From megawatts to money: choosing the right capacity valuation
Chapter 3 of 4 · Designing a data-center capacity calculator
After a calculator estimates capacity, it still needs to answer another question: what kind of value does that capacity represent for this organization?
The PhysaFlow implementation supports hosting revenue and GPU owner cost as separate valuation bases. This distinction matters because each describes a different economic perspective on the same modeled capacity.
Establish eligibility before showing a scenario
The built-in hosting revenue scenarios are available to colocation organizations or users who explicitly confirm hosting eligibility. The calculator also accepts a site-specific revenue-per-MW-year value or range.
This keeps a business-model assumption attached to the output. An internal enterprise facility should not automatically inherit the revenue assumptions of a hosting provider.
Using the illustrative 11.64 MW of modeled stranded capacity and the reference hosting scenario of $1.6 million per MW-year produces an annual revenue opportunity proxy of $18.624 million. This is arithmetic from the supplied scenario, not evidence of realizable revenue. The provided files do not establish the provenance of the scenario price.
Separate opportunity from recovery
A recoverable share is an additional input. Without it, the calculator does not produce potentially recoverable capacity or recoverable revenue.
For illustration, a 25% recovery assumption would produce 2.91 MW of potentially recoverable capacity and a $4.656 million annual revenue proxy under the same price scenario. The 25% is an added hypothetical assumption, not part of the supplied example or a recovery prediction.
This distinction deserves visible space in the product. Users need to see which number comes from measured inputs and which depends on an intervention assumption.
Give GPU equivalents their own boundary
GPU-hour equivalents require an installed GPU count, GPU-scoped IT capacity, a hardware model, and confirmation that the scope matches the facility boundary. The reporting dates determine the hours in the calculation. These outputs express equivalents; they do not establish actual idle GPU-hours from workload telemetry.
When a GPU hourly owner cost is available, the calculator can value that equivalent. If both hosting revenue and GPU owner cost are available, it requires a valuation choice before displaying dollar results.
The presentation explicitly says the two valuations must not be added. They are alternative interpretations, not independent benefits. A well-explained result helps the reader choose the interpretation that fits the decision—and keeps the capacity assumptions visible underneath it.
This chapter is part of the PhysaFlow calculator series overview. The PhysaFlow case study explains the surrounding positioning and product work. Continue the reasoning with defining usable and modeled stranded capacity.