SpaceX Commits to Exclusively Using Nvidia for AI Computing
On August 4, SpaceX held its first earnings call as a publicly traded company and announced that it will exclusively use Nvidia's hardware for all future artificial intelligence (AI) computing needs. CEO Elon Musk cited the superiority of Nvidia’s Vera Rubin architecture and emphasized strong collaboration with the chipmaker. This decision has significant implications for both companies' financials, given SpaceX's massive capital expenditures on AI infrastructure. SpaceX spent $15.8 billion in Q2 2024 alone on AI computing, representing a dramatic increase from previous quarters. The company expects to spend around $37 billion more this year and aims to reach over 10 gigawatts of compute capacity by the end of 2027. Despite not having filed any formal contracts with Nvidia in its recent SEC filings, SpaceX has already secured large-scale GPU commitments for cloud services through customer agreements. The exclusivity deal raises concerns about pricing power and supply chain risk due to reliance on a single supplier. Additionally, SpaceX is selling compute capacity it hasn't yet acquired, potentially exposing the company to delivery risks if Nvidia fails to meet its obligations.
