How the estimate works
The capacity assessment is a deterministic scenario model. The same inputs produce the same result. Its coefficients are model assumptions; they are not presented as validated industry benchmarks or measured PhysaFlow deployment outcomes.
Capacity model
Usable IT capacity is the minimum of all available valid grid, commissioned electrical and cooling IT constraints. Grid IT capacity is gross deliverable power minus intentional reliability headroom, divided by reporting-period PUE. Missing constraints are excluded and disclosed. Invalid supplied inputs must be corrected.
Utilization is measured average IT load divided by usable IT capacity. Loads above usable capacity are invalid. Unused capacity is not automatically stranded. The modeled rate is unused rate multiplied by a cooling scenario coefficient: air 0.485, liquid/immersion 0.340, hybrid 0.4125. Tier and visibility adjustments are zero. The result is bounded between zero and unused capacity. Unknown cooling permits raw capacity results only.
Economic assumptions
Annual revenue opportunity uses site-specific revenue assumptions or eligible hosting/colocation sensitivity cases of $1.2M, $1.6M, or $2.355M per IT MW-year. This is a revenue proxy, not universal financial loss. Ranges remain ranges. Recoverable capacity and revenue require a supplied site-specific recoverable share; there is no default recovery percentage.
GPU equivalent and reporting period
GPU equivalents use reported installed inventory, matching GPU IT capacity, hardware model, and actual reporting hours. Stranded capacity is allocated proportionally to the GPU share of usable capacity. Owner cost is supplied separately. Revenue and GPU owner cost are alternative valuations and are never added together. Dates define a UTC interval with the start included and end excluded; leap years and partial periods use their actual hours.
The homepage scenario
The homepage economics slider illustrates a stranded-capacity range of 30–40% of selected facility capacity and an implied annual value of $12.5M per stranded MW. At 100 MW, this gives 30–40 MW and $375M–$500M per year. These are illustrative website assumptions, separate from the calculator’s engineering model and hosting revenue scenarios. They are not measured outcomes or a recovery guarantee. The capacity breakdown in the problem section separately demonstrates the engineering model at 80 MW usable IT capacity, with utilization and air-cooling assumptions disclosed in its technical drawer.
Limits and next steps
The model does not simulate thermal transients, redundancy, network dependencies, workload eligibility, or site-specific control limits. Use it to scope further measurement. Engineering review and operating data are needed to establish usable capacity and approve any changes.