Checking publisher feeds…
Banning & the Pass
Saved views & sharing
Grid demand & supply
Checking California ISO…
California ISO / full demand and supply dashboardWhat does a workload require?
Explore a hypothetical image-generation workload. The starting 1 Wh per image is an editable example, not a measured average or a claim about any AI service. This scenario is independent of the selected geography.
Computing equipment only. If your measurement already includes facility overhead, do not apply PUE again.
Facility electricity ÷ IT electricity.
Water consumed on site per unit of IT energy.
1 kWh computing + 0.3 kWh facility overhead.
Uses IT energy × water intensity. Changing PUE alone does not change this estimate.
Operational estimates only. Excludes model training, equipment manufacturing, networks and off-site water for power generation. No images are generated here.
Show the math, limitations and research
IT kWh = image count × assumed Wh per image ÷ 1,000. Facility kWh = IT kWh × PUE. On-site liters = IT kWh × WUE. US gallons = liters ÷ 3.785411784.
These are proportional scenarios with constant assumptions. In practice, model, resolution, inference steps, hardware, batching, cooling and weather change the result. Image counts should include discarded generations and retries. The data-center load represents electrical capacity used, not GPU occupancy.
WUE uses liters per IT kWh; it is not multiplied by PUE. On-site consumption is not the same as withdrawals or a permit’s water allocation. No assumption connects a chosen AI service to Banning’s power or water supply. There is no universal per-image figure.
The 2024 FAccT study by Luccioni, Jernite and Strubell reports a mean of 2.907 kWh per 1,000 image-generation inferences across its tested models. It used an A100 GPU and sequential requests without batching. That historical benchmark is context, not this calculator’s default or an estimate for a current commercial service.
Read the 2024 inference-energy study (PDF) ↗
Read Berkeley Lab’s data center research ↗Water, in reported gallons
Monthly supplier reports from the California State Water Resources Control Board. These are service-area reports, not live meters or city/county totals. Supplier selection is separate from the news-area filter.
Retained monthly reports
Local retention covers Banning, Beaumont–Cherry Valley and Yucaipa Valley. Each successful capture checks up to 24 calendar months for changed values and quality flags. Missing reports are not deleted or treated as zero. This is separate from the state’s current dataset.
Reading retained reports…
Check this supplier’s current state records ↗What the numbers mean
Net potable supply excludes water delivered to other entities. Residential use covers single- and multi-family use. Neither measure identifies AI or data-center consumption, and supply is not the same as customer consumption. Supplier service areas can cross city boundaries.
The state’s residential gallons-per-person-per-day figure uses reported residential gallons, population and days in the month. Climate, landscaping, housing and service-area differences mean it is not a fair efficiency ranking between suppliers. Same-supplier year comparisons are descriptive, not an explanation of causes. Before 2023, residential gallons were derived from reported residential share and supply; from January 2023, the dataset uses reported residential volumes.
Source-flagged or invalid values are excluded from headline numbers, charts and comparisons. Missing values stay missing. Preliminary estimates are labeled and published reports can be revised. The current state view can change when reports are revised. The separate retained view preserves captured versions for three local suppliers, starting from its stated baseline. Smaller systems may not appear in this urban-supplier dataset.
State data dictionary and flag methodology ↗Where the electricity goes
Historical release · through 2024. This is reported energy use, not a live meter. CEC source updated January 27, 2026; snapshot retrieved September 29, 2026. Updates require a reviewed dataset refresh.
Banning municipal utilityUtility service territory; not a city-wide or county-wide total. Separately reported planning-area self-generation is not assigned to this utility. Geography is independent of the news filter.
+10.8% compared with 2023, for the same geography and measure.
A year, month by month
Select a month to inspect it. Bars start at zero; the table includes all three measures. GWh measures energy over time; the live grid panel uses MW for power.
Chart scale: 0–22.12 GWh. Choose a bar to read its value.
Annual sector breakdown
2024 · all reported sectors, regardless of measure selection. Sector detail comes from the annual file.
- Commercial
- 71.41 GWh (44.4%)
- Residential
- 60.65 GWh (37.7%)
- Agriculture and Water Pumping
- 11.95 GWh (7.4%)
- Industrial
- 8.87 GWh (5.5%)
- Transportation, Communications, & Utilities
- 4.47 GWh (2.8%)
- Streetlighting
- 3.21 GWh (2.0%)
- Mining
- 0.18 GWh (0.1%)
Read all monthly electricity values
| Month | Total | Residential | Non-residential |
|---|---|---|---|
| Jan 2024 | 10.46 | 3.92 | 6.55 |
| Feb 2024 | 10.67 | 4.66 | 6.01 |
| Mar 2024 | 10.47 | 3.77 | 6.7 |
| Apr 2024 | 10.43 | 4.33 | 6.1 |
| May 2024 | 10.26 | 3.59 | 6.68 |
| Jun 2024 | 11.32 | 3.91 | 7.41 |
| Jul 2024 | 17.87 | 7.49 | 10.38 |
| Aug 2024 | 22.12 | 9.65 | 12.47 |
| Sep 2024 | 18.6 | 7.11 | 11.49 |
| Oct 2024 | 15.79 | 5.01 | 10.78 |
| Nov 2024 | 12.63 | 4.01 | 8.62 |
| Dec 2024 | 10.11 | 3.21 | 6.9 |
One GWh equals one million kWh. County and utility footprints overlap: do not add them together. SCE covers a broad service territory and is not a Beaumont or San Gorgonio Pass total. These reports do not isolate AI or data-center use.
Source: California Energy Commission (2026), California Electricity Consumption. Annual totals and residential totals were checked against the sum of the 12 monthly values for each displayed year.
What the reservoirs actually hold
Daily reservoir storage from California DWR. Storage is a volume on hand, not consumption, a city allocation, or proof of water available for a proposed project. Missing readings are excluded.
Checking…Reservoir observations are not available in this snapshot. Open DWR current conditions ↗
Why there isn’t a live “Banning gallons consumed” counter
No verified city-wide live consumption feed is connected. Reservoir storage cannot be relabeled as Banning’s water use, and an AI workload cannot be assigned a universal water cost. The resource lab models a scenario; the supplier panel above shows monthly reported supply and residential use, with reporting dates and quality flags. Those reports cannot establish minute-by-minute use.
The line connects available daily observations across the requested 30-day window; gaps can occur. Its vertical axis is cropped to the observed range to show variation, not percent full. Gallons use approximately 325,851.429 US gallons per acre-foot. DWR observations can be revised.
What defenders need to know
Latest additions returned by CISA’s Known Exploited Vulnerabilities catalog. These are software vulnerabilities, not evidence of an incident in Banning. Listed due dates apply to covered federal civilian agencies.
Checking the CISA catalog…
Full CISA catalogOur reporting. Our responsibility.
The source wire above links to other publishers. This section holds original work reviewed and published by our newsroom.
The newsroom has not published its first report yet. External coverage remains available in the live source wire.
Open the reporting archive →