Elon Musk’s artificial intelligence venture xAI is taking an unprecedented step to secure the vast amount of electricity required for its next major data center. According to Dylan Patel from SemiAnalysis, who revealed details in a recent podcast interview, Musk’s company is acquiring a power plant overseas and intends to ship it to the United States. The news, later confirmed by Musk himself in a post on X, underscores the scale of xAI’s energy needs as it prepares to build a facility capable of running one million AI GPUs. As reported by Tom’s Hardware page, this new site could consume up to 1.96 gigawatts of power, roughly the equivalent energy demand of 1.9 million American homes.
The new facility will follow the model of xAI’s existing Colossus supercomputer near Memphis, Tennessee, which already ranks among the most powerful and energy-intensive systems in the world. That data center runs on 200,000 Nvidia Hopper GPUs and draws approximately 300 MW of electricity. To meet those needs, xAI installed 35 gas turbines and incorporated Tesla Megapacks to stabilize power delivery. Yet, powering the next-generation data center, expected to use Blackwell-series GPUs, will require more than four times the capacity of Colossus.
Tom’s Hardware analysis reveals that even under conservative estimates, just the GPUs alone would consume 1 to 1.4 gigawatts. But when factoring in CPUs, memory, networking infrastructure, storage, cooling, and electrical inefficiencies, the total demand rises to between 1.4 and 1.96 GW. While renewable options like solar were considered, the article notes these are largely impractical at such a scale and consistency. A more viable path involves combined-cycle gas turbine plants, exactly the kind of infrastructure Musk may be importing to speed up deployment and sidestep long U.S. regulatory timelines.
The decision to import an entire power plant is a bold move, but experts say it reflects a broader trend in the AI industry. Patel emphasized that leading AI players are gravitating toward centralized, ultra-high-performance compute clusters backed by massive power resources and elite talent. As AI models continue to grow in complexity and size, energy will be the new frontier in competitiveness. xAI’s aggressive energy acquisition strategy may set a new precedent in the arms race for AI infrastructure dominance.