intensive power sources in the country.

Environmental trackers note that permitted emissions are not the same as actual emissions. Plants rarely run at their absolute maximum all year, and operating conditions vary. Still, the limit itself reveals the industrial ambition of the project and the regulatory environment that made it possible.

For context, a power plant with emissions in the tens of millions of tons would be operating at a level that can materially affect regional climate totals, local air quality, and the reputation of the company tied to it. The project also raises questions about whether the AI sector is shifting emissions rather than reducing them.

Comparison with other power sources

Natural gas is usually cleaner than coal on a per-unit-of-electricity basis, but that does not make it clean. When scaled up to serve a giant data center, gas can still become a major source of climate pollution. The issue is not only the fuel type but the sheer size of the project and the cumulative impact of adding more fossil infrastructure to meet digital demand.

That is why the GW Ranch permit has become a flashpoint. It may not be the largest plant by physical footprint, but its emissions cap places it in a category that climate policy experts and investors are likely to watch closely.

How this fits into the AI energy race

The Amazon project is part of a wider AI energy race that is reshaping power markets in the United States. As companies compete to build larger models, serve more users, and expand their cloud offerings, the demand for fast, dependable electricity has become a competitive advantage.

That reality is changing how companies think about infrastructure. Instead of waiting for utilities to modernize the grid, some are opting for vertically integrated solutions: acquire land, secure permits, build generation, and connect it directly to a data center campus. The result is a hybrid of tech campus and industrial power station.

This is also why policymakers are paying closer attention. If AI growth is met largely with gas-fired generation, then the industry’s carbon footprint could rise faster than efficiency improvements can offset it. That concern is heightened by the Trump administration’s efforts to loosen restrictions on polluting power plants, which could make it easier for similar projects to move forward.

What Amazon has said about climate and growth

Amazon has not abandoned its climate goals, but the company’s messaging has become more defensive as energy demand has grown. The core claim is that the long-term commitment still stands even if the operating environment has changed dramatically since the company first embraced the Climate Pledge.

That explanation may satisfy investors focused on execution, but it does little to address the underlying problem: Amazon is expanding AI infrastructure at a speed that appears to be outpacing its clean-energy transition. The company, like many others, is trying to reconcile two competing imperatives — build faster and decarbonize later.

In the company’s view, the world has shifted since the climate pledge was first announced, but Amazon says its target has not changed.

That distinction matters because climate targets are only as credible as the infrastructure choices behind them. If the company’s new capacity is powered by a plant that may emit tens of millions of tons of carbon dioxide, the pledge becomes harder to defend as more than a distant aspiration.

What happens next?

The immediate next step is construction and operational rollout, followed by scrutiny from climate groups, local stakeholders, and investors who are increasingly attentive to the environmental cost of AI. If the plant begins operating at scale, its emissions will likely become a benchmark for future debates over data-center power policy.

More broadly, the project may influence how other tech companies plan their next generation of campuses. If Amazon’s approach proves effective, rivals could copy it. If it becomes a public-relations liability or draws stronger regulatory pushback, it could accelerate interest in renewables, storage, nuclear, or other lower-carbon options that can support always-on computing loads.

What to watch

1. Whether the facility stays within its permitted emissions range.
2. How quickly the data center comes online and at what capacity.
3. Whether Amazon adds renewable energy or storage to offset some of the site’s footprint.
4. Whether state and federal regulators tighten scrutiny of AI-linked fossil projects.
5. Whether other major tech firms pursue similar private power arrangements.

Timeline of the Amazon West Texas project

The project has developed as part of Amazon’s broader push to secure power for its growing AI and cloud infrastructure footprint. The timeline below summarizes the key milestones reported so far.

Why this story matters beyond Amazon

This is about more than one company or one power plant. It is a window into the future of digital infrastructure, where the appetite for AI may increasingly collide with climate policy, local air quality, and the economics of energy production.

If Amazon can justify a project of this size for a data center, it may help normalize a new model in which the biggest technology companies behave less like software firms and more like industrial utilities. That shift could reshape emissions trends across the tech sector for years.

For consumers, the story is a reminder that the conveniences of generative AI, cloud computing, and always-on digital services rest on a physical foundation of land, turbines, fuel, and water. The cloud, in other words, is getting a lot less invisible.

And for Amazon, the challenge is especially acute. The company built a reputation on logistics, scale, and relentless expansion. Now it is discovering that those same strengths come with a visible environmental cost that is harder to explain when the power source is a gas plant capable of rivaling the dirtiest in the country.