Kyle Reidhead, an investment analyst at Milk Road, stated on October 15th (Beijing time) that SpaceX possesses an advantage in the AI computing power market that is difficult for emerging cloud service providers like CoreWeave and Nebius to replicate: its GPUs can be used for model training and inference within its xAI platform, or leased to external customers. This flexibility means that when customer demand is insufficient, SpaceX can convert its computing power for internal use, thus enabling it to sign shorter-term, higher-rent contracts, while other cloud service providers rely more on long-term lease agreements and debt financing. Simultaneously, SpaceX can leverage cash flow from its Starlink and rocket launch businesses to support computing power expansion. Reidhead cited a Bloomberg Intelligence model, stating that SpaceX's return on invested capital for its AI business is projected to rise from -4% in 2026 to approximately 25% in 2030, surpassing AWS and Google Cloud. He believes that this difference in business and financing structure could give SpaceX an advantage in the AI computing power competition.
