xAI Targets 1.5 GW Power for AI Supercomputer, Reportedly Deploys 200,000 H100 GPUs in Rapid Build-Out

Elon Musk's artificial intelligence venture, xAI, is reportedly undertaking an exceptionally rapid and large-scale build-out of compute infrastructure, aiming for a 1.5 gigawatt (GW) power capacity to fuel its advanced AI models. A recent social media post by Musk detailed ambitious plans, including the construction of a 1 million square foot GPU data center within six months and the deployment of 200,000 H100 GPUs, dubbed "Colossus 1," in just 122 days. This aggressive expansion underscores the intense global race for AI compute power.

The post highlighted xAI's unconventional approach to securing power, stating, "power? lol just buy a dead Duke Energy plant in Mississippi," and claiming the ability to "spin turbines without a permit" by relocating operations slightly to bypass regulations. Furthermore, xAI reportedly secured 600 megawatts (MW) of Solaris turbines, accounting for "67 % of Solaris order book," in a move described as "literally Airbnb for gas." This strategy suggests a focus on immediate and flexible power solutions.

Musk's post also detailed the substantial energy demands and solutions, noting the operation of "119 chillers humming at 200 MW" and running "1.1 GW across state lines on Tesla batteries lined up like Duracell bunnies on meth." The ultimate goal is to reach 1.5 GW, enabling the training of a model capable of drafting "OpenAI's bankruptcy filing in perfect limerick," a clear jab at a key competitor. This scale of deployment showcases the critical role of energy infrastructure in the AI arms race.

The narrative from Musk's post positions xAI as a hyper-agile player in contrast to rivals. It claimed that while "entire OpenAI campus still arguing over lunch menu," xAI "already poured 40k yards of concrete." Similarly, "Oracle still waiting for concrete to cure," while "Meta still drawing blueprints," and "Anthropic typing safety essays while my turbines outrun their ethics board." This framing emphasizes xAI's speed and execution over traditional development cycles.

Industry analysis confirms a fierce competition among AI companies like OpenAI, Meta, and Anthropic to acquire and deploy vast quantities of advanced GPUs and the necessary power infrastructure. Companies are investing billions in data centers and energy solutions to support the training of increasingly complex AI models. Tesla's Megapack systems are indeed utilized for large-scale energy storage and grid support, aligning with xAI's reported use of Tesla batteries for power distribution.