Amazon is building a massive gas power plant in Pecos County, Texas, to power its new AI data center. The plant has received a permit to emit up to 33 million tons of CO2 – more than even the largest coal plant in the United States, according to The Verge. This reveals the infrastructure reality of AI scaling: tech giants are mobilizing enormous energy resources to keep their models running.
Key facts
- The facility is called GW Ranch and will be powered by 35 natural gas turbines generating 7.65 gigawatts of capacity
- The CO2 permit of 33 million tons exceeds all other individual US power plants
- The plant is not connected to the public grid, but feeds power directly into the Amazon data center
- Amazon has confirmed it purchased the site and will buy power from GW Ranch
Climate pledge under strain
The investment directly contradicts Amazon's Climate Pledge: Jeff Bezos committed in 2019 to make the company carbon-neutral by 2040. Yet Amazon's emissions have climbed for years due to AI demand. An Amazon spokesperson told the New York Times:
"The world looks different now than when we co-founded the climate pledge, but our commitment hasn't changed."
Even if GW Ranch operates at a fraction of its permitted capacity, it would significantly increase Amazon's total emissions. While power plants rarely emit at their maximum permitted levels, the fact that these limits are so generous reveals regulatory laxity.
The broader trend: Tech giants build their own power plants
Amazon is not alone. Meta and Google are also constructing their own power plants to supply their data centers – turning to gas, coal, and other fossil fuels. The reason: AI models consume staggering amounts of electricity. The Trump administration has further loosened restrictions on such facilities, accelerating the trend.
The Texas gas plant represents an extreme case: it is not connected to the public power grid, but rather a dedicated power plant for a single data center. This approach is economically and technically viable only at this scale.
What this means for European companies
For German mid-market firms and European tech companies, the message is clear: US giants are investing in massive infrastructure to scale AI models – and accepting substantial environmental costs to do so. This creates a competitive advantage through sheer scale and capital that European companies struggle to match. At the same time, it underscores that AI energy consumption remains a central cost and sustainability challenge – including for German companies building AI infrastructure. The question remains: how long can regulators and society tolerate this energy hunger?
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
