AI Data Centers · Tools
Interactive simulators for AI infrastructure
Calculators and comparators built for the questions operators actually ask: what does this rack draw, how cool can I keep it, which GPU per dollar wins. No login. Results stay client-side.
GPU Comparator
Side-by-side comparison of H100, H200, B100, B200, GB200, MI300X, MI325X, TPU v5/v6 — TFLOPS, HBM, NVLink, FP8/INT4, power, $/TFLOPS.
Open tool →OperatePUE Calculator
Compute Power Usage Effectiveness from IT load, cooling overhead, and ambient temperature. Compare air vs. liquid vs. immersion cooling.
Open tool →DesignPower Budget Planner
Plan a campus power envelope from rack density × row count × cooling factor. Includes typical AI-rack profiles (40 kW air, 132 kW liquid).
Open tool →DesignCapacity Model
Convert MW available → GPU count → tokens/sec → useful training throughput. Sanity-check vendor capacity claims.
Open tool →OperateIncident Replay
Replay real AI-data-center incidents (overheat, transformer failure, water-leak) and see how recovery procedures cascade through the stack.
Open tool →Looking for lessons instead?
The tools above assume you already understand the concepts. If you want the underlying material first, start with the lesson tracks.
Browse lessons →