AI infrastructure spending is moving beyond chips and data centers into power generation itself.
The Decoder reports that Nvidia is investing up to $3 billion in Lancium, a power infrastructure developer with four gigawatts already under contract in Texas and plans for far more. The same report says Amazon is backing a large gas-fired power plant in Pecos County, Texas, to support data center demand.
The figures show how the AI buildout is changing the bottleneck. Model training and inference need GPUs, but those GPUs need reliable electricity, transmission, cooling, and sites that can be approved quickly. Power access is becoming a strategic asset for AI companies and their suppliers.
The climate implications are significant. The reported Amazon-backed plant could become one of the country’s largest greenhouse-gas sources if built as described. The next phase of AI competition will therefore be measured not only in model scores, but in energy choices and local infrastructure tradeoffs.