Exercises where the controls actually run and the rest is honestly labelled as simulation.
Size the cluster, see where the money actually goes, and find the volume at which self-hosting an open-weight model stops losing money against a hosted API.
20 minutes to size one workload; longer if you argue with the defaults, which you should
Start with your headcount, finish with a defensible number. Cost per successful outcome — and why people, not tokens, usually dominate the bill.
25 minutes with your own numbers; five minutes to see the shape
Cost the five drivers, turn released hours into a defensible benefit, compare buy, extend and build, and see ROI, payback and sensitivity.
30 minutes with your own numbers; five minutes to see the shape