Worked example · nothing withheld
Read the deliverable before you pay for it.
This is the complete output of a Capacity & Density Envelope Study — the document, the walkthrough deck, the proposal, and the tables behind every number — run on a synthetic European colocation hall. Not a sample, not an extract. The whole thing.
The hall does not exist. Every figure is modelled from invented inputs and none is a measurement of a real site — which is exactly why we can publish it in full.
Deployable today
0
racks · rack feed / tap-off rating binds first
On the grid connection alone
28
racks, after the costed ladder
With behind-the-meter supply
62
racks · 160 weeks, set by transformer capacity
Fourteen sections and an appendix: the answer, the headroom ladder, time to power, every architecture compared, sensitivity, economics, risks, the model's own accuracy record, and the assumptions and provenance behind the headline figures.
Open →The board summaryOne page a sponsor can forward without editing: the answer, what decides it, how firm it is and which single input it depends on most, and what to decide next.
Open →The walkthrough deckNine slides from the same solved model, for the session with your engineering team. Arrow keys to advance; print to PDF to send.
Open →The proposalWhat a proposal looks like when the engine has already run: the binding constraint and the rack counts before the fee, and the data request in section three.
Open →The tender specificationWhat you send to suppliers once the study says what binds. Every duty derived from the model and naming the constraint it came from, quoted at this site's own water temperature rather than a manufacturer's reference condition.
Open →And the schedule that makes bids comparable
Four quotations normally answer four different questions. The specification ships with a machine-readable response schedule: the supplier returns these fields in these units, and the bids are then compared in racks and weeks rather than only in euros — because the item exists to unlock compute on a date, and the cheaper, slower bid is the expensive one.
Check our arithmetic
The tables behind the document, as files your own engineers can open. Every headline quantity in provenance.csv carries a digest and the digests it was derived from, so any number can be walked back to the assumption underneath it. A study you cannot check is a study you have to take on trust, and nobody sensible does that with a capital decision.
This example runs at 83% intake completeness, with 5 inputs still assumed — and the document names every one of them rather than hiding them in a caveat.
Now run it on your own hall.
Seven numbers you already know, and the engine names the constraint that binds yours. Free, and it takes a minute.