Builder
You’re the developer now.
Every choice has consequences. This is the numbers view. The digital twin is the campus. Screening model with low / expected / high ranges — not a permit.
Facility load
7.25MW
6.15–8.96
Low · expected · high
PUE
1.38
1.23–1.60
Low · expected · high
WUE
0.12L/kWh
0.02–0.45
Low · expected · high
Annual energy
37,717MWh
22,412–59,764
Low · expected · high
Annual water
833k gal
93–4,270
Low · expected · high
Land
4.0acres
2.2–8.3
Low · expected · high
Racks (approx.)
100
56–250
Low · expected · high
Permanent ops jobs
6
4–10
Low · expected · high
Power is not energy
5 MW IT is about 1,250 homes at an ERCOT peak-hour snapshot (250 homes per MW). Annual energy at expected utilization is about 3,592 U.S. households (10,500 kWh/year). Those are different comparisons. Neither is a San Antonio meter.
Potable vs all water
Expected potable share of facility water: 10%. Reclaimed water still leaves the recycled system. It does not vanish.
Where did this number come from?
- Liquid at the chip, dry rejection outdoors. Not automatically water-intensive.
- On 110°F days a dry cooler may need adiabatic assist. That is a water choice at the peak, not all year.
- Climate assumed: ASHRAE 2A hot-humid. Dry coolers are viable but not free on design days.
- These are screening ranges, not a design. Real numbers require drawings, hours of wet operation, and a meter.
- Redundancy adds equipment and land more than it adds annual energy. Standby plant is not a second full load.
- PUE/WUE bands drawn from LBNL 2024 fleet averages, operator disclosures, and ASHRAE 2A climate penalties — then widened. Not a CFD study.
- Jobs: Hamm Institute-style intensity (construction peak vs ops FTE). Construction is temporary.
- Utilization is the silent multiplier. An empty hall uses a fraction of nameplate.
Change the assumptions. See what happens.
Data last checked 2026-09-04 · Current public data is a snapshot, not a live meter
Show me the receipts