SpaceX Expands Colossus 2 Compute Campus

Colossus 2 shows that the AI race is increasingly a power, cooling, construction, and networking race—not merely a chip race.

• –A 400 MW expansion is infrastructure at hyperscale, aimed at supporting future Grok training and high-volume inference.
• –SpaceX’s vertically integrated approach could bring capacity online faster than conventional greenfield data-center projects.
• –More GPUs do not automatically produce better models; utilization, data quality, networking, and research execution remain decisive.
• –The campus could also become a strategic compute platform for outside AI companies, strengthening SpaceX’s infrastructure business.
• –Local permitting, emissions, grid stability, and community opposition may become the project’s biggest constraints.