“Tesla projects its AI training compute capacity will increase from its current 100,000 H100-equivalent GPUs to 275,000 by March 2026 with the launch of its Cortex-2 cluster.”