SpaceX Targets Massive AI Expansion With 440,000 NVIDIA Vera Rubin Systems
During its third-quarter 2026 earnings call, Elon Musk outlined SpaceX’s ambitious strategy to expand its artificial intelligence computing infrastructure and strengthen Grok’s competitiveness.
The company aims to allocate approximately 50% of its computing capacity to internal operations and 50% to external customers, adjusting the balance according to demand. Rather than committing to long-term rental agreements, SpaceX prefers shorter-term contracts at prevailing market rates, anticipating that more capable AI models will increase the value of computing resources.
One of the most significant announcements was SpaceX’s plan to deploy a training cluster featuring 440,000 NVIDIA Vera Rubin computing systems by the end of Q2 2027. The planned infrastructure is intended to accelerate Grok’s development and improve its competitive position.
Musk also highlighted a growing challenge for the AI industry: expanding computing capacity increasingly depends on electricity supply, cooling systems and supporting infrastructure, rather than chips alone.
Why it matters: The AI race is no longer just about developing smarter models. Access to reliable power, large-scale computing infrastructure and efficient deployment could become equally important competitive advantages.
These are company targets and forward-looking plans, not guarantees of delivery or future performance.
