On data centers, weaponry over livingry, the severed water cycle, and the signs of a civilization that needs to cool itself.
Ask whether data centers are "warming the planet" and the arithmetic will reassure you: their direct heat is a whisper next to greenhouse forcing. But that is the wrong instrument for the question.
Earth is a coupled heat–water–vegetation circulation system. What matters locally is not how much energy arrives, but how the land partitions it — into latent heat that lifts water skyward through leaves, or sensible heat that simply bakes the air. A forest and a city can receive the same rain and the same sun, and send them down utterly different roads. The data center is only the newest organ of that city — the hottest, densest, most continuous one — stitched into a body that was already running a fever.
Toggle the landscape. Watch where the energy goes.
Illustrative partition, not measured values — the shape of the shift is what the literature describes: urbanization pushes the surface energy balance toward sensible heat and away from latent heat.
A 1‑gigawatt computing campus compresses roughly a gigawatt of continuous heat rejection into a small footprint. Unlike solar heating or most daily urban activity, the thermal load does not set with the sun.
A city is a thermal disturbance in its landscape: built surfaces absorb and re-emit solar heat, while buildings, vehicles, and industry add waste heat. A hyperscale campus is a concentrated thermal disturbance nested within that urban field — round-the-clock heat, sealed ground, cooling-water demand, and rapid runoff inside an already stressed regional climate. Atmospheric studies show that concentrated urban heat can deepen mixing, organize convergence, and modify heavy rainfall. How far a data-center cluster extends that effect remains an open, testable question. Tap the chain.
The same powered environment also changes the local electromagnetic environment through substations, cabling, servers, and communications equipment. This is a separate engineering and exposure-management consideration from waste heat — not evidence by itself of ecological or health harm — and should be measured against established guidance rather than folded into the climate mechanism. WHO guidance on electromagnetic fields distinguishes measurable biological effects from confirmed health hazards.
These are urban studies, not data-center studies — that specific modeling barely exists yet. Which is exactly the gap.
Simulations of a record rainfall event varied anthropogenic surface heat directly. More heat meant more turbulence, buoyancy, and convection — and heavy precipitation responded more sensitively than light rain did.
In a modeled Pearl River Delta rainstorm, anthropogenic heat and urban land-use change each independently intensified the rain.
Added anthropogenic heat increased sensible heating, boundary-layer mixing, instability, upward motion, and convergence — and increased rainfall in all four events examined.
Modeling an extreme NYC thunderstorm with anthropogenic heat included produced a stronger urban heat island and stronger low-level moisture convergence, enhancing the precipitation.
Honesty clause: none of this attributes any particular flash flood to data centers, and the dominant driver of rising extreme precipitation globally remains greenhouse warming — warmer air holds more vapor. The narrower claim is the live one: giant heat clusters are a plausible, testable additional local forcing.
While the campuses were being poured, the sky was already keeping records. What follows is the warming baseline — the loaded dice — onto which gigawatt heat sources are now being bolted. None of these events are attributed to data centers. They are the context.
A slow-moving system dropped record daily rainfall on several cities; meteorologists put the event's statistical likelihood at once in 500 to 1,000 years.
Training storms triggered flash-flood emergencies, mudslides, and swift-water rescues — one of several one-in-200-year rainfall events to hit the Northeast that year.
Enough water to bury North Carolina three and a half feet deep, fueled by a record-warm Gulf. One storm accounted for more than thirty flash-flood emergencies.
Devastating flash floods swept central and eastern Europe in the hottest year in 175 years of observation — 1.55 °C above the pre-industrial baseline.
Deadly flash floods from rainfall that attribution science found roughly thirteen percent more intense because of warming.
The Guadalupe River disaster — torrential rain over the July 4 weekend produced one of the deadliest inland floods in United States history.
A subway station rain gauge beat its previous one-hour record in half the time, in a summer when flash-flood warnings ran at the fastest pace ever logged.
Weeks of record rainfall from stacked tropical systems left at least 219 dead or missing across the South Central Coast and Highlands.
Sand Run near Buckhannon rose more than eight feet in under five hours, crushing a flood-crest record that had stood since 1985. All 55 counties were placed under a state of emergency, in a July when nearly half the year's US flash-flood emergencies were issued in a single month.
One square kilometer and a dial with two directions. The center is the post-industrial default: cleared, sealed, inert. Turn right and you pour power onto the ground — the city's move, perfected by the campus. Turn left and you give the ground back its function: wetland, canopy, cold water, and the compounding returns of a system that fosters life.
Order-of-magnitude sketch, not an engineering model. Rightward: essentially all electricity becomes heat on the fixed footprint, with evaporative cooling near the USGS illustrative rate. Leftward: mature-forest latent cooling on the order of tens of megawatts per km², sequestration in the hundreds of tonnes per year, recharge as a fraction of annual rainfall — and a life index that stands in for what resists metering: the compounding synergies of a system optimizing itself.
When a cooling tower drinks an aquifer, the water is not destroyed. It rejoins the sky and rains out somewhere — perhaps three hundred miles downwind. The globe keeps its water. The watershed does not.
The USGS notes that many data-center cooling systems consume water through evaporation rather than returning it locally — illustratively, around two million gallons a day for a 100‑MW facility, depending heavily on cooling technology. In Michigan, the Huron River Watershed Council warns that proposed facilities may draw groundwater or municipal supplies connected to Great Lakes waters, raising questions of recharge, wetlands, baseflow, and drought resilience — while impervious development accelerates stormwater on top of it.
So a region can hold two truths at once:
More atmospheric moisture, sharper convection, heavier bursts of rain, faster flashier runoff.
Less infiltration, falling recharge, thinner baseflow, warming spring-fed streams, cold-water refugia lost.
That paradox is no longer abstract in the upper Midwest. The U.S. Drought Monitor, valid August 18, 2026, reported drought degradation across parts of Michigan and Wisconsin after another week with little to no rainfall, including an area of extreme drought in northern Wisconsin. Earlier in August, the Monitor described 90-day precipitation deficits of roughly four to eight inches across much of northern Wisconsin and adjacent Michigan, with some areas receiving only a fraction of normal rainfall.
Anyone walking these landscapes can see the expression of that deficit before opening a map: browned grasses, stressed leaves, hardened surface soils, shallow wetlands pulling back. Those observations are local and qualitative, not proof of a particular cause. But they are exactly the surface symptoms expected when precipitation fails to keep pace with atmospheric demand and stored soil water is progressively spent.
The key hydrological idea is evaporative demand. Warm air can contain more water vapor, yet it can also pull harder on soil and vegetation. More moisture in the atmosphere therefore does not guarantee more moisture in the ground. If the land receives heat faster than rain and infiltration can replace what is lost, soil moisture declines even while the atmosphere carries more total vapor.
What happens when regional warming, urban heat islands, impervious land, groundwater withdrawal, evaporative cooling, and clusters of continuously operating data centers are added to that already-thirsty system? Do they measurably increase local evaporative demand, shift water from terrestrial storage into atmospheric transport, intensify boundary-layer mixing, or help redistribute rainfall into shorter, heavier, more spatially uneven events?
Data centers are especially interesting because several mechanisms arrive together. Electricity ultimately leaves the equipment as heat. Large campuses can reject that heat continuously through air or cooling systems. Some cooling designs consume water by evaporation, moving locally withdrawn water into the atmosphere rather than returning it to the same watershed. The campus footprint can also replace vegetation and permeable soil with roofs, roads, substations, and hardscape. Each mechanism is established individually. What remains unresolved is their cumulative regional climatic importance when many large facilities are concentrated in the same landscape.
This distinction matters. The Codex is not claiming that data centers caused the 2026 Michigan–Wisconsin drought or that recent flash floods can be assigned to a server campus. Atmospheric circulation and greenhouse-driven warming remain dominant large-scale forcings. The narrower proposition is both stronger and more useful: thermal infrastructure may act as an additional local or regional amplifier by changing how heat and water are partitioned between soil, vegetation, groundwater, runoff, and the lower atmosphere.
In that frame, the apparent contradiction resolves itself. A hotter system can become simultaneously wetter above and drier below: more vapor overhead, less water banked in soil; heavier rain when conditions finally break, but less slow recharge between storms. Water has not vanished. Its residence time, phase, pathway, and location have changed.
Viktor Schauberger — the Austrian forester they called the Water Wizard — spent a lifetime insisting that water is a living circulation, ruined the moment we straighten it, expose it, and heat it. He watched deforested valleys lose their springs and called the resulting landscapes exactly what they were: wounded.
He would recognize the thermal disturbance instantly. Not as a novel technology problem, but as the oldest mistake wearing new hardware: taking a landscape whose genius was cooling, spiraling, sinking, storing — and replacing it with a machine whose primary physical by-product is concentrated heat, released continuously into its surroundings.
“Comprehend and copy nature.”Viktor Schauberger · Kapieren und kopieren
His core dichotomy maps onto this moment with uncomfortable precision. Our civilization runs almost entirely on what he called the explosive principle — centrifugal, heating, expanding, degrading. The forest runs on the implosive — centripetal, cooling, condensing, ordering. Cold power.
Translated out of his poetry and into hydrology: cold groundwater, shade, infiltration, and transpiration are the stabilizing components of a watershed's energy and water budgets. Cold water carries more dissolved oxygen; groundwater is a vast thermal buffer; the canopy is surface-area architecture for moderating everything above and below it. Schauberger said water wants to stay cold. Science says cold water is where the system keeps its stability. These are the same sentence in two dialects.
To embody cold power is not to abandon computation. It is to demand that our densest machines be designed the way a forest is designed — heat recaptured and put to work, water returned to the ground it came from, canopy and infiltration built back around the footprint, rhythm instead of relentlessness. In step with nature, or in debt to it.
Not "are data centers warming the Earth?" — a question the global average answers too quickly and too smugly. Ask instead:
At what concentration does continuous anthropogenic heat rejection — combined with land-cover conversion and consumptive cooling-water use — measurably alter a region's boundary-layer dynamics, precipitation distribution, groundwater recharge, and watershed thermal regime?
For a place like the Great Lakes basin, answering it means one coupled model tracking waste-heat flux, cooling-tower moisture, forest removal, impervious acreage, groundwater withdrawal, lake breezes, nighttime boundary layers, downwind rainfall, and the temperature of the streams the trout still need. That would test the idea — rather than dismissing it with a thermometer.
The urban thermal disturbance predates the server. Long before the first rack was bolted down, the modern city had already declared war on its own water cycle — sealing the ground, entombing its streams, and trading transpiration for exhaust. The campus does not invent the wound. It moves into one.
A dense downtown is a landscape flipped inside out. Ground that once drank now sheds; streams that once meandered now run in pipes beneath the streets, engineered for velocity instead of storage — hydrologists call the result urban stream syndrome: flashy floods, starved baseflow, eroded channels, water too warm for anything that evolved in it. Meanwhile the surface energy budget inverts. With almost nothing left to transpire, the sun's whole delivery becomes sensible heat, stored in asphalt all day and re-radiated all night.
The consequences are no longer subtle. Zhengzhou, July 2021: a year's worth of rain fell on a city of ten million in three days; the pavement handed it straight to the subway, and more than three hundred people died. Jakarta: decades of impervious sprawl cut the land's capacity to absorb its own rivers while over-pumping sinks whole districts — parts of the city dropping as much as 25 centimeters a year, roughly forty percent now below sea level, so severe that Indonesia is building itself a new capital. Mexico City: sinking as much as half a meter a year in places from aquifer collapse, even as it floods in the wet season and pipes drinking water in from other basins. Sealed above, drained below — the signature of the wound is always the same.
And this is the host into which we are now stitching gigawatt-class heat organs. A data center in a forest clearing is a stress. A data center in a sealed, pre-heated, stream-entombed metropolis is a stress multiplied by everything around it.
The sky's records get the cameras. The ground's records go unphotographed — because the ground's records are absences. Before redesign, the quiet ledger:
A city built the old way is a wound that never closes: pavement sheds the rain it should drink, buried streams pipe the surge it should store, and every roof and lot converts sun into sensible heat with nothing transpiring it away. Add a computing campus and you've stitched a furnace into the scar tissue. The redesign is not exotic. It is the forest's operating principles, translated into infrastructure — a city that absorbs, stores, cools, and releases slowly. A sponge.
Each graft below replaces a piece of scar with a piece of skin:
Nearly all of a data center's electricity can be recovered as heat. Microsoft's Espoo scheme is built to warm a quarter-million Finns — roughly 40% of regional heating demand; Google's Hamina facility gives 80% of local district heat away free; Meta's Odense campus has fed 12,000+ homes since 2019. The IEA estimates recovered data-center heat could warm about a tenth of European homes by 2030.
Liquid cooling keeps the water in the pipe instead of sending the aquifer into the sky — and returns hotter, more usable heat as a bonus. The consumptive withdrawal that severs the local cycle becomes optional.
Depaving, bioswales, and rain gardens intercept the burst and hand it downward — converting flash runoff back into the slow, cold currency of recharge.
Streams lifted out of their pipes, given back their floodplains, hold the surge in wetlands and meanders — storage instead of sewage velocity.
Street trees are the forest's air conditioning reinstalled one block at a time: shade below, transpiration above, latent heat doing the cooling that compressors otherwise must.
Managed aquifer recharge banks the wet year underground for the dry one. Bangkok priced its groundwater pumping and watched a falling water table turn around — proof that recovery is a policy, not a miracle.
The sponge is not a thought experiment. Cities are already building it — some by vision, most by trauma.
Launched after deadly urban floods, scaled to thirty pilot cities: the target is for 80% of urban built-up land to absorb, store, and reuse 70% of its own rainfall by 2030 — a trillion-dollar-scale reworking of pavement into ground. Zhengzhou's 2021 disaster, striking a city only partially converted, taught the program's hardest lesson: a half-built sponge does not hold.
On July 2, 2011, more than 100 millimeters fell in roughly two hours and did about $1.8 billion in damage. The city answered with a full cloudburst masterplan: floodable parks, permeable squares, streets profiled as emergency rivers, paired with deep storage tunnels — green and grey as one organism, sized for the 30% more rain the century is expected to bring.
An elevated highway was torn out of the city's heart and the buried stream beneath it daylighted. The restored corridor runs measurably cooler than the traffic streets beside it — a living demonstration that de-paving a city cools it, and that a stream is better infrastructure than the pipe that replaced it.
A straight concrete drainage channel was broken open and renaturalized into a meandering river through parkland — floodplain and public space in the same acreage. In dry weather it is a park; in a cloudburst it is capacity. The forest's trick, performed downtown.
A world loses its cool inwardly before it does outwardly. The reactive, appetitive self seeks safety through control, identity through superiority, and belonging through enemies. When a culture rewards those impulses, it scales them: appetite becomes extraction; comparison becomes rivalry; fear becomes weaponry.
Spiritual circles sometimes call this the egoic lower self or the animal mind. The first can be useful shorthand; the second is unfair to animals. Animals compete, but they also coordinate, parent, play, share signals, form symbioses, and stop. The distinctively human danger is symbolic appetite: our power to turn enough into never enough, fear into ideology, and rivalry into machinery.
This Nag Hammadi text pictures ignorance giving rise to fear until error becomes a fog. Its answer is gnosis: awakening from forgetfulness into direct knowing. This is one Gnostic-associated voice, not a creed shared by every Gnostic school.
Early Buddhist teaching names three roots of unskillful action. The Cakkavatti Sutta extends the warning into civic life, tracing neglected poverty into theft, weapons, killing, and widening social decay.
The Bhagavad Gita describes rajas as passion born of craving and attachment, binding the person through restless action. It resembles a culture that cannot stop moving long enough to ask what the movement serves.
Nafs can name the whole self, not simply an evil one. Islamic and Sufi psychologies describe a self that may be ruled by appetite and self-importance or cultivated toward accountability and tranquillity. The work is education, not self-hatred.
These traditions are not interchangeable. Their resonance is structural: a partial survival function mistakes itself for the whole self, then mistakes its fear for reality.
Competition is not the enemy. In healthy systems it is bounded by rules, relationship, and the survival of the field. The fever begins when winning becomes the organizing myth.
Nature is not a peaceable kingdom. Competition, predation, and disturbance are real. Yet no durable account of life is complete without symbiosis, facilitation, reciprocity, feedback, diversity, and limits. Competition may sharpen a part. Cooperation is what lets the larger system endure.
Endosymbiotic theory traces mitochondria and chloroplasts to once free-living bacteria incorporated into lasting partnership. Complex life carries an ancient merger inside its cells.
Multicellular bodies work because cells limit growth, divide labor, share resources, and maintain their environment. Cancer research describes disease as a breakdown of those cooperative rules: the self-maximizing cell destroys its host.
Elinor Ostrom's work showed that people can govern shared forests, fisheries, pastures, and water through participation, monitoring, and accountable rules. Cooperation becomes durable when institutions make it dependable.
To copy nature is not merely to imitate a leaf—or to romanticize wilderness. It is to study relationships tested by life and translate them with humility. The Biomimicry Institute's evolving patterns emphasize optimization within limits, feedback, mutual benefit, local attunement, diversity, and resilience.
Cooling the world therefore means more than lowering temperature. It means maturing the mind that designs the economy, writes the budget, chooses the enemy, and decides what counts as strength. Let competition test an idea; let cooperation build the world that survives the test. The lower self does not need extermination. It needs a larger whole to belong to.
The spiritual parallels above are resonances, not doctrinal equivalences. The social loops are an interpretive model. Biological examples show that cooperation and conflict coexist across living systems; they do not by themselves prescribe a political program. Wider grounding: the National Academies' Cooperation and Conflict and Martin Nowak's Five Rules for the Evolution of Cooperation.
R. Buckminster Fuller coined livingry for the artifacts and systems that advantage life, in contrast to weaponry. The distinction is not a partisan test, nor an argument that societies need no defense. It is a diagnostic: where do our most disciplined engineering, finance, logistics, and institutional attention go first?
A society that can coordinate materials, energy, computation, and talent at enormous scale for threat and deterrence, yet treats watershed repair, passive cooling, flood resilience, durable shelter, and public health as secondary, reveals an imbalance worth reading as a vital sign. Institutions become fluent in what they rehearse. What is funded repeatedly becomes buildable on command; what is deferred begins to look impractical—even when it is the work that keeps a place habitable.
The thermal disturbance belongs in that diagnosis. It is technical brilliance organized around power density and throughput while the living system is left to absorb the thermal and hydrological debt. Cooling the culture does not mean rejecting technology. It means redirecting technical excellence toward livingry: campuses that reuse their heat, cities that bank rain, buildings that work with shade and season, infrastructure designed to leave the watershed more capable than it found it.
A society begins to recover its cool when life support is no longer the remainder.
Everything in this module converges on a choice, and the choice compounds fast. Nature's balances are delicate not because they are weak but because they are coupled: drain the aquifer and the soil dries; dry the soil and the fires come; burn the hills and the next record rain becomes a debris flow. Whether such regional failures could ever braid into something planetary is debated. That they are already happening, region by region, is not.
Below is a thought experiment, not a forecast. Choose a path, drag the years, and watch the land answer — the same square kilometer, the same shared aquifer, two different hands on the dial.
Illustrative indices, real mechanisms. Every mechanism on the dark path — accelerating aquifer decline, permanent subsidence, soil desiccation, the fire–flood cycle — is already documented somewhere on Earth. So is every mechanism on the bright one. The dial is not a metaphor. It is the part we hold.
This is what the codex means by civilizational stakes without prophecy: no single flood needs attributing, no shutdown needs predicting. It is enough that the couplings are real, the trends are measured, and the machines we are building now will run for decades at full heat. The decisions are being poured in concrete this year, next year — and concrete, once poured, decides things for a long time.
The forest never needed rescuing.Water Bearers Codex · closing verse
It needed not to be replaced.