[Musk: SpaceX Plans to Lease Out Approximately Half of Its AI Computing Power, Aiming to Build a 440,000-Device Training Cluster by Q2 Next Year] During SpaceX's Q3 earnings call, Musk stated that the company aims to allocate its AI computing power roughly half for internal use and half for external leasing, with the ratio dynamically adjusted based on demand. The company prefers signing short-term rental agreements at market prices rather than locking into long-term contracts. He believes that as model capabilities improve, the value of computing power per unit of energy consumption will increase, and the shortage of computing power will persist. Musk revealed that SpaceX plans to launch a training cluster consisting of 440,000 Nvidia Vera Rubin computing devices by the end of Q2 2027 to enhance Grok's competitiveness. He noted that the primary bottlenecks for computing power expansion have shifted more toward power supply, cooling, and infrastructure, rather than chips.