As the AI infrastructure race intensifies, Elon Musk’s xAI is scaling up its physical footprint by acquiring another large building, known as MACROHARDRR, near Memphis to expand its flagship Colossus data center project. The acquisition marks xAI’s third major facility in the region, strengthening its plan to build a tightly connected AI supercomputing campus capable of supporting massive model-training workloads. With this expansion, the company is pushing toward a total power capacity of about 2 gigawatts, positioning the Colossus project among the most energy-intensive and ambitious AI infrastructure efforts currently underway.
The new building is located in Southaven, Mississippi, just across the state line from Memphis, and sits close to xAI’s existing and planned Colossus sites. Its proximity is intentional, since placing several facilities close to one another allows xAI to link them through ultra-fast networking and operate them as a single, unified computing environment. This setup is crucial for training large AI models, where performance increasingly hinges on how efficiently tens or even hundreds of thousands of processors can work together in coordination rather than as separate, disconnected systems.
Notably, Colossus itself first came into focus in 2024, when xAI converted a former industrial facility in Memphis into a massive AI training center. That site was rapidly equipped with tens of thousands of GPUs, making it one of the largest AI compute clusters in operation at the time. Since then, the firm has signaled its intention to expand far beyond the initial build, with long-term plans that could eventually involve hundreds of thousands to around one million GPUs spread across multiple buildings in the region. And now, the newly acquired MACROHARDRR facility is expected to complement both the original Colossus site and Colossus 2, a second data center already under development nearby. Together, these locations form the foundation of a multi-building AI campus designed for continuous expansion.
Meanwhile, power and energy infrastructure are central to the project’s scale. A total capacity of around 2 gigawatts places Colossus in a class typically associated with heavy industrial operations rather than conventional data centers. Supplying that level of electricity requires dedicated power arrangements, including new generation capacity and specialized cooling systems to manage the heat produced by densely packed AI hardware.
The move comes as the AI industry more broadly races to secure its own computing infrastructure instead of relying solely on shared cloud providers. Major players are investing tens of billions of dollars in data centers to support increasingly large and complex models. But at the same time, the rapid expansion of these facilities is drawing scrutiny over their impact on local power grids, water usage for cooling, and long-term environmental sustainability.