Elon Musk confirms he diverted 12,000 Nvidia H100 GPUs from Tesla to X

Elon Musk has confirmed that he redirected 12,000 Nvidia H100 GPUs intended for Tesla to his other venture, X, due to unavailability of operational space at Tesla's Texas factory, which is still under construction. This diversion is believed to set back Tesla’s progress in acquiring approximately $500 million worth of these high-demand GPUs, crucial for advancing its AI and robotics ambitions. This development was initially revealed in a CNBC report, which highlighted potential internal conflicts and Musk’s prioritization of X over Tesla regarding the allocation of technology resources.

According to Musk, the Texas factory’s south extension, which is nearing completion, is set to accommodate 50,000 H100 GPUs for Full Self-Driving (FSD) training. This configuration would significantly support Tesla's substantial investment in AI, which includes roughly $10 billion this year alone, with about half dedicated to internal projects like the Tesla-designed AI inference computer and the Dojo supercomputer. Moreover, the redirection was facilitated by an Nvidia memo which adjusted the original delivery schedule of Tesla's H100s to later dates in compensation for the diverted units.

Reflecting on the implications of this move, Tesla's AI capabilities might face delays, yet the company remains committed to a hefty AI-related expenditure and aims to lead in AI and robotics. As Tesla plans to ramp up its capabilities by the end of the year, the operational ability of the Giga Texas extension will be critical. Musk's strategic decision, while controversial, aligns with his broader vision for integrating advanced AI within Tesla’s ecosystem despite the immediate logistical hurdles.