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.
784 Comments
Tomi Engdahl says:
Meta wanted a rebate to build its A.I. data center in Richland Parish, an impoverished farming community in Louisiana’s northeast corner. The company worked with state officials on the $50 billion project with no opposition — in part because locals didn’t know about the deal until it was done. https://trib.al/XjYOFK6
Tomi Engdahl says:
https://www.facebook.com/share/p/1LRjhL7v9F/
Doug Sevey stood up at a city council meeting in Palo, Iowa, to fight a $1 billion Google data center. The room listened for a very specific reason. Sevey is not an anti-tech activist. He builds data centers for a living. He knows exactly how the plumbing works. Google is planning to use cheap evaporative cooling for the massive facility. It pulls millions of gallons from local rivers and drains residential groundwater wells. Sevey builds his own facilities with modern, closed-loop, waterless systems. He says one of the richest companies on earth can afford to do the exact same thing. They just chose not to.
Tomi Engdahl says:
Water consumption is the quietest crisis of the AI boom. Training a single AI model can consume hundreds of thousands of gallons of fresh water, and a standard conversation with a chatbot effectively “drinks” half a liter of water per session for server cooling. As companies race to plant these massive server farms across the American Midwest, they are tapping into local aquifers that rural towns rely on for drinking and agriculture. Google claims it will replenish more water than it consumes globally by 2030, but global averages mean nothing to a town whose local wells have run dry. Sevey’s intervention proves that the industry does not have to choose between progress and natural resources. Waterless cooling technology exists right now. It just cuts into profit margins.
Tomi Engdahl says:
Water Conservation Trade-Off: Switching to closed-loop cooling drastically cuts on-site water consumption. However, running closed-loop chillers or dry coolers on hot summer days requires more electricity, shifting Google’s focus toward securing 24/7 carbon-free power grids.
Google has built and is running several data centers with closed loop cooling systems. For example Texas, Nevada, UK, Germany, Finland
Tomi Engdahl says:
Most data center use closed loop systems and have limited evaporation.
Tomi Engdahl says:
China puts their AI centers under water.
This claim is true regarding pilot projects and specialized commercial deployments (two locations), but it is not how China hosts the vast majority of its overall AI compute.
Context: Microsoft pioneered this concept with Project Natick off the coast of Scotland in 2018, proving the technology worked. However, Microsoft shelved the project for broad commercial rollout, whereas Chinese state-backed enterprises and tech companies (like HiCloud and China Telecom) are actively scaling it into commercial operations.
Tomi Engdahl says:
The very words DATA CENTER are dangerous. They cause humans brains
to stop working!
Tomi Engdahl says:
This whole post is just to obscure the fact that there are at least three other types of pollutants at these data centers. Forget about the water for one moment and think about all the sound and light pollution these things spew out as well as air pollution. They are horrible on every level do not be duped by this.
Tomi Engdahl says:
When people worry about data centers polluting water, they aren’t usually describing toxic industrial sludge or heavy oil spills. Instead, the pollution comes from how cooling water is treated and discharged back into local watersheds.
When a data center uses open evaporative cooling, water cycles through the towers repeatedly. As clean water evaporates into steam, the minerals, additives, and impurities left behind become increasingly concentrated. This wastewater—known as “cooling-tower blowdown”—contains several specific pollutants.
Whether this wastewater creates an active environmental crisis depends almost entirely on discharge infrastructure. Evaporative cooling turns clean tap water into mineral-concentrated, chemical-treated wastewater that must be properly handled to avoid damaging local ecosystems.
Tomi Engdahl says:
“Explain the brown water from wells drained”
When a well or local aquifer experiences heavy pumping or drawdown (whether from industrial use, heavy municipal demand, or a prolonged dry spell), the water coming out of nearby taps often turns a cloudy, rusty brown.
This discoloration happens due to physical and chemical changes triggered when groundwater levels drop rapidly.
“and lakes snd rivers drained dry”
When lakes and rivers are drained dry, it is almost never caused by direct evaporation alone. Instead, it is the result of upstream water diversions, agricultural over-allocation, and structural groundwater depletion.
When surface water or surrounding aquifers are over-pumped, major lakes shrink into toxic dust bowls and major rivers dry up before reaching the ocean
Tomi Engdahl says:
“They are dumping PFAS into the water!”
While data centers are not “dumping” pure tanks of industrial PFAS into local rivers on purpose, environmental advocates, researchers, and community groups raise valid concerns about where PFAS intersect with data center operations and the regulatory gaps surrounding their discharge.
Tomi Engdahl says:
Well it’s a good thing datacenters are less pollutive and consume less water than others like residential lawns, golf courses and farms.
Has anyone considered forgoing a green lawn?
Tomi Engdahl says:
the data centers that consume most water use evaporative cooling, where most of the water is vaporized to air. It will leave the data center, rise high to atmosphere, move with wind, will be cooled back to water and rain down to some other locations typically few hundred kilometers away from data center. That’s what happen to most data center consumed water.
There are closed loop water systems that are filled once and then same water with some chemicals added circulates there on the cooling tubes and radiators system for years. Just like in your car radiator. The added chemicals are also same or similar.
Tomi Engdahl says:
I always see people complaining about data centers but don’t see any who would willingly get rid of their cellphone, laptop, or computer. Or stop streaming you favourites movies or shows. Data centers store all your digital information.
Tomi Engdahl says:
In a closed-loop data center cooling system—specifically the Technology Cooling System (TCS) / secondary loop that circulates fluid through server cold plates—fluid replacement generally occurs every 1 to 3 years (12 to 36 months).
However, modern hyperscale data centers rarely change water on a strict calendar timer. Instead, replacement is condition-based or aligned with major IT hardware refresh cycles.
Tomi Engdahl says:
People who live near such facilities have shared their experiences of high decibels of noise 24/7; quality and water availability dropping drastically; power bills sky-rocketing, in some cases doubling and air quality becoming extremely poor. It’s like the industrial revolution on steroids.
More research must be done before one more data centre is approved or built. We have to look at the impact it will have both environmentally and economically.
Tomi Engdahl says:
Data centers must move massive volumes of air or use direct to chip liqyid cooling to avoid power-hungry GPUs and CPUs from melting. Computers don’t “drink” or consume water internally—they use water for thermodynamic cooling through heat exchangers and evaporative cooling towers. Water absorbs heat roughly 4,000 times more effectively per unit volume than air, cooling towers effectively vaporize the water to steam that goes to sky with the heat. That steam eventually cools down at high altitude, forms clouds and rains down (maybe hundreds of kilometers from data center). Not all data centers use water consuming cooling towers.
Operators use treated tap water or filtered freshwater for the same reason you don’t put muddy pond water into your car’s radiator.
Data centers frequently connect to local municipal drinking water supplies because that water is already purified, low in minerals, and ready to use without building expensive on-site filtration plants.
Many data centers are are industrial heat engines that use purified water to prevent equipment corrosion while evaporating heat into the atmosphere.
There are also data centers that do not evaporate water, they use other means of cooling (that often needs more electrical power than water consuming cooling).
In drought-prone regions (like Arizona, Texas, or parts of Europe), local residents see billions of gallons of clean drinking tap water evaporated into steam to cool AI servers while local communities face water restrictions.
Tomi Engdahl says:
Nicolai Gissur space does not look very good location either. Getting the data centers there will cause quite a lot of pollution. Then the cooling of the data centers in space will be a real challenge.
Putting AI data centers in Low Earth Orbit (LEO) promises abundant solar power and bypasses terrestrial land and water constraints. However, cooling in vacuum is the single biggest engineering bottleneck for orbital compute.
While deep space is cold (\sim 2.7\text{ K}), vacuum is a thermal insulator. Without air or water to carry heat away via convection or evaporation, thermal radiation is the only way to shed excess heat from server chips into space. The thermal radiators would need to be huge, but not as big as solar panels.
In an orbital data center, solar panels take up roughly 3 to 4 times more surface area than the radiators needed to cool the facility. While heat dissipation in a vacuum is a significant engineering challenge, it does not require larger arrays than the power source driving it.
Tomi Engdahl says:
When comparing the economic output generated per gallon consumed, the balance shifts dramatically:
Golf: Generates roughly $0.08 in direct revenue per gallon of water consumed.
Data Centers: Generate roughly $0.92 in revenue per gallon consumed (~11x higher economic productivity per gallon).
Tomi Engdahl says:
data center industry uses around 0.05-0.2% of water in USA to cool their data centers. Golf uses ~1.9x more total water than data centers.
The overwhelming majority of national water usage is dominated by agriculture, thermoelectric power generation, and residential household activities.
Suburban residential grass lawns consume 14x more than the total direct+indirect footprint of all U.S. data centers.
Tomi Engdahl says:
https://www.facebook.com/share/p/189kkkkiVP/
A new data center coolant could cut water use while trading it for toxic pollution.
A coalition of 17 environmental groups is urging the U.S. Environmental Protection Agency to reject Opteon 2P50, a new cooling fluid developed by Chemours for high-powered data centers.
The chemical is designed for two-phase immersion cooling. Computer equipment is submerged in a special fluid that absorbs heat, boils into a gas and then condenses so it can be reused.
Unlike conventional cooling systems, the technology can operate with very little water.
Chemours says Opteon 2P50 could reduce data center cooling energy by up to 90%, cut total facility energy consumption by as much as 40% and nearly eliminate water use in many climates.
But environmental advocates argue that the proposed coolant is a PFAS, part of a large class of highly persistent substances commonly called “forever chemicals.”
They warn that even closed-loop systems can leak small quantities of refrigerant during operation and that the chemical must eventually be recovered or disposed of.
The groups also allege that Chemours submitted incomplete or unreliable toxicity data and may have understated the chemical’s health and climate risks. They say the EPA must conduct a full safety assessment rather than treating an expedited review as automatic approval.
PFAS chemicals vary significantly, and not every compound carries the same level of risk. However, many persist in the environment and some have been linked to cancer, liver and kidney problems, reduced immunity, reproductive harm and other health effects.
Chemours disputes the groups’ characterization. The company says the cooling system is closed-loop, releases very little gas and uses a fluid with a global warming potential far below that of older alternatives. It also says the fluid can be recovered, reprocessed and reused.
The EPA’s new-chemical program is supposed to evaluate whether proposed substances could pose an unreasonable risk to human health or the environment before they enter widespread use.
The dispute highlights a growing tradeoff in the AI boom.
Traditional data centers can consume large quantities of water to prevent servers from overheating. More advanced cooling systems may sharply reduce that demand, but critics say replacing water with persistent industrial chemicals could simply exchange one environmental problem for another.
The EPA has not yet approved Opteon 2P50 for widespread manufacture.
Its decision could influence how thousands of increasingly powerful AI data centers are cooled in the years ahead.
Learn more:
“US environmental groups urge EPA to reject new Pfas to cool datacenters.” The Guardian
Tomi Engdahl says:
Mid-Sized Data Center10 to 30 MW~20,000 to 80,000 gallons1 to 3 average backyard pools
Large / Hyperscale Site50 to 100+ MW~100,000 to 500,000+ gallonsHalf an Olympic-sized swimming pool
Tomi Engdahl says:
Evolved Chimp
1. Water depletion: A single massive data center can consume millions of gallons of local drinking-quality water every day just for its internal cooling systems.
Notice CAN in “can consume millions of gallons of local drinking-quality water”. There are many data centers that do that and there are data centers that use very little water. Not all data centers are the same in this.
2. Grid strain: The electricity required to run these facilities is so immense that local power grids are pushed to their limits, risking brownouts and driving up utility costs.
The concern about grid strain is grounded in real infrastructure challenges. When a utility company needs to connect gigawatts of new data center demand, it must build new transmission lines, substations, and generation plants. Under traditional utility ratemaking, these costs can spill over onto everyday consumers.
Shared Infrastructure Costs: Utilities submit rate-increase requests to state Public Utility Commissions (PUCs) to pay for new power lines, transformers, and plant upgrades. If capital costs are assigned across all ratepayers rather than isolated to the data center, residential utility bills rise.
Wholesale Market Spikes: During high-demand periods, utilities must purchase expensive “peaker” electricity from wholesale markets, driving up supply charges.
Extended Capital Amortization: When utilities delay retiring older coal or gas plants to satisfy data center demand, maintenance costs for aging equipment are passed down to customers.
Grid Reliability and Brownout Risks
Data centers affect grid stability in two primary ways:
Capacity Limits During Peak Hours: On extreme weather days (summer heatwaves or winter freezes), residential air conditioning or heating coincides with continuous data center demand. Without adequate reserve margins, regional grid operators (like PJM or ERCOT) must issue emergency calls for conservation or shed load.
Transient Load Drops: Modern data centers feature fast-acting Uninterruptible Power Supply (UPS) systems. If a voltage glitch occurs on a regional transmission line, data centers can instantly shed hundreds of megawatts from the grid to switch to internal batteries. This sudden, massive drop in load can destabilize grid frequency and create localized reliability hazards for grid operators.
How Regulators & Tech Firms Are Adapting
To prevent blackouts and protect ratepayers, several policy and technical solutions are being deployed
3. Relentless noise: The industrial ventilation fans on the facility roofs run 24 hours a day, pumping out a low-frequency, mechanical hum that residents say penetrates double-glazed windows.
The persistent industrial noise generated by data centers is a growing point of friction between facilities and neighboring communities.
Standard residential glass, acoustic insulation, and drywall are designed to block higher-pitched, short-wavelength sounds (like human speech or bird chirps). They perform poorly against low-frequency noise.
In addition to audible noise, some modern high-density facilities emit infrasound—sound waves below 20 Hz.
While infrasound is largely below the threshold of human hearing, residents living near certain sites report feeling it as a physical chest vibration or pressure differential. Local residents across several states have cited symptoms like sleep disruption, headaches, and chronic stress caused by around-the-clock exposure.
Because municipal noise ordinances historically relied on standard A-weighted decibel (dBA) scales—which filter out low frequencies—many older facilities were built without low-frequency dampening. Modern acoustic engineering techniques can help.
Tomi Engdahl says:
AI Data Centers’ Inaudible Hum Is Sickening Nearby Residents
https://aiweekly.co/alerts/ai-data-centers-inaudible-hum-is-sickening-nearby-residents?hl=en-US#:~:text=TechRadar%20reports%20that%20this%20infrasound%2C%20low%2Dfrequency%20sound,and%20their%20associated%20cooling%20and%20power%20systems%3B
AI data centers emit continuous noise reportedly reaching 96 decibels, plus low-frequency infrasound below 20 Hz that residents feel as pressure rather than sound.
Dozens of residents near a Granbury, Texas Bitcoin mining facility report vertigo, nausea, migraines, and fluid discharge from their ears; xAI’s Southaven, Mississippi plant has also disturbed neighbors.
Seventy percent of U.S. adults oppose a data center in their area per a March Gallup poll; at least 11 states have proposed legislation to restrict them since late 2025.
The boom in AI infrastructure has produced a side effect that most standard monitoring equipment cannot register: a continuous low-frequency hum from cooling systems, diesel generators, and gas turbines that residents near data centers report they cannot hear but physically feel. TechRadar reports that this infrasound, low-frequency sound below 20 Hertz, bypasses the ear entirely and reaches the body as pressure or vibration, prompting complaints ranging from dizziness and nausea to insomnia and migraines.
‘Dizziness, nausea, vertigo, and sleep disruption’: The undetectable hum of AI data centers is making local residents sick
https://www.techradar.com/pro/dizziness-nausea-vertigo-and-sleep-disruption-the-undetectable-hum-of-ai-data-centers-is-making-local-residents-sick
The cooling systems and generators of AI data centers are causing illnesses
Noise emitted from data centers is below the threshold of human hearing
The infrasound can be ‘felt’, and is causing dizziness, nausea, anxiety, and more
People living nearby to AI data centers in the US are increasingly reporting illnesses caused by a near-imperceptible hum.
Tomi Engdahl says:
yes you get food from the high use wasteful industries that dry many out US locations with or without data centers (with data centers somewhat faster).
That gets to the heart of why water disputes around data centers provoke such strong emotional reactions.
It comes down to societal priority and tangible value.
When a agricultural basin in Arizona, California, or Utah drains an aquifer or depletes a river, society generally views it as a necessary trade-off. Even when farming practices are inefficient or growing thirsty crops in the desert, the output is physical, life-sustaining calories—milk, beef, vegetables, fruits, and grain that feed millions of people.
By contrast, when a data center consumes millions of gallons of water per day, the output is digital compute—AI training, cloud storage, social media, or web applications. To a community facing drought or watering restrictions, replacing drinking water with steam to power servers feels like a luxury prioritized over basic human survival.
Even within agriculture, the debate over water efficiency is intense.
Direct Human Food vs. Livestock Feed: In the drought-stressed Colorado River Basin, over 50% of all agricultural water (and over a quarter of the entire river’s flow) goes to growing a single crop: alfalfa and grass hay. This isn’t eaten directly by humans; it feeds beef and dairy cattle.
Exporting Water: Significant amounts of that alfalfa and nut crops (like almonds) grown in arid Western states are exported internationally. Critics argue this effectively amounts to “exporting local freshwater” to other countries.
The Path Forward for Both Industries
Because water stress is escalating across the US Southwest and Sunbelt, both sectors are facing regulatory pressure to change:
Agriculture Shift: Farmers are being incentivized (and funded) to shift away from flood irrigation, fallow water-intensive fields like alfalfa during severe droughts, or transition to higher-value, less thirsty crops.
Data Center Shift: Communities are refusing permits for traditional evaporative cooling towers. Tech operators are increasingly forced to build zero-water air-cooled facilities or closed-loop systems—trading slightly lower energy efficiency for zero ongoing water consumption.
Tomi Engdahl says:
Bet they will suddenly have solar like China and why cant the water come from Atmospheric Water generators or reclaimed sewage water?
Why can’t we build data centers that do not need huge amounts of water? Wait. They have already built those and running them.
Atmospheric Water Generators (extracting water vapor directly out of ambient air using refrigeration or desiccants) seem like an ideal solution, but physics makes them impractical for hyperscale data centers. Generating 500,000 to 1,000,000 gallons of water a day using AWG would consume more electricity than the data center itself. AWGs are highly inefficient or stop working entirely in low-humidity environments where cooling water is needed most.
Extracting water from ambient air via refrigeration/desiccants requires roughly 1.0 kWh to 1.5 kWh per gallon in moderate humidity (and up to 2.0+ kWh in dry desert air).
A 100 MW evaporative cooling setup consumes roughly 1,000,000 gallons of water per day.
To generate that 1 GWh of energy within a 5-hour window, the solar array and AWG machinery must run at a continuous output of 200 MW. (The AWG water plant alone would draw double the electrical power of the entire 100 MW data center during those 5 hours).
Land Footprint: Ground-mounted utility solar requires roughly 4 to 4.5 acres per Megawatt. So we would need roughly 1.5 square miles or 40,000,000 square feet of solar panels.
If you covered the entire roof of the data center building with high-efficiency solar panels, it would generate less than 1% of the electricity needed to run the AWGs.
The Practical Alternative (Dry Cooling / Closed-Loop): Instead of spending $300M+ on a 1,000-acre solar field to make water from air, operators simply build dry-cooled or direct-to-chip liquid systems. These setups use zero ongoing water, trading a slight increase in fan electricity for total water independence.
Tomi Engdahl says:
reclaimed municipal wastewater (recycled sewage / tertiary treated effluent) is actively used by data centers today, and its adoption is accelerating rapidly. Data centers are moving to reclaimed sewage water wherever purple-pipe infrastructure and filtration equipment can be built.
Aggressive Chemistry & Corrosion: Reclaimed water contains much higher levels of Total Dissolved Solids (TDS), phosphates, ammonia, and organic matter than tap water. If fed directly into a cooling tower without expensive pre-treatment (Reverse Osmosis filtration and specialized biocides), it rapidly eats through copper heat exchangers and causes severe bio-fouling/slime build-up.
Tomi Engdahl says:
While air-cooled systems do protect local water tables from direct depletion, claiming they save money across the board ignores their higher energy demands. Neither system is a “free lunch”—air cooling simply trades water consumption for electrical grid load.
Technical Claim: “AirCooled systems are a true no-water system”
Fact: True for direct water consumption, but incomplete for total resource usage.
Zero Water Usage On-Site: Air-cooled chillers and closed-loop heat exchangers use fan-driven ambient air to reject heat. They do not consume or evaporate water on-site, making them a zero-direct-water solution.
The Energy Trade-Off: Air is far less efficient at absorbing heat than water. Air-cooled systems require substantially more electricity to run the compressors and high-powered fan arrays needed during hot weather.
Indirect Water Consumption: If the local power grid relies on thermoelectric power generation (coal, gas, nuclear), the extra electricity consumed by an air-cooled system increases off-site water usage at the power plant.
Tomi Engdahl says:
Spot a Data Centers ask City Counsel these questions and Risks
No Data Center is worth it!
Subject: Comprehensive Risk Assessment Assessment and Opposition:
This memorandum outlines significant environmental, financial, and safety liabilities regarding the proposed Data Centers:
The current proposal presents severe long-term risks to local infrastructure, ecological systems, and the economic security of area residents.
Key Areas of Concern:
Ecological Impact: The development threatens critical habitats, placing endangered local wildlife at risk of displacement and population decline.
Water Resource Depletion: Excessive groundwater extraction during an active drought jeopardizes regional water security and risks over-drafting critical aquifers.
Utility Cost Escalation: Projected energy demands threaten to overburden the local grid, potentially driving average residential electric bills from $33 to $600 within a three-year period.
Groundwater Contamination: The proposed wastewater management and “recycling” protocols involve discharging industrial backwash into the ground, creating a high risk of chemical infiltration into local aquifers under the guise of “waste mitigation.”
Decommissioning Liabilities: The absence of an upfront decommissioning bond leaves local taxpayers financially vulnerable. If operations cease, the community will be left to fund the removal of hazardous materials, including vehicle-sized lithium batteries, industrial diesel generators, and hazardous electronic waste.
Infrastructure and Thermal Hazards: The facility introduces severe risks of natural gas line ruptures and dangerous thermal emissions. Continuous operations will generate permanent thermal signatures and relentless noise pollution from 24/7 cooling fans and backup generators.
Fire Escalation Risk: The immense cooling and energy requirements, paired with regional water scarcity, severely compromise local fire suppression capabilities, creating a critical safety hazard.
Conclusion: the Data Center project presents a disproportionate financial and environmental burden to life.
The Financial stakeholders and planning commissioners must immediately account for these extensive liabilities on their balance sheets and halt the current expedited approval process until a comprehensive, independent impact study is completed.
https://www.facebook.com/share/p/1GeMnN1zD6/
Tomi Engdahl says:
When comparing the total water consumption of evaporative cooling versus closed-loop (air-cooled) systems, you must look at the life-cycle water footprint:
Direct Water Usage: Water evaporated or consumed on-site at the facility during cooling.
Indirect Water Usage: Water consumed off-site by power plants to produce the electricity required to run the facility and its cooling equipment.
U.S. Electricity Grid Water Intensity: In the United States, generating 1 kWh of electricity consumes an average of roughly 1.2 gallons (~4.5 liters) of water at thermoelectric and hydroelectric power plants.
Efficiency vs. Water Trade-off: Evaporative cooling uses the heat of vaporization to cool servers efficiently (lowering electricity draw / PUE to ~1.1–1.2). Closed-loop/air-cooled systems use zero direct water but require more electricity to run heavy compressors and fans in warm ambient temperatures (raising PUE to ~1.4–1.8)
On-Site Local Impact vs. Off-Site Regional Impact
Evaporative Cooling: Draws heavily on local municipal or groundwater supplies. Up to 85% of that water is evaporated directly into the atmosphere at the local site and lost from the local watershed.
Closed-Loop / Air-Cooled: Eliminates local water table strain entirely. However, the increased electrical demand shifts the water burden off-site to regional power generation stations.
When accounting for the U.S. average power grid, the total water footprints are surprisingly close:
On average, closed-loop air-cooled systems can achieve a slightly lower total water footprint (or roughly equal) compared to evaporative systems because direct cooling tower evaporation loses vast quantities of water on-site.
Grid Dependent Nuance:
On dirty/thermoelectric grids (coal/gas with wet cooling): Closed-loop systems consume significantly more off-site grid water, narrowing or reversing the gap.
On clean/renewable grids (solar/wind/zero-water energy): Closed-loop systems become vastly superior in total water efficiency, approaching near-zero total water consumption because the electricity used generates no water footprint.
Tomi Engdahl says:
From a local perspective, prioritizing air cooling (closed-loop) over evaporative cooling shifts the burden directly from local community resources to the developer and the electric grid.
When a facility uses closed-loop or air-cooled systems:
On-Site Direct Water Use = Zero: Municipal water, local wells, aquifers, and river basins are not drawn down for cooling towers.
Agricultural Protection: Local water tables, farm irrigation rights, and livestock supplies remain untouched by the facility’s daily cooling operations.
Local Ratepayer Defense: Community water utility infrastructure doesn’t need costly expansions to supply millions of gallons per day to a single site.
Higher Energy Demands Fall on the Operator
Air cooling requires significantly more electricity—especially during high summer heat—to drive industrial fan arrays and refrigeration compressors.
Corporate Cost: The facility operator absorbs the higher electric bill and must purchase additional energy to keep servers cool.
Grid Capital: The developer/utility must build or contract dedicated power infrastructure (substations, transmission lines, power purchase agreements)
While air cooling successfully protects the local water table, “higher energy demand” doesn’t entirely isolate the community if the broader grid isn’t prepared:
Grid Strain & Power Costs: If the extra electrical demand forces local power utilities to construct expensive new power generation or transmission capacity, those costs can sometimes be passed to regional electricity ratepayers unless strict utility protections are enforced.
Indirect Water Draw at the Power Plant: If the regional electric grid relies on fossil-fuel or nuclear plants that use evaporative cooling, the facility’s increased electricity usage still consumes water—it just happens off-site at the power plant rather than in the local town.
When local water preservation is the top priority for a community, insisting on air-cooled or closed-loop systems forces the developer to handle their own thermal burden through energy purchases rather than consuming local community water.
Higher energy demand falls on the builder/user as usually more cost. That’s why evaporative cooling has been liked by builders in USA. Thrend is towards closed loop systems with very low water usage.
On-Site Local Impact vs. Off-Site Regional Impact
Evaporative Cooling: Draws heavily on local municipal or groundwater supplies. Up to 85% of that water is evaporated directly into the atmosphere at the local site and lost from the local watershed.
Closed-Loop / Air-Cooled: Eliminates local water table strain entirely. However, the increased electrical demand shifts the water burden off-site to regional power generation stations. Cooling tower evaporation loses vast quantities of water on-site, which could be at the data center or power generation station.
When accounting for the U.S. average power grid, the total water footprints are surprisingly close:
On average, closed-loop air-cooled systems can achieve a slightly lower total water footprint (or roughly equal) compared to evaporative systems.
Grid Dependent Nuance:
On dirty/thermoelectric grids (coal/gas with wet cooling): Closed-loop systems consume significantly more off-site grid water, narrowing or reversing the gap.
On clean/renewable grids (solar/wind/zero-water energy): Closed-loop systems become vastly superior in total water efficiency, approaching near-zero total water consumption because the electricity used generates no water footprint.
Tomi Engdahl says:
https://meche.mit.edu/news-media/vapor-collection-technology-saves-water-while-clearing-air?hl=en-US#:~:text=The%20technology%20was%20originally%20envisioned%20by%20professor,and%20plumes%20from%20power%20plant%20cooling%20towers.
Tomi Engdahl says:
Technically, yes—water vapor can be captured. But from a physics and engineering standpoint, doing so defeats the entire purpose of using evaporative cooling in the first place.
https://meche.mit.edu/news-media/vapor-collection-technology-saves-water-while-clearing-air?hl=en-US#:~:text=The%20technology%20was%20originally%20envisioned%20by%20professor,and%20plumes%20from%20power%20plant%20cooling%20towers.
https://www.reddit.com/r/ask/comments/1mff7gq/if_data_centers_use_so_much_water_evaporating/?solution=iQUa%2FbIakC1qGD8%3DiQUa%2FbIakC1qGD8%3D&js_challenge=1&token=747a162344eed3b0fb0dea7cb14a4f2c&jsc_orig_r=&hl=en-US
Tomi Engdahl says:
What if the data centers would consume enormous amounts of energy but almost no matter. And considerable part of energy they take is reused for heating purposes (district heating, commercial greenhouse). Would that be a good idea?
Tomi Engdahl says:
Claim
Around these AI facilities Calves are being born still born
The claim regarding stillborn calves near data centers stems from recent public hearings and local opposition testimony in agricultural communities (such as in Texas and Alberta, Canada).
Local ranchers and political candidates have cited anecdotal accounts from farmers living adjacent to newly built industrial data centers who reported sudden drops in herd fertility, miscarriages, or stillbirths. While these claims are actively raised in town hall meetings and local government ordinances, there are no peer-reviewed veterinary or scientific studies confirming a direct cause-and-effect link between data centers and stillborn calves.
However, animal scientists and agricultural experts recognize several real physiological mechanisms through which nearby mega-data centers can stress livestock and disrupt cattle reproduction:
1. Chronic Low-Frequency Noise & Acoustic Stress
2. Groundwater Contamination & Nitrate Spikes
3. Thermal Stress (Micro-Climate Spikes)
Summary
While the specific claim of stillborn calves is currently based on anecdotal reports from farmers fighting local zoning projects rather than confirmed scientific studies, the underlying concerns—continuous low-frequency noise, heat stress, and potential well-water degradation—are legitimate veterinary risk factors for cattle health and pregnancy.
Tomi Engdahl says:
That classic proverb captures the core paradox of the entire tech expansion: Road to hell is paved with good intentions.
Tomi Engdahl says:
https://www.facebook.com/share/p/1BRu5cR1ek/
Real estate analysts continue studying how large industrial facilities, including data centers, may influence nearby residential property values. Factors such as continuous equipment noise, visual impacts, increased traffic, and large-scale infrastructure can affect buyer perceptions in some communities.
However, the effect varies widely depending on the facility’s design, location, landscaping, noise mitigation measures, and local market conditions. There is no universal rule that homes within a specific distance lose a fixed percentage of value.
As AI infrastructure expands, developers are placing greater emphasis on community engagement, improved architectural design, and environmental mitigation to reduce potential impacts on surrounding neighborhoods.
#DataCenters #RealEstate #Community #Infrastructure #Technology
Tomi Engdahl says:
Tomi Engdahl, indeed. All of the golf courses in the United States consume roughly 500 billion to 550 billion gallons of water annually. For comparison: The Scale: This total is roughly 7 to 8 times larger than the projected annual water consumption of all U.S. data centers combined in 2028 (approx. 68–73 billion gallons).
Tomi Engdahl says:
Large numbers of golf courses are now using effluent water.
The shift toward using effluent (recycled/reclaimed wastewater) in data centers is happening rapidly, but data centers face far tougher technical and logistical hurdles than golf courses.
Tomi Engdahl says:
https://www.facebook.com/share/p/1BSfqgKG42/
Major technology companies are increasingly signing long-term nuclear energy agreements to supply electricity for expanding artificial intelligence data centers. Reliable, carbon-free nuclear generation provides continuous power that complements renewable energy sources such as solar and wind.
As AI computing grows, electricity demand from data centers is rising rapidly. Long-term agreements with nuclear power producers help companies secure stable energy supplies while supporting emissions reduction goals.
The trend highlights how advanced computing is reshaping investment in the global energy sector.
#ArtificialIntelligence #NuclearEnergy #DataCenters #Technology #CleanEnergy
Tomi Engdahl says:
Using mineral oil for cooling data centers—specifically via liquid immersion cooling—is a technique designed to replace air conditioning and water-evaporative cooling towers.
Instead of blowing chilled air over hot servers, entire server racks are completely submerged in a tub of non-conductive (dielectric) fluid. The fluid absorbs heat directly from CPUs, GPUs, and memory chips, carrying it away to an external heat exchanger.
Immersion systems operate as sealed loops. They do not rely on evaporative cooling towers.
Extreme Hardware Density: Air cooling maxes out at around 30 to 40 kilowatts (kW) per rack. Immersion cooling can easily handle 100+ kW per rack.
While early DIY setups and early-stage crypto-mining farms used standard off-the-shelf mineral oil (or transformer oil), modern enterprise data centers rarely use raw mineral oil anymore.
Industry standards (such as the Open Compute Project guidelines) favor engineered synthetic hydrocarbons (like Polyalphaolefins or isoparaffins) and synthetic esters over commodity mineral oil
Tomi Engdahl says:
we actually can use urine to cool data centers—we just process it first through municipal sewage systems.
Piping raw urine directly into a data center causes extreme corrosion, toxic ammonia fumes, and pipe blockages.
However, by routing urine through municipal wastewater treatment, tech facilities can safely convert human liquid waste into millions of gallons of clean industrial cooling water without destroying equipment or harming local drinking water supplies.
Tomi Engdahl says:
It’s 2026 and nobody haven’t figured out a good ol radiator 20 times the size of a car to recycle It’s water and cool the data down……..
There’s more to this everyone..
Yes, it’s called a closed loop system. Modern chips have taken this a step further and use refrigerant to cool, water cooling is quickly becoming a thing of the past.
Tomi Engdahl says:
Jerry Gruber most people dont realize how dirty drinking water actually is when you are considering using it to cool tech. Drinking water in a cooling system is a recipe for failure. They also confuse using water to cool, in a closed loop non evaporative environment, vs consuming water where it evaporates into the air to cool. All of this also assumes water is planned to be used, vs air cooling or refrigerant cooling methods. The latest chips are being designed to use far less water than our legacy systems did. Most of the people claiming this amount of water usage is irresponsible, don’t understand how much water their power plants evaporate on an hour by hour basis.
Tomi Engdahl says:
they use that to fill the system the first time. The vast majority dont drain water or evaporated it. They recirculate the water over and over. Some of the hyperscales that use cooling towers do evaporate some, but that’s on the smaller condenser loop, not the cooling loop. The water that is drained from data centers goes back to the water treatment plant and gets reused. Golf courses and places like almond farms use that water and it evaporates or runs off with pesticides and fertilizers.
Also US golf courses consume 3-4 times the amount of water that all combined US data centers use.
Tomi Engdahl says:
“Open loop hasn’t been used since 1989/1990″ is a myth at least for huge data centers. Might or might not be true for smaller data centers.
Open-loop evaporative cooling towers have remain one of the most widely used cooling methods for large hyperscale data centers for much longer. Hyperscalers (Microsoft, Google, Meta, AWS) did not phase out open-loop cooling on a single historic date, but rather initiated a multi-phase transition away from open-loop evaporative cooling starting around 2021–2022, accelerating rapidly through 2024–2026.
Tomi Engdahl says:
When people point to agriculture in water conservation debates, alfalfa is almost always Exhibit A. It sits at the exact intersection of massive local water consumption and global trade economics. The Colorado River Basin: Alfalfa alone consumes ~26% of all water used across the entire Colorado River Basin—more than all the cities, homes, commercial buildings, and industrial facilities (including data centers) in the basin combined.
Alfalfa is not often eaten directly by humans (some seeds and sprout are eaten); it is grown mainly as high-protein forage for dairy cows, beef cattle, and horses. Alfalfa remains highly profitable for farmers.
The controversy intensifies over alfalfa exports. Critics argue that growing water-intensive crops in drought-prone U.S. states and shipping them overseas is functionally equivalent to exporting billions of gallons of fresh water to foreign countries.
While data centers draw intense community attention because they plug directly into municipal water grids, agriculture—and alfalfa in particular—is where the vast majority of physical freshwater is actually consumed.
Tomi Engdahl says:
In an open-loop evaporative cooling system (the kind that consumes millions of gallons) the possibility not to be able to recycle comes to the fundamental reasons boil down to thermodynamics, chemistry, and physics. The cooling effect relies on latent heat of vaporization. To catch that vapor and condense it back into liquid water, you would need massive condenser/refrigeration units. The electricity required to condense that steam would destroy the energy efficiency gains of evaporative cooling in the first place.
Most new data centers actually recycle water efficiently by using closed loop cooling system that does not consume much water in operation. The same water circulates in the data center pipes for years.
Tomi Engdahl says:
Water rights are property and state law determines who access. Data centers have to purchase those water rights from the farmer. FACT
Tomi Engdahl says:
Old evaporative cooled data centers (those whic use lots of water) vaporize most of the water they take it. It goes to the sky where it comes back down about week later typically few hundred mails away. Same happens to water used for growing crops and when you water your lawn.
Many new data centers use closed loop cooling where pipes are filled once (few swimming pools of water) and that same water circulates on the data center for few years. After that it might need to be replaced with new water (and old water needs to be cleaned/ treated).