The AI boom is creating a problem that has little to do with artificial intelligence itself: power.
Amazon is backing a massive data-center project in West Texas that will be supplied by its own 7.65-gigawatt natural-gas power plant. The project highlights just how much electricity is required to operate the infrastructure behind modern AI systems.
The facility is planned for Pecos County and is designed to operate alongside a large AI data-center campus. Instead of depending entirely on the existing electrical grid, the project is being developed with dedicated power generation for the data centers.
That approach illustrates the enormous energy requirements of modern AI infrastructure.
Training and running large AI models requires huge numbers of GPUs and specialized processors. These machines consume substantial amounts of electricity, while the facilities housing them also require cooling systems and other supporting infrastructure.
The scale of Amazon’s project is particularly striking. The planned power plant has a capacity of 7.65 gigawatts and is expected to use 35 gas turbines. Reports indicate that the facility could be authorized to emit around 33 million tons of carbon dioxide annually.
The project therefore raises an uncomfortable question for the technology industry: how much physical infrastructure will be required to keep the AI revolution running?
AI companies have spent years competing over increasingly powerful models, GPUs, and data. But the next major competition may be over something much more basic: electricity.
Data-center operators are already looking for new ways to secure enough power as AI workloads grow. Some companies are investing in nuclear energy, renewable generation, batteries, and dedicated power infrastructure. Amazon’s Texas project demonstrates another approach: building a large gas-powered facility specifically around the needs of an AI data center.
The development also exposes the environmental trade-offs behind AI.
Artificial intelligence is often presented as software that exists in the cloud, but that cloud is ultimately made of physical computers sitting inside enormous buildings. Those computers require electricity around the clock.
As AI models become larger and AI agents perform increasingly complex tasks, demand for computing power is expected to continue increasing.
The race to build better AI may therefore become a race to build more data centers and find enough energy to operate them.
Amazon’s Texas project is a striking example of how far that race is already going.
