Tell us the node type, GPU count, topology, and target date. We'll return a capacity quote with the hardware and facility assumptions spelled out.