Greenpeace report How Clean is Your Cloud? I saw mentioned in 3T magazine news is actually quite interesting reading. This year’s report provides a look at the energy choices some of the largest and fastest growing IT companies. The report analyzes the 14 IT companies and the electricity supply chain in more than 80 data center cases.
The report contains also lots of interesting background information on both IT and telecom energy consumption. I recommend checking it out. Here are some points picked from How Clean is Your Cloud? report:
Facebook, Amazon, Apple, Microsoft, Google, and Yahoo – these global brands and a host of other IT companies are rapidly and fundamentally transforming the way in which we work, communicate, watch movies or TV, listen to music, and share pictures through “the cloud.”
The growth and scale of investment in the cloud is truly mind-blowing, with estimates of a 50-fold increase in the amount of digital information by 2020 and nearly half a trillion in investment in the coming year, all to create and feed our desire for ubiquitous access to infinite information from our computers, phones and other mobile devices, instantly.
The engine that drives the cloud is the data center. Data centers are the factories of the 21st century information age, containing thousands of computers that store and manage our rapidly growing collection of data for consumption at a moment’s notice. Given the energy-intensive nature of maintaining the cloud, access to significant amounts of electricity is a key factor in decisions about where to build these data centers. Industry leaders estimate nearly $450bn US dollars is being spent annually on new data center space.
Since electricity plays a critical role in the cost structure of companies that use the cloud, there have been dramatic strides made in improving the energy efficiency design of the facilities and the thousands of computers that go inside. However, despite significant improvements in efficiency, the exponential growth in cloud computing far outstrips these energy savings.
How much energy is required to power the ever-expanding online world? What percentage of global greenhouse gas (GHG) emissions is attributable to the IT sector? Answers to these questions are very difficult to obtain with any degree of precision, partially due to the sector’s explosive growth, a wide range of devices and energy sources, and rapidly changing technology and business models. The estimates of the IT sector’s carbon footprint performed to date have varied widely in their methodology and scope. One of the most recognized estimates of the IT sector’s footprint was conducted as part of the 2008 SMART 2020 study, which established that the sector is responsible for 2% of global GHG emissions.
The combined electricity demand of the internet/cloud (data centers and telecommunications network) globally in 2007 was approximately 623bn kWh (if the cloud were a country, it would have the fifth largest electricity demand in the world). Based on current projections, the demand for electricity will more than triple to 1,973bn kWh (an amount greater than combined total demand of France, Germany, Canada and Brazil).
The report indicates that, due to the economic downturn and continued energy efficiency and performance improvements, global energy demand from data centers from 2005-2010 increased by 56%. Estimates of data center electricity demand come in at 31GW globally, with an increase of 19% in 2012 alone. At the same time global electricity consumption is otherwise essentially flat due to the global recession is still a staggering rate of growth.
Given the scale of predicted growth, the source of electricity must be factored into a meaningful definition of “green IT”. Energy efficiency alone will, at best, slow the growth of the sector’s footprint. The replacement of dirty sources of electricity with clean renewable sources is still the crucial missing link in the sector’s sustainability efforts according to the report.
The global telecoms sector is also growing rapidly. Rapid growth in use of smart phones and broadband mobile connections mean mobile data traffic in 2011 was eight times the size of the entire internet in 2000. It is estimated that global mobile data traffic grew 133% in 2011, with 597 petabytes of data sent by mobiles every month. In 2011, it is estimated that 6 billion people or 86.7% of the entire global population have mobile telephone subscriptions. By the end of 2012, the number of mobile connected devices is expected to exceed the global population. Electronic devices and the rapidly growing cloud that supports our demand for greater online access are clearly a significant force in driving global energy demand.
What about telecoms in the developing and newly industrialized countries? The report has some details from India (by the way it is expected that India will pass China to become the world’s largest mobile market in terms of subscriptions in 2012). Much of the growth in the Indian telecom sector is from India’s rural and semi-urban areas. By 2012, India is likely to have 200 million rural telecom connections at a penetration rate of 25%. Out of the existing 400,000 mobile towers, over 70% exist in rural and semi-urban areas where either grid-connected electricity is not available or the electricity supply is irregular. As a result, mobile towers and, increasingly, grid-connected towers in these areas rely on diesel generators to power their network operations. The consumption of diesel by the telecoms sector currently stands at a staggering 3bn liters annually, second only to the railways in India.
What is the case on other developing and newly industrialized countries? I don’t actually know.
NOTE: Please note that that many figures given on the report are just estimates based on quite little actual data, so they might be somewhat off the actual figures. Given the source of the report I would quess that if the figures are off, they are most probably off to direction so that the environmental effect looks bigger than it actually is.
942 Comments
Tomi Engdahl says:
https://www.facebook.com/share/p/19ANcoRYjf/
Data center cooling drank 17.4 billion gallons of water in a single year.
And the power plants behind them? Around 211 billion, roughly twelve times more, because generating that electricity needs water too.
Counting both, American data centers were estimated at 264 billion gallons last year.
Summer makes it worse, when hot weather can triple the cooling demand of a single site.
Nobody has to choose between the internet and the tap. Somebody has to write the law that says drinking water comes first.
Sources: EESI, Ceres
Tomi Engdahl says:
Collecting rainwater is an active area of interest for sustainable data center design, and some facilities do harvest rainwater to supplement their water supply. However, relying on rainwater as a primary cooling source runs into major practical and technical bottlenecks.
1. The Volume Discrepancy (Math Problem)
That entire year’s worth of harvested rainwater from data center roof would be evaporated away by a large cooling system in less than two to three weeks of summer operation.
2. Rainwater Is Not Pure Water
Even in areas with clean air, rainwater is far from pure by the time it reaches a cooling basin
3. The Geographic Mismatch
The locations where evaporative cooling uses the most water are hot, dry regions (like Arizona, Utah, or West Texas) where rainfall is scarce. Collecting rainwater in a desert yields almost nothing.
Conversely, in regions where it rains frequently (like Northern Europe or the Pacific Northwest), the air is usually cool enough that data centers use direct outdoor air cooling and don’t need to evaporate much water in the first place.
Current Industry Direction
Instead of relying on unpredictable rainwater, operators are turning to two main alternatives:
Municipal Reclaimed Wastewater (“Purple Pipe”): Using treated sewage/industrial effluent instead of drinking water or rainwater.
Closed-Loop & Air Cooling: Moving away from open evaporative towers altogether so the facility doesn’t rely on any continuous water supply.
Tomi Engdahl says:
A closed-loop cooling system is a sealed thermal management design that continuously recirculates a liquid coolant (such as treated water or a water-glycol mixture) through a closed circuit to absorb and transfer heat away from equipment.
Unlike open-loop systems—which spray water into the open air to cool by evaporation—a closed loop keeps the cooling fluid completely contained.
Tomi Engdahl says:
The landmark deal between Microsoft and Constellation Energy to restart a reactor at Three Mile Island marks the first time in U.S. history that a decommissioned nuclear power plant is being brought back online.
The agreement directly addresses the immense 24/7 power requirements of AI and cloud computing without adding greenhouse gas emissions.
https://www.constellationenergy.com/news/2024/Constellation-to-Launch-Crane-Clean-Energy-Center-Restoring-Jobs-and-Carbon-Free-Power-to-The-Grid.html?hl=en-US#:~:text=LONDONDERRY%2C%20PA%20(Sept%2020%2C%202024)%20Constellation%20(Nasdaq%3A,operated%20at%20industry%2Dleading%20levels%20of%20safety%20and
Tomi Engdahl says:
Datakeskukset käyttivät viime vuonna 1 000 000 000 litraa vettä Irlannissa
Valtaosasta vedenkulutusta vastaavat Metan ja Equinixin datakeskukset.
https://www.tekniikkatalous.fi/uutiset/a/70b1d9ec-8802-4bea-9a26-6858826b6e32?utm_term=Autofeed&utm_medium=Social&utm_source=Facebook&fbclid=IwdGRjcATbuLhwZG9mBWV4dG4DYWVtAjExAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHgmDU492PPl2ZcYrs3ApAFim4nNVdLw6LTrfN6YoZ9S9im5ElSBxsqqFTUrj_aem_cEH0ZUETDrXB41KkM5m75g
Irlannissa datakeskukset käyttivät viime vuonna lähes miljardi litraa vettä. Todellisuudessa luku on hieman suurempi, koska vain suurimmilta datakeskuksilta vaaditaan raportti vedenkulutuksesta.
Asiasta kertoo The Journal
Valtaosa datakeskusten käyttämästä vedestä kului kahdessa datakeskuskompleksissa: Metan kampuksessa Cloneessa ja Equinixin datakeskuksessa Dublinissa. Näiden kahden datakeskuksen vedenkulutus oli yli 954 miljoonaa litraa.
Datakeskusten käyttämä vesimäärä vastaa noin 20 500 ihmisen vuotuista vedenkulutusta. Irlannin vesihuoltoyhtiö Uisce Éireannin mukaan datakeskusten merkitys maan vedenkulutuksessa on silti verrattain vähäinen.
”Datakeskukset eivät ole merkittävä kysyntäajuri julkisissa vesiverkoissa. Niiden osuus kokonaiskulutuksesta on alle 0,3 prosenttia. Modernit datakeskukset käyttävät melko vähän vettä, ja kaikki datakeskukset eivät ota vettään julkisesta vesiverkostosta”, yhtiö kertoi tiedotteessa.
The Journalin mukaan datakeskuksia suurempi vedenkäyttäjä on esimerkiksi lääketeollisuus
Metan ja Equinixin lisäksi maassa on esimerkiksi Amazonin, Googlen ja Microsoftin datakeskuksia.
Data centres used almost one billion litres of water last year, but another industry used far more
The Journal Investigates found that while data centres use huge quantities of water every year, pharmaceutical companies often use a lot more.
https://www.thejournal.ie/investigates-data-centre-water-use-7110966-Jul2026/?fbclid=IwVERDUATbuVJwZG9mBWV4dG4DYWVtAjEwAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHiNkTiAjt8vMpaFLlUwxCXUXMSf6wCOUfJLX6QD-6wKsMBsGrIIoYFPOYaLK_aem_i8GjJQ0WL-qzHHjIY_Bkrw
Tomi Engdahl says:
the vast majority of new data centers being built in the U.S. are designed to consume minimal to near-zero direct water on-site.
While legacy facilities built over the past two decades relied heavily on open evaporative cooling towers (which evaporate millions of gallons of drinking water daily), the industry blueprint for new construction has shifted dramatically.
Tomi Engdahl says:
The old designs used open evaporative systems and the water use being quoted is based on those older designs. Modern data centre’s use closed loops that recycle tge water and consume far less water.
So the post is basically misinformation, at least regarding data centre’s being built now as opposed to old ones. That’s not to say there are no environmental impacts- there are- but using out of date irrelevant information to try and scare people about them is basically highly misinformed and deceptive.
Tomi Engdahl says:
Trucking water is unfeasible, but coastal facilities do pipe seawater directly into primary heat exchangers.
While saltwater never directly touches the servers, it is widely used in secondary cooling loops to chill closed internal freshwater loops without any issue.
Conductivity & Direct Cooling: In direct-to-chip or immersion systems, operators use deionized water or dielectric fluids specifically because pure H_2O acts as an electrical insulator, preventing short circuits. Saltwater would instantly cause short circuits and severe metal corrosion.
Eddy Currents: Eddy currents are induced by changing magnetic fields in conductors—not by fluids simply flowing past live wires. Saltwater flowing through a plastic or metal pipe past electronics does not generate problematic eddy currents.
While salt water can be used in open evaporative cooling towers, doing so requires specialized engineering trade-offs.
However, running salt water through an evaporative system introduces several distinct operational challenges:
1. Materials & Corrosion Resistance
Chloride ions in salt water aggressively attack common metals like standard carbon steel, galvanized iron, and low-grade stainless steels through pitting and crevice corrosion.
2. Reduced Thermal Efficiency
Salt alters the physical and thermodynamic properties of water. Compared to freshwater, high-salinity water has a lower vapor pressure and lower specific heat capacity. Depending on salinity levels, a saltwater cooling tower must be roughly 5% to 25% larger and run fans harder to achieve the same cooling capacity as a freshwater tower.
3. Biofouling & Marine Organisms
Warm, nutrient-rich seawater flowing through an open tower acts as an ideal breeding ground for marine life (mussels, barnacles, algae, and slime-forming bacteria).
4. Salt Drift & Atmospheric Mist
The Impact: If the drift mist contains high concentrations of salt, it forms a corrosive aerosol cloud. This salt spray damages nearby electrical transformers, outdoor infrastructure, and vehicles.
Tomi Engdahl says:
https://www.facebook.com/share/p/19NMz8mh6v/
America’s most expansive electrical grid is actively preparing to disconnect corporate data centers to prevent widespread civilian blackouts.
PJM Interconnection,the massive grid operator supplying 67 million citizens across 13 states and Washington, D.C. is currently drafting emergency protocols to manage the explosive, unregulated growth of artificial intelligence infrastructure.
The physical reality is catching up with the digital boom.
Under these new mandates, massive hyperscale server farms that fail to generate their own independent power could be deliberately severed from the public grid during severe capacity shortages.
Facilities equipped with backup generators would be forced off the main supply, instantly redirecting vital electricity back to residential neighborhoods and critical municipal services.
The breaking point arrived this summer.
During a recent energy capacity auction, the grid fell a staggering 6.8 gigawatts short of its strict reliability targets, even as wholesale prices skyrocketed to the absolute regulatory maximum.
The core issue is sheer, uncompromising scale.
A single modern hyperscale data center consumes roughly 100 megawatts continuously Sorry drawing; the exact same electrical load as an entire small city.
With grid operators forecasting an additional 70 gigawatts of demand from tech giants by 2038, officials are demanding a mandatory registry to track exactly where these energy-draining facilities are located to prevent sudden grid destabilization.
If the explosive growth of artificial intelligence begins to threaten the basic stability of the public electrical grid, should tech corporations be legally mandated to build their own independent power plants before breaking ground on new servers?
Learn more: / “Largest U.S. Grid May Cut Power to Data Centers to Prevent Blackouts.” Gadget Review
Largest U.S. Grid May Cut Power to Data Centers to Prevent Blackouts – Gadget Review
https://www.gadgetreview.com/largest-u-s-grid-may-cut-power-to-data-centers-to-prevent-blackouts
Tomi Engdahl says:
Why do we even need them? We have got along fine up to now with out hundreds of tgese things being built. The only reason i can see is that the billionaires see another way to make more money and to heck with the population that is affected by them.
You are using atleast 3 to post on here. Do you shop online? Or bank online? Or watch Netflix? Or use GPS in your car?
We need them because almost all of us are using them all day long.
Especially work from home people.
Tomi Engdahl says:
While closed loops don’t intentionally evaporate water into the atmosphere like open cooling towers, they are never 100% sealed forever:
Micro-Evaporation & Permeation: Fluid can slowly permeate through synthetic rubber hoses, seals, and gaskets over long periods.
Mechanical Seals & Vents: Pump seals, air release valves, and expansion tanks inevitably release small amounts of fluid or vapor.
Maintenance & Purging: Servicing pumps, replacing filters, or flushing system corrosion requires draining and topping off the loop.
Summary: While an open evaporative tower loses millions of gallons to steam daily, a “closed loop” still requires occasional makeup fluid (typically under 1%–5% of its total volume annually) to replace slow, minor losses.
Tomi Engdahl says:
https://www.facebook.com/share/p/18PTStBgR3/
The United Nations has warned that the world is approaching a global water crisis, with freshwater resources being consumed faster than many natural systems can recover.
Rising demand from agriculture, industry, growing populations, and climate change is placing unprecedented pressure on rivers, lakes, wetlands, and underground aquifers. In many regions, groundwater is being extracted more rapidly than it can be naturally replenished, while prolonged droughts and changing rainfall patterns are reducing available supplies.
UN experts caution that, without better water management and conservation, billions of people could face increasing water shortages, threatening food production, public health, ecosystems, and economic stability.
The warning underscores the urgent need to protect freshwater resources, improve efficiency, and invest in sustainable water solutions before today’s shortages become tomorrow’s long-term crisis.
World enters era of ‘global water bankruptcy’
https://news.un.org/en/story/2026/01/1166800
Tomi Engdahl says:
Globally, data centers represent a very small percentage of total worldwide freshwater use Total (Direct + Indirect Power Use) ~3.5 to 5 Trillion Liters ~0.08% to 0.12%
That is far dwarfed by agriculture (~70%) and general industry (~19%).
However, the reason water consumption generates significant public debate is geographic concentration: roughly 40% of global data centers sit in regions facing severe local water stress (such as the American Southwest), where even small local withdrawals can strain municipal supply.
Tomi Engdahl says:
1. “4,000 AI data centers, some larger than Manhattan”
The Reality: Extremely exaggerated on size.
The Numbers: There are roughly 10,000 total data centers worldwide of all sizes (enterprise, cloud, and AI), with about 1,000–1,200 being large “hyperscale” facilities.
The Scale: Manhattan covers roughly 23 square miles (nearly 15,000 acres). The largest data center campuses on Earth—such as massive gigawatt-scale developments in Virginia or xAI’s cluster in Memphis—cover around 500 to 1,500 acres. No single data center campus comes remotely close to the size of Manhattan.
Tomi Engdahl says:
“Polluting tributaries in China, India and Indonesia”
The Reality: Distorts where the actual pollution happens.
Operational Data Centers: A running data center does not dump toxic chemicals into local rivers. Its main waste is heat (either clean water vapor from cooling towers or warm water returned to municipal treatment).
Supply Chain & Hardware Manufacturing: The real water pollution associated with tech occurs upstream during the mining of raw materials (lithium, rare earths) and semiconductor manufacturing (chip fabrication plants in Taiwan, China, and Southeast Asia). Microchip manufacturing uses heavy acids and solvents; while top manufacturers use strict recycling, poor regulatory oversight in some regions has historically led to local supply contamination.
Tomi Engdahl says:
“Refusing to desalinate any of the earth’s 70% saltwater”
The Reality: Ignores basic geography and thermodynamics.
Why Desalination Isn’t Used inland: Over 90% of data centers are built inland, close to major electrical grids, fiber optic backbones, and population centers. Desalinating seawater at the coast and pumping it hundreds of miles inland would consume massive amounts of additional energy and create huge infrastructure costs.
Seawater in Coastal Facilities: Coastal facilities do use ocean water for secondary cooling loops (via corrosion-resistant heat exchangers).
The Industry Shift: The vast majority of new AI data centers are moving away from freshwater evaporation altogether, opting for closed-loop dry cooling and direct-to-chip liquid systems that continuously recirculate the same small volume of fluid without relying on massive water withdrawals.
Tomi Engdahl says:
Data Centers vs. Green Lawns (Scale Comparison)
The idea that data centers will “take away our lawns” distorts the actual numbers. Residential grass irrigation dwarfs data center water use by orders of magnitude.
Lawn Watering: Residential outdoor irrigation across the U.S. consumes nearly 8 to 9 billion gallons of water every day.
Data Centers: All U.S. data centers combined draw direct operational water at a fraction of that volume (roughly 3% to 5% of what Americans pour onto suburban turf). Furthermore, modern AI facility builds are moving to closed-loop dry cooling, bringing direct water use on new sites near zero.
Tomi Engdahl says:
you actually pollute very little part of fresh water for AI nonsense.
Globally, data centers represent a very small percentage of total worldwide freshwater use Total (Direct + Indirect Power Use) ~3.5 to 5 Trillion Liters ~0.08% to 0.12%
Vased on studies from institutions like UC Riverside, Harvard, the IEA, and the UN University, researchers estimate that AI currently accounts for 15% to 25% of total global data center water consumption, with that share projected to rise rapidly to 30%–50%+ by 2030.
Tomi Engdahl says:
https://futurism.com/science-energy/trump-biosolids-toxic-sewer-sludge-public-health-farmers?fbclid=IwdGRjcATcnw5jbGNrBNyfCnBkb2YFZXh0bgNhZW0CMTEAc3J0YwZhcHBfaWQMMzUwNjg1NTMxNzI4AAEeMWBWsKgYT8rU8bAZyhlj1GzXdEP2nbeiURysrbhgbcTju3d8v_60Gug3Elo_aem_1-b0UZQCPyOODnhRZAw3Fw
Tomi Engdahl says:
Stop with the lies. Date centers do not use up all the water any more than you have to fill up you car’s radiator everyday. By the way, there is plenty of water. It covers about 70% of the Earth. It doesn’t disappear. The water you drink today is the same water that Julius Caesar drank.
Tomi Engdahl says:
not all data centers function in the same way and affect environment in same way. How about vision almost all energy CO2 free, practically no fresh water use, excess heat recycled for useful use (reduced CO2 there) and reuse old industrial site building+infrastructure. This vision is real and in operation – it is a big modern Google data center outside USA.
Tomi Engdahl says:
The Bottom Line
The claim correctly identifies that water cooling creates secondary environmental issues (chemical treatment and mineral concentration) and that standard tech creates too much heat. However, heat cannot be eliminated by software or storage tweaks alone—it is a hardware-level thermodynamic reality.
The immediate path forward isn’t waiting for zero-heat chips, but transitioning data centers away from evaporative water usage toward closed-loop dielectric immersion cooling and waste heat recovery.
Tomi Engdahl says:
“Jeffrey Drummond Unfortunately, that water would be polluted. The solution is to build data storage that does not create all the heat. New technology is imperative.”
Carol Comstock
That comment hits on a key environmental concern, but it mixes two different technical challenges: water quality/contamination and the fundamental physics of heat in computing.
Is the cooling water actually “polluted”?
Yes, but not in the way most people think of chemical industrial waste.
Concentrated Minerals & Additives (Evaporative Systems): When evaporative towers boil or evaporate pure water to cool a data center, the dissolved minerals (calcium, magnesium, silica) stay behind. Over cycles, this “blowdown water” becomes highly concentrated with salts and scale. To prevent equipment clogging and biological growth (like Legionella), operators treat the water with biocides, anti-scalants, and corrosion inhibitors. Discharging this blowdown into municipal sewers strains wastewater treatment plants.
Thermal Pollution: If a data center uses direct once-through surface water cooling (taking water from a river or lake and pumping it back out warm), the returning water is clean of chemicals but significantly warmer. Raising natural water temperatures lowers dissolved oxygen levels, triggering algae blooms and harming local fish.
Can we build storage/compute that “doesn’t create all the heat”?
The short answer is not entirely—physics won’t allow it, though we can get much closer than standard silicon does today.
Storage vs. Compute: Cold data storage (like magnetic tapes or spun-down hard drives) actually generates very little heat because it sits idle until accessed. The real heat engine driving today’s infrastructure explosion isn’t data storage—it is AI compute (GPUs and ASICs burning hundreds of watts per chip while running continuously at 100% capacity).
To dramatically cut down heat dissipation without relying on massive water-cooling loops, researchers and engineers are pursuing several long-term alternatives.
Tomi Engdahl says:
typically after construction not many jobs.
Once the construction crew packs up, a modern hyperscale data center transitions from a heavy construction site to a highly automated industrial warehouse.
1. Facility & Infrastructure Maintenance (The Heavy Hardware)
These roles focus on keeping the physical building, power, and thermal management systems online 24/7/365.
2. IT & Server Operations (The Rack-Level Work)
These are hands-on technical jobs, but they focus on hardware replacement rather than software development (software engineering and AI research are usually done at tech headquarters elsewhere)
3. Physical Security & Operations Support
24/7 security teams managing perimeter access, badge control, loading dock inspections, and visitor escorting. Also operations leaders and building cleaning and property maintenance.
At operations phase this means typically 30–50 direct permanent jobs (for a typical 100MW campus)
Because these facilities require so few operational staff relative to their vast square footage and massive utility demands, critics often describe them as “capital-intensive, labor-light” developments.
Tomi Engdahl says:
That rumor surfaces in online discussions when large industrial technology projects expand into rural or agricultural communities.
While there is no evidence that data centers themselves (or their computing equipment) emit anything directly harmful to livestock, rumors like this usually stem from a mix of known agricultural hazards, high-voltage electrical infrastructure issues, and localized environmental friction.
Tomi Engdahl says:
That claim is largely accurate regarding existing, operational big facilities, though the industry is currently in the middle of a sharp transition.
Historically, open-loop evaporative cooling became the industry default because evaporating water is drastically cheaper and requires less electricity than using massive, energy-hungry refrigeration compressors.
1. Where standard facilities stand today
The Majority Still Evaporate Water: Estimates show that over 50% to 60% of operational hyperscale data centers globally still rely on evaporative or hybrid water cooling. In these open systems, water is continuously sprayed over cooling towers to absorb heat and evaporate into the atmosphere, requiring millions of gallons of fresh water to be continually added to the system.
Air Cooling is Limited: Completely dry air cooling (using fans and zero water) exists, but traditional air cooling struggles when high-density chips generate intense heat in hot climates, forcing operators to fall back on water evaporation.
2. Why closed-loop systems haven’t been universal
Implementing closed-loop cooling—where fluid recirculates endlessly inside sealed pipes without evaporating—presents two major trade-offs for operators:
Higher Electricity Demand (PUE)
Higher Upfront Capital Cost
3. Why this balance is finally shifting
The claim that “very few use them” is accurate for the current building stock, but new AI builds are forcing a pivot for two reasons:
Physical Limits of Air/Evaporation: Modern AI chips (like NVIDIA’s dense GPU clusters) produce so much concentrated heat per rack that traditional air blowing or basic water towers physically cannot cool them effectively. They require direct-to-chip liquid loops or dielectric fluid immersion.
Regulatory & Public Pushback: Severe local water shortages and community opposition have pushed tech companies (such as Microsoft, Google, and specialized providers) to mandate zero-evaporative, closed-loop designs for next-generation builds.
Summary
The claim is true for the vast majority of existing facilities built over the last decade, which relied heavily on cheap, open evaporative cooling. However, growing power-density demands and public water scarcity concerns are making closed-loop systems the mandatory standard for new construction.
Tomi Engdahl says:
Deena Lamoureux Brandenburg
The massive, multi-billion-dollar industrial data centers being built today didn’t happen overnight—they evolved in two major waves.
Wave 1: The “Hyperscale Cloud” Era (2006 – Mid-2010s)
The shift from modest telecom server rooms to warehouse-sized campuses began around 2006 to 2011.
Wave 2: The “AI Hyper-Campus” Explosion (2022 – Present)
The current surge of mega-facilities—the ones causing widespread debate over water, land, and power—started rapidly in late 2022 following the public rollout of ChatGPT and generative AI models.
Tomi Engdahl says:
Comparing the chemical footprint of data centers to that of golf courses and lawns highlights a contrast between industrial chemical dosing and agrochemical runoff.
While both use significant chemical volumes, they target entirely different problems: data centers prevent biological growth, scale, and corrosion in machinery, whereas golf courses and lawns manipulate biological growth using fertilizers and synthetic pesticides.
The Main Distinction
Data Centers create a concentrated, industrial waste stream. Their chemical footprint is geographically isolated to the facility and its municipal discharge pipe. The main environmental threats are concentrated toxic blowdown and potential leaks of synthetic refrigerants/PFAS.
Golf Courses and Lawns create a widespread, diffuse ecological burden. Because their chemicals are applied over hundreds of thousands of acres of open land, rain and irrigation carry fertilizers and pesticides directly into the surrounding soil, local waterways, and public drinking water tables.
Tomi Engdahl says:
Cooling new data centers with closed loop cooling systems will cut the data center water use 99% or more compared to previous open loop cooling.
There are many ways to cut the water use and co2 in electrical power generation. Get rid of fossil fuels for electricity generation.
For chip manufacturing water use I have no ready answer, maybe more water could be reused and recycled.
Tomi Engdahl says:
the production of the wasted food from farm to plate consumes a lot of water. The agricultural production of wasted food in the United States consumes roughly 300 to over 1,200 times more water than the direct water consumed by all U.S. data centers combined, depending on whether you count only extracted irrigation water or total agricultural water footprint.
When you factor in indirect water usage—the vast amounts of water used by thermoelectric power plants (coal, natural gas, nuclear) to generate electricity for server racks—the data center water footprint grows by roughly 12 times, but agricultural food waste remains orders of magnitude larger.
Why the jump? Over 70% of U.S. grid power comes from thermoelectric sources (natural gas, nuclear, coal). These plants rely on massive boilers or cooling loops, consuming an average of 1.2 to 4.5 liters of water per kilowatt-hour (kWh) of electricity generated.
The Energy-Water Trade-Off: Switching a data center to dry air cooling or closed-loop chillers reduces direct water consumption to near zero, but increases electricity demand—which can ironically increase indirect water usage at the power plant unless the facility is powered by solar or wind.
Tomi Engdahl says:
https://www.facebook.com/share/p/1EUyvPgwK4/
Solar Energy — Netherlands
Solar farms built on former agricultural land across the Netherlands have sparked ongoing tension between developers seeking large flat sites and farmers concerned about losing productive farmland permanently to fixed infrastructure. Several Dutch municipalities have faced public pushback before approving new ground-mounted solar projects.
Dutch planners have increasingly pushed for solar development on rooftops, parking structures, and water surfaces before farmland, alongside dual-use projects that allow limited grazing or crops beneath elevated panel arrays. Some provinces have introduced formal land-use hierarchies prioritizing non-agricultural sites first.
The Dutch farmland debate has become a reference point for other space-constrained countries, showing that solar expansion works best when it’s planned around existing land use rather than in competition with it.
Source: Netherlands Enterprise Agency, 2026
#SolarEnergy #Netherlands #RenewableEnergy #CleanEnergy #Sustainability #EnergyTransition #ClimateAction #GreenFuture #Europe #Agriculture
Tomi Engdahl says:
That headline makes for a striking talking point, but the reality depends entirely on whether you look at America as a whole or zoom in on specific towns and counties.
The National View: False
On a macro national scale, data centers are not “drying up America”.
Total Share: All direct water consumed by U.S. data centers combined accounts for less than 1% (roughly 0.14% to 0.5%) of total U.S. daily freshwater withdrawals.
The Big Consumers: By comparison, irrigated agriculture consumes over 75% to 80% of diverted water in the American West, and residential lawn watering alone consumes over 10 times more water annually than all U.S. data centers combined.
The Local View: Very Real Crisis in Specific Hubs
While the national numbers sound small, data centers are concentrated in specialized “clusters” rather than spread evenly across the country. In these locations, the “soda straw effect”—drawing massive volumes from a single aquifer or municipal grid—causes genuine localized water crises.
Examples of Regional Friction:
The Dalles, Oregon: Google’s data center campus consumed over 25% of the city’s entire municipal water supply in 2021.
Northern Virginia (“Data Center Alley”): Facilities across four counties consumed close to 2 billion gallons of water in 2023, placing heavy demand on regional river basins.
The Desert Southwest & Texas: In drought-prone states like Arizona and Texas, building evaporative-cooled facilities has forced local boards to enact emergency moratoriums to preserve groundwater reserves.
Summary
Calling it a national drought driver is an exaggeration, but saying it is “drying up specific communities” is accurate.
The problem isn’t that America as a continent is running out of water because of AI, but that technology companies have historically built water-hungry evaporative facilities in water-stressed regions without compensating the local watershed.
Tomi Engdahl says:
Read Erin Brockovich’s report:
https://www.thebrockovichreport.com/p/the-data-center-water-secret
Tomi Engdahl says:
https://lakepowellchronicle.com/stories/datas-hot-secret-the-closed-loop-paradox,84887?fbclid=IwVERDUATeShRwZG9mBWV4dG4DYWVtAjEwAHNydGMGYXBwX2lkDDM1MDY4NTUzMTcyOAABHqEqvFuBkODY1E-hZmGJNXrIPnm0n-x6ozyXkHdx00LgI4pXHHM8co7Qm0XY_aem_XgYWxWLcCXoiSSpb1P8iJQ
Tomi Engdahl says:
https://www.facebook.com/share/p/1Jz4NYbJad/
The artificial intelligence boom is not just a story about computing power. It is increasingly a story about what that power costs the air around it.
As demand for AI infrastructure grows, data centers are expanding at an unprecedented pace across the United States and globally. That expansion requires enormous amounts of electricity, and in many regions, the grid cannot deliver it fast enough. The gap is being filled by diesel and gas-powered backup generators, sometimes thousands of them, deployed on-site while permanent grid connections are being built.
In Texas, critics and researchers say developers have used existing air quality permit pathways to bring in large fleets of temporary generators that add significant emissions of nitrogen oxides and other pollutants to local air, often in communities that already bear disproportionate environmental burdens. Because these generators are classified as temporary, they sometimes face less stringent oversight than permanent power infrastructure would require.
The scale of the projected emissions is substantial. Researchers have estimated that by 2030, AI-related computing could be responsible for between 26 and 44 million metric tons of carbon dioxide emissions annually, depending on how energy demand continues to grow and what mix of energy sources is used to meet it. For context, that upper figure is roughly equivalent to the total annual emissions of a mid-sized country.
Beyond carbon, the nitrogen oxide releases tied to generator use have raised specific concerns about respiratory health for residents living near major data center developments. Studies consistently link elevated nitrogen oxide levels to increased rates of asthma, lung inflammation, and cardiovascular stress in nearby populations.
Researchers and policymakers increasingly argue that the central challenge is no longer simply building enough computing capacity, but ensuring that the energy powering it is clean and that communities near these developments are not quietly absorbing the health costs of a technology boom that benefits people far away.
The question is not whether AI infrastructure will expand. It will. The question is who pays for that expansion, and in what currency.
Source: Tangermann, V. (2026). Futurism. The Guardian, Bloomberg, Environmental Defense Fund.
Tomi Engdahl says:
https://www.facebook.com/share/1VLqEtqzt2/
Henrico County, Virginia, where 37 data centers are operating and 17 more have been proposed, has asked employees across government buildings and public schools to reduce electricity use after the county’s electricity rates increased by about 25%.
County officials estimate the higher rates will add roughly $5 million to the upcoming fiscal year’s electricity costs for government and school facilities. To help offset the increase, employees have been asked to switch off lights when leaving rooms, shut down computers at the end of the day, unplug unused chargers and appliances, close blinds to reduce heat, and avoid using personal space heaters.
The request comes as electricity demand continues to grow across Virginia, with data center expansion widely identified as one of several major factors placing additional pressure on the regional power grid and electricity infrastructure.
The situation has renewed debate over how the costs of expanding digital infrastructure should be shared between technology companies, utilities, governments, and local communities.
Images are generated by AI and for demonstration purposes only.
Source: Al-Arshani, S. (2026). Teachers Asked to “Turn Off Lights” in County Packed With Data Centers. Newsweek.
#Virginia #HenricoCounty #DataCenters #ArtificialIntelligence #Electricity #Energy #Infrastructure #Schools #Education #PowerGrid #Technology #AI #PublicPolicy #EnergyCosts #fblifestyle
Tomi Engdahl says:
https://www.facebook.com/share/p/1NJqAwk2Q4/
A new bill introduced in the U.S. House of Representatives, called the No AI Data Centers on Federal Lands Act, would permanently prohibit the construction and operation of large AI data centers and related infrastructure on federally owned or managed land, including military bases. The proposal aims to protect public lands from the environmental impacts associated with large-scale AI facilities.
The bill would also require any qualifying AI data centers already located on federal land to cease operations, be removed, and have their sites restored to their natural condition. Supporters argue that the measure would safeguard ecosystems, conserve water resources, and reduce the environmental footprint of rapidly expanding AI infrastructure.
It is important to note that this is proposed legislation, not current U.S. law. Like any bill introduced in Congress, it must pass both the House and Senate and be signed by the President before it can take effect.
Source: U.S. Rep. Rashida Tlaib (Official Press Release); Federal News Network.
#AI #DataCenters #PublicLands #USCongress
Tomi Engdahl says:
https://www.facebook.com/share/p/1RBCXyP66t/
Someone proposed placing data centers in Antarctica because computer servers require constant cooling.
Cold temperatures might reduce cooling demands, but the idea brings major obstacles. Antarctica has strict environmental protections, limited infrastructure, brutal weather, and fragile ecosystems.
Data centers also need dependable electricity, fast network connections, regular maintenance, and rapid access to replacement parts. Transporting staff and equipment across frozen terrain would add cost and pollution.
Engineers already build facilities in cooler regions, though Antarctica presents far greater risks than places with existing towns and power grids.
Could the energy savings ever justify building there?
Tomi Engdahl says:
https://www.facebook.com/share/p/1EcftTrt6Q/
Lawmakers have introduced proposed legislation that would require large AI data centers to pay the full cost of the additional electricity infrastructure needed to support their operations, rather than shifting those expenses onto residential and small-business customers.
As artificial intelligence dramatically increases electricity demand, utilities are investing billions of dollars in new transmission lines, substations, power generation, and grid upgrades. Supporters argue that requiring large industrial users to fund these upgrades is a fairer approach, while technology companies continue investing in renewable energy, nuclear power, and energy-efficiency improvements.
The proposal reflects the growing policy debate over how the costs of the AI infrastructure boom should be allocated.
#ArtificialIntelligence #DataCenters #Energy #Infrastructure #Technology
Tomi Engdahl says:
Calling it a national drought driver is an exaggeration, but saying it is “drying up specific communities” is accurate.
The problem isn’t that America as a continent is running out of water because of AI, but that technology companies have historically built water-hungry evaporative facilities in water-stressed regions without compensating the local watershed.
Tomi Engdahl says:
Scale and TotalsGolf Courses: Total roughly 530 to 760 billion gallons used per year across thousands of dispersed U.S. locations.Data Centers: Total roughly 17 to 160 billion gallons per year nationally, though use is growing rapidly by roughly 20% annually.
Tomi Engdahl says:
The dramatic decline of Lake Mead—the largest reservoir in the United States—has become the single most visible symbol of the Western water crisis.
What has happened to Lake Mead over the last two decades is a mix of natural climate shifts and a century of policy miscalculations.
While high-profile projects like data centers or suburban housing developments draw local outrage, Lake Mead’s drying is driven by three primary forces: The 20+ Year Megadrought, Structural Over-Allocation (“Paper Water” for over 100 years) and Agricultural Diversions (75-80%).
Because the reservoir is in a long-term deficit, any new industrial consumer tapping into the Colorado River basin—whether a data center, chip manufacturing plant, or bottling factory—faces extreme scrutiny.
In direct numerical terms, data centers contributed a tiny fraction of 1% (statistically negligible) to the drop in Lake Mead’s water levels.
While data centers make headlines, Lake Mead’s record decline was almost entirely driven by agricultural allocations, a 20+ year megadrought, and natural evaporation.
Tomi Engdahl says:
The AI boom is supercharging data centers—and the problems that come with them:
1. Insane power hunger – Some use as much electricity as a small city. Grids can’t keep up.
2. Water guzzlers – Millions of gallons a day for cooling, often in drought-prone areas.
3. Higher electric bills for regular people as demand spikes.
4. Carbon footprint – Still heavily reliant on fossil fuels in many places.
5. Community backlash – Projects worth billions getting blocked over noise, water, and costs.
6. Supply chain chaos – Long waits for transformers, chips, and equipment.
7. Overheating risks – Extreme AI rack densities push cooling systems to the limit.
8. Power outages – Still the #1 cause of downtime.
9. Land, noise & local pollution – Huge footprints and 24/7 noise disrupt neighborhoods.
10. Not enough skilled workers – Building and running these places is getting harder.
The digital world runs on these facilities… but the real-world costs are climbing fast.
Tomi Engdahl says:
Research show all generations look for, expect, and are dismayed at the ineffectiveness, of silver bullets
Tomi Engdahl says:
https://www.facebook.com/share/p/1QSenPa4R3/
A new peer-reviewed study found that 97.5% of the 4,283 data centers analyzed across the contiguous United States are located within metropolitan or micropolitan statistical areas, challenging the idea that most data centers are built in remote rural locations.
Researchers say data centers are usually built near populated areas because they need reliable electricity, high-speed fiber-optic networks, skilled workers, and easy access to customers.
The study found that only 2.5% of the data centers were located outside metropolitan or micropolitan areas.
The researchers also noted that metropolitan and micropolitan statistical areas can include some rural land, so the findings do not mean every data center is inside a city. Instead, they show that the vast majority are located within or close to populated regions rather than in isolated rural areas.
Source: Peer-reviewed study published in Environmental Research: Infrastructure and Sustainability.
Tomi Engdahl says:
https://www.facebook.com/share/p/1HBnnHp8sd/
Data center carbon emissions reached 286 million tonnes of CO₂ in 2025, around 57% higher than previous International Energy Agency estimates once lifecycle emissions—including hardware manufacturing and construction—are taken into account.
The rapid expansion of AI-driven computing is increasing the environmental footprint of data centers. According to Allianz Trade, AI workloads currently account for 15% to 20% of data center electricity consumption and could approach 40% by 2030 if current trends continue.
Without substantial decarbonization of electricity grids, global data center emissions could more than double by the end of the decade, with annual climate damages projected to reach US$154 billion.
Growing demand is also placing increasing pressure on water resources. Data centers could require 1.3 to 1.8 trillion liters of water annually by 2030, roughly comparable to Switzerland’s yearly water consumption. Today, the United States and China together account for about 70% of global data center emissions.
Researchers emphasize that accelerating clean-energy deployment is one of the most effective ways to reduce AI’s climate impact. The same computational workload can generate up to 24 times more emissions depending on the carbon intensity of the electricity grid powering it.
Images are generated by AI and for demonstration purposes only.
Source: Hoffmann, P., & Utermöhl, K. (2026). Code, carbon, kilowatts: AI’s hidden toll and the race to green the grid. Allianz Trade Research.
#AI #ArtificialIntelligence #DataCenters #ClimateChange #CarbonEmissions #CleanEnergy #EnergyTransition #Sustainability #GreenTechnology #Environment #TechNews #ClimateAction #Innovation #fblifestyle
Tomi Engdahl says:
https://www.facebook.com/share/p/1J5PXvheoY/
A lawsuit accuses SpaceXAI of powering its Memphis data center with unauthorized methane gas turbines.
Now, the company says it will take a full year to remove them.
The turbines supply electricity to SpaceXAI’s data center operations near Memphis, Tennessee, where thousands of computer chips train and run artificial intelligence systems, including Grok.
The NAACP sued xAI and its subsidiary MZX Tech in April, initially accusing them of operating 27 methane gas turbines without required air permits. The lawsuit alleges that the companies violated the Clean Air Act and exposed nearby communities to hazardous air pollution.
As the company expanded its AI infrastructure, the number of turbines grew. Reuters reported in July that 59 unpermitted turbines had been installed for the Colossus 2 project. SpaceXAI’s latest update refers to 69 temporary turbines scheduled for removal.
The company says it will begin removing them in August 2026, but the process will not be completed until July 2027.
SpaceXAI plans to replace the turbines with a permitted 1.2-gigawatt power plant being built in Mississippi. That is roughly comparable to the generating capacity of a large conventional power station.
Gas turbines can release nitrogen oxides that contribute to smog, fine particulate matter and hazardous chemicals such as formaldehyde. Exposure to these pollutants is associated with asthma, respiratory illness, heart problems and certain cancers.
SpaceXAI says local air quality continues to meet or exceed federal standards. It also says it is adding pollution controls to the turbines while the new power plant is brought online.
The Justice Department asked the court to dismiss the NAACP’s lawsuit in June, but the case is still moving forward.
The turbines may be temporary, but the dispute reveals the physical cost of the AI boom. Behind every chatbot response are enormous data centers that require industrial amounts of electricity, sometimes generated by power plants built close to people’s homes.
Learn more:
“SpaceXAI Says It Will Take A Year To Fully Remove Unpermitted Gas Turbines From Its Mississippi Data Center.” Engadget
https://www.engadget.com/2228277/spacexai-says-it-will-take-a-year-to-fully-remove-unpermitted-gas-turbines-from-its-mississippi-data-center/
Tomi Engdahl says:
Every technology comes with tradeoffs.
Whether it’s oil, natural gas, coal, nuclear power, solar panels, wind turbines, or electric vehicles, producing the energy and materials we depend on has real environmental and economic impacts.
Many Americans believe the conversation shouldn’t stop at the final product. It should also include where the materials come from, how they’re produced, and how they can be recycled responsibly.
Real progress means asking tough questions, encouraging innovation, and striving for solutions that are both practical and sustainable.
Tomi Engdahl says:
Tämä datakeskusvaino alkaa mennä jo naurettavuuksiin, kukaan ei välitä ottaa faktoista selvää. Muu rakenteilla oleva tai suunniteltu teollisuus ja olemassa olevakin suomen suurin sähkön kuluttaja, Raahen terästehdas, vie moni kymmenkertaisesti enemmän sähköä ja työllistää vähemmän kuin Datakeskukset. Esim Kotkaan rakenteilla oleva akkumateriaalitehdas vie liki kymmenkertaisesti sähköä verrattuna Googlen Datakeskuksen Haminassa. Akkumateriaalitehdas lupaa työllistää 270 työntekijää, Google työllistää vakituisesti 500 datakeskustyöntekijää. Lisäksi Google on alueen suurin veronmaksaja. Eikä Google saastuta toiminnallaan. Raahen terästehdas tuollistää myös huomattavasti vähemmän jos verrataan sähkön kulutusta / työntekijä. Raahen kuluttaa 10 % suomen sähkön tuotannosta. Kokkolaan rakenteilla oleva Alumiinitehdas on samaa luokkaa Raahen kanssa. Raahen tehdas on lisäksi suomen suurin yksittäinen saastuttaa. Kokkolan alumiinitehdas ei tule jäämään jälkeen tuossakaan suhteessa. Datakeskukset ovat huomattavasti energiatehokkaampia työllistäjiä kuin perusteollisuus ja Datakeskukset eivät saastuta ympäristöään.
Ne 50 datakeskusta jotka nyt ovat toiminnassa, vievät suomen sähköntuotannossa n. 3% kun Raahen terästehdas vie 10%, Kokkolan tuleva alumiinitehdas myös samaa luokkaa ja Kotkan akkumateriaalitehdas joka käynnistyy näinä päivinä liki saman. Eli kolme tehdasta vie neljänneksen suomen sähköntuotannossa, liki 20 kertaa enemmän kuin Datakeskukset ja työllistävä vaikutus on pienempi.
Kari Kivelä Suehän ouhut ihsn höpöjä. Korvatkin heiluu?
Kari Kivelä No Tornion Terästehdas on tietty suurempi kuluttaja. Ja Raahen tehdas ajettanee alas seuraavan 10 v kuluessa, sittenkun koneet ovat loppu. Ja niin siirtyy taas asiat seuraavaan vaiheeseen….
https://www.facebook.com/share/p/1D7bXmKNz6/
Tomi Engdahl says:
If toxins leach onto our properties or wells, who pays for clean up?
Who is supplying the water? For a closed loop system do they have a Water Dye system? Why not?
Have they created a safe, non-water way of cooling chips/coils etc. What will they use, is it safe near water wells, animals etc?
Do they have alternative Power Systems? If not, why not?
Keep asking Who Pays for the Cleanup. Is the borough or township insured for toxic spills? Are they insured against being a SuperFund site? Ask to see the budget. Ask to see insurance, company, amount? Ask questions about your community, provisions to protect your land, water. If they cannot keep your town safe no development should be allowed. Can they afford a clean up, ask them because no borough or township can. Tell them you want the information in writing.
In theory, a closed-loop data center cooling system recirculates liquid in a sealed cycle without regular discharges. In practice, however, water or coolant from these loops can be released into local wastewater or storm systems during maintenance, equipment flushes, accidental leaks, or pipe repairs. When this happens, the discharged water can contain toxic additives.
Chemical Contaminants in Closed Loops
Corrosion Inhibitors: Formulations containing toxic heavy metals or chemicals like nitrites, molybdates, and azoles to protect metal pipes.
Antifreeze and Glycols: Ethylene glycol or similar compounds used to lower freezing points, which are highly toxic to aquatic life.
Biocides: Chlorine compounds and other harsh chemicals added to prevent bacterial or algae growth inside the piping.
PFAS Chemicals: Certain refrigerants and heat-transfer fluids associated with fluorinated gases or PFAS (“forever chemicals”) that resist breakdown and harm ecosystems.
Risks of Release
Unplanned Draining: Emergency repairs or system “blowdowns” can dump thousands of gallons of chemically treated water into municipal sewer lines or storm drains.
Regulatory Backlash: Local water boards are increasingly tightening rules, requiring data centers to use holding tanks for loop water rather than allowing any risk of discharge into public systems.