What the record shows
A documented result from existing technology somewhere else, with its source at the bottom of the page. It is not a result from Node Zero, which isn’t built yet.
What the Record Shows
The benefits we design for, held to what has actually been measured — with the limits beside them.
Published sources only · Last checked October 2026
How to Read This Page
Three columns for every claim.
The honest limit
What the same record doesn’t show, or where it points the other way. We put it right beside the benefit, not in a footnote.
What we must measure
What Node Zero has to measure and publish before we claim the benefit for ourselves. Measure first, claim second.
The Water Comes First
The test every other benefit has to pass.
Nothing below counts if the water is harmed. This is the standard we are measured against, and the record that tells us why it matters.
The rule, and the warning
What the record shows
South Carolina’s water law covers private and artificial waters, groundwater included — not only lakes and rivers.18 For lakes, its rules limit warming to 5 °F above natural conditions (weekly average), never above 90 °F, and require freshwater dissolved oxygen to average at least 5.0 mg/L a day.8
The honest limit
Even at Lake Keowee, South Carolina, a reservoir with water flowing through it, power-station cooling heat left the water column 1–4 °C warmer and pushed its layering deeper (studied 1971–82).9 Flooded pits can hold permanently layered, oxygen-free bottom water, and a sudden overturn can kill fish.10 A quarry pit with no stream in or out has no current to carry heat away, and depending on the site, a flooded pit can pass water on into the groundwater around it.19 We found no record of computing heat in an enclosed pit.
What Node Zero must measure first
A full year of temperature, oxygen, and bottom-water chemistry at every depth, how the pit trades water with the surrounding groundwater, and a heat budget set against the pit’s volume, before any hardware enters the water. If the numbers don’t allow it, it doesn’t happen.
Compute & Energy
What water-cooled and sealed computing has actually shown.
Each finding below is about existing, published technology elsewhere — not about Node Zero, which isn’t built.
Energy efficiency
What the record shows
The average data center surveyed in 2025 used about 1.54 units of energy for every unit that reached its computers.1 A U.S. national laboratory has run liquid-cooled computing at about 1.03,2 and Google reports 1.10 for its Berkeley County, South Carolina campus in 2024.3
The honest limit
Natural-water cooling has not been shown to beat the best land practice on energy: Google reports the same 1.10 for its seawater-cooled site in Hamina, Finland.3 We don’t claim energy savings.
What Node Zero must measure first
Energy efficiency over at least twelve months, including a South Carolina summer, with the measurement boundary stated, against both the industry average and the 1.10 already achieved in the state.
Drinking water left alone
What the record shows
In 2024, Google’s seawater-cooled Hamina site consumed about 0.3 million gallons of freshwater. Its Berkeley County, South Carolina site consumed 776.5 million.3
The honest limit
Seawater is left out of those figures by definition, and a quarry pit is not the sea. We found no documented case of computing cooled inside a freshwater lake or flooded quarry. In a closed pit, some of the heat would leave as evaporation, and that is water used too.
What Node Zero must measure first
Freshwater consumed per unit of computing, including any extra evaporation from the pit that our heat causes.
Hardware that lasts longer
What the record shows
In a two-year sea trial (2018–2020), Microsoft reported that servers sealed in a nitrogen-filled vessel failed at one-eighth the rate of an identical group on land.4
The honest limit
The result is reported by the operator, from a single vessel of 864 servers in open seawater.4 It has not been repeated in fresh water.
What Node Zero must measure first
Our own hardware failure rate against a land-based control group, published either way.
Cold water in place of refrigeration
What the record shows
Cornell University reports that its lake-water cooling has cut the electricity used by its campus chilled-water system by about 86% since 2000.5
The honest limit
Moving deep lake water also moves nutrients. Phosphorus became the regulators’ central concern, and Cornell’s permit caps it and required years of study.5
What Node Zero must measure first
Phosphorus and other nutrients at depth and at the surface, before and during operation, so any change is caught.
Putting heat to use
What the record shows
Computing heat is the primary heating source for a U.S. national laboratory’s offices and labs.2
The honest limit
The working examples are in places that need heat most of the year. A U.S. city-greenhouse project vented its computing heat outdoors in spring and fall, and its machines could need shutting down in summer.6 South Carolina’s climate makes reuse harder, so we make no reuse claim.
What Node Zero must measure first
Metered heat delivered to any use, month by month, and the share of heat that still goes to the water.
Renewable power
What the record shows
Microsoft’s sea trial ran on Orkney’s grid, supplied by local renewable power.4
The honest limit
U.S. utility-scale solar produced about 23% of its rated output in 2024 (preliminary).7 Google reports that 31% of its hourly use on the South Carolina regional grid was carbon-free in 2024.3 We found no case of round-the-clock computing run on on-site renewables alone.
What Node Zero must measure first
Hourly on-site generation against hourly load, and grid power drawn by the hour, rather than an annual average.
Land, Food & People
What gardens, growing, trails, and lodging have shown.
The public side of the campus is designed to be shared. Here is what the record supports, and what it doesn’t.
Growing food under cover
What the record shows
A modeling study found hydroponic lettuce could yield about 11 times more per square meter than field lettuce, using about 20 liters of water per kilogram instead of 250.11
The honest limit
The same study estimated about 82 times more energy per kilogram, and its hydroponic figures were calculated, not measured.11
What Node Zero must measure first
Kilograms harvested per square meter per year, by crop, with the water and energy used for each kilogram.
Drip irrigation
What the record shows
In a state-funded Texas field trial, subsurface drip used 41% less water than flood irrigation and 38% less than center pivot.12
The honest limit
Using less water on one field does not guarantee water is saved across a watershed, because less water returns to streams and aquifers.13
What Node Zero must measure first
Water applied for each kilogram harvested.
Community gardens
What the record shows
In a randomized trial in Colorado, new community gardeners ate more fiber, were more physically active, and reported less stress and anxiety than a waiting-list group.14
The honest limit
The effects were modest, and the trial found no change in fruit and vegetable intake or body weight.14
What Node Zero must measure first
Gardeners and hours, produce grown and where it went, and a validated before-and-after survey for any health claim.
Farmers markets and SNAP
What the record shows
SNAP households made about 2.2 million purchases at farmers markets and direct-marketing farms in fiscal 2021, up 162% since 2017.15
The honest limit
National growth says nothing about how many local families one market will serve.
What Node Zero must measure first
SNAP/EBT transactions and dollars once the market is authorized, and the share of vendors who are local.
Time in nature
What the record shows
In a survey of nearly 20,000 people in England, those who spent at least 120 minutes a week in nature were more likely to report good health and high wellbeing.16
The honest limit
This is an association from a survey, not proof that a trail causes better health.16
What Node Zero must measure first
Trail counts and visitor surveys. No health claim without our own before-and-after data.
Eco-cabins and local benefit
What the record shows
Nature tourism keeps its money local only when ownership, staff, and supplies are local. One studied fishing lodge lost 86.1% of its local tourism revenue to outside supplies, staff, and profits.17
The honest limit
We found no U.S. study that measures the local economic benefit of small rental eco-cabins.
What Node Zero must measure first
Cabin revenue, and the share of spending and wages that stays in the county.
Not on This Page Yet
Where the record isn’t strong enough to quote.
- Earth-covered cabins. The best measured data on earth-sheltered and Earthship-style buildings come from dry or temperate climates, not humid South Carolina. We are researching the record for the building method we actually intend before saying anything about it.
- Swimming safety. Unsupervised quarry swimming is a known drowning hazard. Water recreation on our campus is planned only with trained lifeguards on duty and tested water, and we’ll publish the safety log rather than borrow anyone else’s numbers.
Sources
Where every finding comes from.
- Uptime Institute, Global Data Center Survey 2025, July 2025. uptimeinstitute.com
- National Renewable Energy Laboratory, Lessons Learned from the World’s Most Energy Efficient Data Center (NREL/PR-2C00-70433), 2018. docs.nlr.gov
- Google, 2025 Environmental Report (2024 data): data center efficiency, water use by location, and grid-region carbon-free energy tables. www.gstatic.com
- Microsoft Research, Project Natick and Project Natick: Facts and figures (operator-reported). www.microsoft.com
- Cornell University Facilities, Lake Source Cooling FAQ, June 17, 2020 (operator-reported). fcs.cornell.edu
- University of Notre Dame, Cooperative Computing Lab, “Green Cloud” project note, January 11, 2010. ccl.cse.nd.edu
- U.S. Energy Information Administration, Electric Power Monthly, Table 6.07.B, capacity factors (2024, preliminary). www.eia.gov
- South Carolina Department of Environmental Services, Regulation 61-68, Water Classifications and Standards (agency copy; the State Register is the official text). www.des.sc.gov
- Oliver & Hudson, effects of power-station cooling on Lake Keowee, South Carolina, Water Resources Bulletin 23(2), 1987. pubs.usgs.gov
- Boehrer & Schultze, “On the relevance of meromixis in mine pit lakes,” 7th International Conference on Acid Rock Drainage, 2006. www.imwa.info
- Barbosa et al., “Comparison of Land, Water, and Energy Requirements of Lettuce Grown Using Hydroponic vs. Conventional Agricultural Methods,” Int. J. Environ. Res. Public Health 12(6), 2015. pmc.ncbi.nlm.nih.gov
- Texas Water Development Board / El Paso Water, Comparison of Drip Irrigation to Flood and Center Pivot Irrigation Systems in Hudspeth County, 2021. www.twdb.texas.gov
- Ward & Pulido-Velázquez, “Water conservation in irrigation can increase water use,” PNAS 105(47), 2008. doi.org
- Litt et al., community gardening randomized trial (CAPS), Lancet Planetary Health 7(1), 2023. doi.org
- USDA Food and Nutrition Service, 2021 SNAP Retailer Management Year End Summary. fns-prod.azureedge.us
- White et al., “Spending at least 120 minutes a week in nature is associated with good health and wellbeing,” Scientific Reports 9:7730, 2019. doi.org
- Butler et al., “Can Fishing Tourism Contribute to Conservation and Sustainability via Ecotourism?” Sustainability 12(10):4221, 2020. doi.org
- South Carolina Pollution Control Act, S.C. Code § 48-1-10, definition of “waters” (unannotated online code). www.scstatehouse.gov
- Geological Survey of Finland, Mine Closure Wiki: Mine flooding. mineclosure.gtk.fi
Figures checked against their sources in October 2026. Results marked operator-reported were published by the company or institution that ran the project and have not been independently verified. If you find an error, tell us through the contact page and we’ll correct it in the open.
The record is a starting point, not a promise.