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.
1,014 Comments
Tomi Engdahl says:
many parts of USA are in the real danger of running out drinking and irrigation water – with or without those data centers!
Data centers play a part in water shortage – in most locations they are a very small part of the water shortage problem.
Even if we would shut down all the around 5000 data centers today, that water shortage problems overall would not go away (could tempararily ease situation on some locations).
Tomi Engdahl says:
Lol
Data centers, lol
are you kidding
Meanwhile
Yes, recent climate change is primarily driven by human activities, especially the burning of fossil fuels.
(https://www.un.org/en/global-issues/climate-change)
Key Causes
Burning Fossil Fuels:
Using coal, oil, and gas for energy and transport releases massive amounts of carbon dioxide (\(\text{CO}_{2}\))
.Deforestation:
Cutting down forests reduces the number of trees available to absorb \(\text{CO}_{2}\) from the air.Agriculture:
Farming and livestock production release strong heat-trapping gases like methane (\(\text{CH}_{4}\))
Tomi Engdahl says:
Many data centers in USA use lots of fossil fuels to power them.
Transportation is the single largest direct source of greenhouse gas emissions in the U.S. (~28–29%). Data centers do not burn gasoline or jet fuel on-site (aside from monthly generator tests); their emissions are indirect, created at grid power plants that burn natural gas and coal to feed the electrical grid.
Data center related emissions are around half or slightly more of aviation co2 emissions.
Tomi Engdahl says:
In Finland some closed paper mills have been successfully converted to data centers.
Google’s data center in Hamina, Finland, is a major hyperscale facility located in a converted former paper mill on the Gulf of Finland.
Established in 2009 when Google acquired the former Stora Enso Summa paper mill, opening the first operational phase in 2011.
It supplies heat to Hamina city and also uses seawater cooling drawn directly from the Gulf of Finland to handle the heat not needed to heat up the city
It is powered by roughly 97–98% carbon-free energy, heavily backed by regional wind power contracts.
Collaborates with Haminan Energia to capture excess server heat and redirect it into the local municipal district heating network, supplying roughly 80% of the town’s annual heating needs.
Hamina, Suomi – Googlen palvelinkeskuksen sijainti https://share.google/jEULu4Unpe2O2Muwa
Tomi Engdahl says:
Why Tech Companies Build in Dry Regions (The Trade-Off)
Operators don’t choose arid climates out of spite; they do so due to specific geographical and electrical grid factors:
Solar Energy Availability: Water-scarce regions (like the American Southwest) often have the highest density of cheap, abundant solar power, allowing tech firms to meet renewable energy targets.
Low Humidity: Dry air makes evaporative cooling systems run significantly more efficiently than in hot, humid environments, reducing overall power consumption.
The Water vs. Power Trade-off: Operators face a physical choice—use water to cool the facility (evaporative), or use significantly more grid electricity to run massive fans and refrigeration compressors (dry cooling).
Tomi Engdahl says:
The Industry Shift
Because local municipal pushback in drought-hit counties has created major project delays, hyperscalers are aggressively moving away from open evaporative cooling towers in new U.S. builds:
Reclaimed / Non-Potable Water: Many facilities in places like Phoenix or Austin are mandated to use treated municipal wastewater or industrial effluent rather than drawing from local drinking water aquifers.
Transition to Dry Cooling: Newer architectures use sealed, closed-loop liquid systems or air-chilled radiators, dropping operational water consumption down to near zero at the expense of slightly higher power consumption.
Tomi Engdahl says:
What Data Centers Actually Compute: Over 99% of global data center capacity is built for everyday consumer and enterprise workloads:
AI Workloads: Training and inference for Large Language Models, computer vision, and autonomous systems.
Cloud Services: Enterprise software, medical record databases, video streaming (YouTube, Netflix), and social media platforms.
Scientific & Industrial Computing: Climate modeling, drug discovery, financial market clearing, and supply chain management.
Tomi Engdahl says:
Daily and Annual ScaleAI Data Centers (U.S.):
All U.S. data centers combined (including non-AI cloud traffic) use an estimated 17 billion to 90 billion gallons of water a year for direct cooling, with AI representing roughly 15% to 20% of that total. Projections estimate total data center water demand (direct and power-grid indirect) could reach 469 billion gallons by 2028.
Individual Facilities: A single large AI data center can consume between 1 million and 5 million gallons of water per day—comparable to the municipal water use of a small town of 10,000 to 50,000 people.
Comparisons to Other Water Uses in the U.S.Agriculture & Livestock:
U.S. agriculture uses over 27 trillion gallons of irrigation water annually. Farming and the meat/livestock industry consume hundreds of times more water than all projected AI needs combined.
Golf Courses: U.S. golf courses use over 500 billion gallons of water a year for turf irrigation—roughly 2 to 10 times more than the total annual footprint of all U.S. data centers and their associated energy generation.
Residential Lawns: Americans use over 2 trillion gallons of water a year just to water private lawns.
Everyday Goods: Producing consumer beef, cotton clothing, or almonds requires significantly more “blue water” (consumed freshwater resources) per year than the entire global and domestic AI infrastructure.
Tomi Engdahl says:
Modern closed-loop data centers do not just throw the water away.
How water is treated depends entirely on the design of the cooling system,
1. Closed-Loop Systems (100% Recirculated)
2. Seawater & River Thermal Systems (Direct Flow)
Where Water Is “Lost” (Evaporative Towers)
The reason people often think data centers “throw water away” comes from older or legacy evaporative cooling towers:
In those specific designs, water is sprayed into warm exhaust air, and heat is rejected through evaporation (turning liquid water into water vapor/steam).
That water is lost to the local atmosphere as steam and must be continuously replenished with tap water.
Because of environmental concerns and municipal water restrictions, modern AI facilities are rapidly moving away from evaporative designs in favor of closed-loop systems and dry air coolers that keep and recycle 100% of their operational fluids.
Tomi Engdahl says:
Tomi Engdahl Data centers primarily release or utilize PFAS (per- and polyfluoroalkyl substances), commonly known as “forever chemicals”, through cooling systems, hardware components, and electronic waste.How Forever Chemicals Are Released or UsedCooling Fluids and Refrigerants: Facilities use specialized fluorinated gases (such as hydrofluorocarbons or HFCs) and emerging liquid/immersion cooling fluids (like hydrofluorolefins or specific developmental products like Opteon 2P50) to manage high-density server heat. These chemicals can leak into the air, or seep into wastewater and local groundwater during maintenance or equipment failure.Semiconductors and Hardware: The millions of microchips and server components inside a data center rely heavily on PFAS during their manufacturing wafer-fabrication processes (including photoacid generators).E-Waste Disposal: Because high-powered AI servers degrade and must be replaced every few years, discarding old microchips and hardware creates massive amounts of electronic waste that can leach toxic residues into local environments.
Tomi Engdahl says:
https://www.facebook.com/share/p/19aZjnN4vc/
The server that stores your photos also warms a Finnish home. In Finland, a data center operated by Yandex (now rebranded) uses waste heat to warm 5,000 homes in the city of Mäntsälä. The heat is captured from the servers and fed into the local district heating network. The data center is powered by 100% renewable wind and hydro electricity. This is the circular energy economy in action: the same energy that powers the internet is used twice — once for computation, once for warmth. The project has been so successful that the city has attracted additional data centers, creating a cluster of heat‑recovering facilities. Finland’s cold climate and advanced district heating networks make it an ideal location. The heat from your video stream is now part of the city’s heating system. The cloud is not just storing data; it’s warming homes. The future of data centers is as much about heat as about bytes. The server rack is a radiator, and the city is the beneficiary. #datacenterheat #finlandenergy #districtheating #circulareconomy #cleancloud
Tomi Engdahl says:
https://www.facebook.com/share/p/1Bu7Y5gLyH/
Nuclear plants use far less land than solar farms, but land is only part of the story.
A one thousand megawatt nuclear plant needs about one square mile of land, while solar panels producing the same power need forty five to seventy five square miles. Our World in Data reports nuclear uses roughly thirty four times less land than solar per unit of electricity.
That land efficiency is real. So are the tradeoffs. Nuclear plants cost more to build and often take over a decade to come online, while solar can be deployed much faster and at smaller scale. Every energy source involves a different set of compromises.
The comparison matters because land constraints are growing. Large solar farms require vast acreage, sometimes competing with agriculture or natural habitat. Nuclear’s compact footprint leaves more land untouched, but its high capital costs and long construction timelines create financial and planning burdens.
The real question is not which one wins, but which tradeoffs a community is willing to accept. Energy transitions require honest accounting of land use, cost, speed, and environmental impact.
Source: Nuclear Energy Institute & Our World in Data
Tomi Engdahl says:
Your tightening makes this much stronger. Good call removing the 37-73bn by 2028 as a standalone headline – that figure comes from the same LBNL projection that says direct use could double to quadruples by 2028, so without the range and assumptions it reads as too precise.
You asked me to stress-test the three foundation claims. Here’s how they hold up against the underlying research:
17.4 billion gallons direct consumption in 2023
Defensible. This is directly from the LBNL 2024 United States Data Center Energy Usage Report, the most cited federal estimate.
LBNL reported ∼17 billion gallons direct in 2023, with some summaries rounding to 17.4B/17.5B gallons (66B liters). Multiple summaries of the report quote:
> “in 2023, U.S. data centers consumed 17 billion gallons (64 billion liters) of water”
> “directly consumed 66 billion liters (about 17.4 billion gallons) of water in 2023, up from 21.2 billion liters in 2014″
Your phrasing “estimated” is correct – it’s an estimate, and the report itself notes rapid change.
211 billion gallons indirect / associated with electricity
Defensible, with your qualifier essential.
LBNL also estimated:
> “consumed an additional 211 billion gallons (800 billion liters) of water indirectly through the electricity that powers them”
And: “That same report found indirect water consumption… was nearly 800 billion liters in 2023, roughly twelve times the direct figure”
You are exactly right not to add them and label it “AI water use.” The indirect figure:
Applies to all U.S. data center electricity, not just AI
Depends heavily on the generation mix – nuclear and coal need the most water, gas less, wind/solar almost none
Some viral math also incorrectly counts hydro reservoir evaporation as consumption
Your sentence that distinguishes direct cooling from the much larger indirect footprint of electricity generation, and says it should not simply be added, is the correct framing. LBNL’s authors and later reviewers make the same point.
3% to 9% Texas projection by 2040
Defensible and accurately caveated.
From UT Austin May 6, 2026 white paper “Water Use Requirements for Data Centers in Texas”:
> “Data centers could potentially account for 3% to 9% of Texas’ water use by 2040″
> “The estimate, which includes both water used for cooling and to produce the power needed for data centers”
> “Currently, data centers account for less than 1% of the state’s water use.”
For comparison, manufacturing is ∼7% of state water use, so your context helps readers understand scale.
Your inclusion that it depends on growth rate, cooling tech, and electricity source matches the researchers’ own uncertainty language: “There’s a lot of uncertainty surrounding the water use”
Your other two framing claims
Up to ∼5 million gallons/day for a very large hyperscale: This aligns with Guardian’s reporting on the same drought analysis: “Large datacenters, some the size of small towns, can require up to 5m gallons of water a day, equivalent to the water use of up to 50,000 people”. Calling it an upper-end figure, not typical, is accurate.
809 planned / 517 in drought areas: Confirmed as Guardian analysis using Cleanview + NOAA drought data. “About two-thirds of the 809 data centers planned… 517 data centers are set to be built in areas classified as drought-stricken”, and “Of 809 planned datacenters, 517 are in locations that have been in drought conditions throughout the past year”
You handled that correctly by saying “doesn’t mean every facility is currently facing shortage, but… where water availability is already under pressure.”
Overall this version is much more defensible than most pieces circulating. The only improvement I’d suggest is to cite the LBNL report year as 2024 report estimating 2023 use, so readers don’t think it’s 2023 research.
META AI
https://www.facebook.com/share/p/1EY34heSD7/
Tomi Engdahl says:
https://www.facebook.com/share/p/1CdbFZwhgw/
Growing demand for artificial intelligence infrastructure is raising new questions about how data centers use valuable water resources.
Large data centers require cooling systems to keep powerful servers operating, and some facilities rely on water-based cooling methods. Critics argue that using treated drinking water for industrial cooling should face stricter limits, especially in regions dealing with drought or water shortages.
Researchers and industry experts have debated the scale of this water use. Estimates vary depending on the size of the facility, cooling technology and local conditions. Some large facilities can consume millions of gallons of water, while others use more efficient systems designed to reduce demand.
Supporters of data center expansion point to economic benefits, technological progress and investments in cleaner cooling methods. Critics argue that communities should have stronger protections to ensure essential water supplies are prioritized.
As AI growth accelerates, the debate over energy, water and infrastructure will likely become a major part of planning future data centers.
#fblifestyle #Polarbear #AI #Environment #Technology
Tomi Engdahl says:
https://www.facebook.com/share/p/1BV5NCPczp/
What if the ocean could become part of the infrastructure powering the next generation of AI?
Samsung Heavy Industries is developing floating data centers designed to operate offshore. The concept uses seawater for cooling, potentially reducing the need for conventional land-based cooling systems while also easing pressure on land and electricity infrastructure. A 2026 partnership with Lloyd’s Register and Capital Clean Energy Carriers is helping advance the design, with a planned 50 MW-class system and a target for commercialization around 2028.
One important clarification: the current Samsung concept is not a proven system that turns seawater into drinking water. The seawater is primarily part of the cooling strategy. Whether floating data centers can operate reliably and economically at large scale is still being tested.
Would offshore data centers be a smart way to handle the growing demand for AI computing?
Source: Samsung Heavy Industries; Lloyd’s Register
Visuals are AI Illustrated.
#ArtificialIntelligence #DataCenters #Samsung #Technology #Innovation
Tomi Engdahl says:
https://www.facebook.com/share/p/1877fBQWda/
THE BOTTLE SAYS NATURAL. THE AQUIFER IT CAME FROM TOOK 5,000 YEARS TO FILL AND IS BEING EMPTIED IN DECADES.
Groundwater — water stored in underground aquifers in the pores and fractures of rock and sediment — provides approximately 50% of the world’s drinking water, 40% of irrigation water for food production, and baseflow for rivers and wetlands during dry seasons when surface water is insufficient. Unlike surface water, which recharges relatively quickly through the water cycle, deep aquifers accumulate over timescales of hundreds to thousands of years, meaning that extraction rates above the natural recharge rate permanently reduce the aquifer’s stored volume — a process called groundwater depletion that is occurring in virtually every major agricultural and urban region on Earth where groundwater extraction is the primary water source.
NASA’s GRACE satellite mission — which measures gravitational changes caused by water mass movement — has documented significant groundwater depletion in the Central Valley of California, the North China Plain, northern India, the Middle East, and numerous other regions where agricultural and industrial extraction has exceeded recharge for decades. The Central Valley aquifer system, which underlies some of the most productive agricultural land in the United States, is being depleted at rates that in some areas have caused the land above to sink by as much as 28 feet since the 1920s — a process called subsidence that permanently reduces the aquifer’s storage capacity even if extraction is reduced.
The global bottled water industry extracts groundwater for commercial sale in communities that frequently have their own water needs, their own wells, and their own historical relationship with the aquifer being extracted — a relationship that predates the extraction facility by generations but that carries less legal weight than the extraction permit in most jurisdictions. The permit is issued by a government. The aquifer was formed by geology. The community’s well was dug before the permit existed. The water table does not recognize the distinction between who has the document and who has the need.
The blind cave fish has lived in that aquifer for longer than any human settlement above it. It has never seen light. It has always had water. This is the first century in which the second of those two facts is uncertain.
#naturelover #nature #NatureEscape #savenature #SaveAnimals
Tomi Engdahl says:
Big tech = no social skills, no communication skills, and less caring for is and the creatures, and the environment.
Tomi Engdahl says:
Should be cooled by another method, we will run out of water,it’s ok the machines will kill us soon enough
they’re using a closed loop system now. Just turned a Meta plant in Idaho this week..closed loop reuses the same water for 5 to 7 years
Tomi Engdahl says:
https://www.facebook.com/share/p/18e7jXLDMx/
NVIDIA has developed a new cooling system for its next-generation Rubin AI infrastructure that can use liquid heated to 45°C (113°F) to cool powerful AI chips.
That’s hotter than the water in a typical hot tub, which is usually around 38–40°C (100–104°F).
The system uses direct liquid cooling, sending the coolant directly to cold plates attached to the processors and other high-power components.
The coolant is made of 75% water and 25% propylene glycol and moves through a closed loop. It enters the system at up to 45°C and can leave at about 55°C (131°F) after absorbing heat from the chips.
The liquid does not need to be colder than the chips to cool them. It only needs to be cool enough to absorb their heat while keeping the processors within their operating limits.
In suitable climates, NVIDIA says the 45°C design can allow data centers to use dry coolers instead of mechanical chillers, reducing the energy needed for cooling.
NVIDIA also says its closed-loop design can reduce on-site water consumption for cooling to near zero in suitable locations, compared with conventional cooling-tower systems.
That does not eliminate AI’s overall water footprint. Water can still be used to generate the electricity powering data centers and to manufacture the chips themselves.
NVIDIA says its Rubin generation uses 100% liquid cooling, with no fans needed inside the system.
Source: NVIDIA
Tomi Engdahl says:
“They could find a way to use environmental safe antifreeze. If it cools an engine it should cool the data center. Pump, fans, radiator, and a filter.”
There is no reason to use fresh water, they have options, might not be the cheapest but would beat the price of fighting communities and creating regulation.
The setup described (pumps, fans, radiators, filters, and environmentally safe coolant in a closed loop) describes a dry cooling system using direct-to-chip liquid loops.
Environmentally Safe Antifreeze Exists & Is Used: Data centers do not use toxic ethylene glycol. Instead, closed-loop facilities use inhibited Propylene Glycol (PG) mixed with deionized water (typically 20–30% PG) or specialized synthetic dielectric fluids. Propylene glycol is non-toxic, food-safe, and biodegradable.
Car-Radiator Mechanics Work: Next-generation AI racks use Direct-to-Chip (D2C) liquid cooling. Fluid pumps through copper cold plates attached directly to GPUs/CPUs, picks up the heat, flows out to an external closed-loop radiator (a “dry cooler”), and circulates back.
Bypasses Fresh Water Depletion: Because the fluid stays sealed inside metal piping indefinitely, these systems consume zero operational fresh water through evaporation.
While the concept works and is becoming the industry standard for new AI builds, it comes with specific engineering trade-offs that explain why evaporative cooling was used historically: The Energy Penalty and Capital Expense
Tomi Engdahl says:
The Modern Standard
Combining closed-loop plumbing, internal propylene glycol loops, advanced filtration, and local water retention represents the current baseline for sustainable data center engineering. Using these systems allows operators to drastically cut down on municipal freshwater draws, eliminate toxic chemical discharges, and reduce friction with local communities.
Tomi Engdahl says:
https://www.facebook.com/share/p/1Jmb1N5Gz7/
The AI boom has a physical footprint most of us never see, and water is becoming a major part of it.
A new analysis from Rystad Energy estimates that data centers consumed about 222 billion litres of water directly for cooling in 2025. If computing capacity continues expanding without major improvements in water efficiency, that could rise to nearly 644 billion litres every year by 2030.
The reason is simple: servers generate heat. AI hardware can pack enormous computing power into dense racks, and that heat has to go somewhere. Many facilities use evaporative cooling systems where water absorbs heat and some of it is ultimately lost to evaporation.
But 644 billion litres is not inevitable. Rystad estimates moderate efficiency improvements could lower the 2030 figure to around 543 billion litres, while aggressive water-saving measures could bring it down to about 388 billion litres. Technologies such as dry cooling can dramatically reduce direct water use, although they may require more electricity.
Location matters enormously too. AWS reported water-use figures ranging from just 0.02 litres per kWh in Stockholm to 2.85 litres per kWh in Jakarta, showing how climate and cooling design can produce differences of more than 100 times.
Perhaps the biggest concern is where this demand occurs. Rystad projects that regions already experiencing high or extremely high water stress could account for 34% of global direct data-center water consumption by 2030, with Jamnagar and Thane in India among the areas identified as particularly exposed.
And these numbers only cover water consumed directly for cooling. Producing the electricity that powers data centers creates an additional, less visible water footprint.
AI may exist digitally, but the infrastructure behind it is very physical. Every new generation of computing will force us to think not only about chips and electricity, but also about the water required to keep those machines running.
Source: Rystad Energy.
Tomi Engdahl says:
Data centers primarily release or utilize PFAS (per- and polyfluoroalkyl substances), commonly known as “forever chemicals”, through cooling systems, hardware components, and electronic waste.How Forever Chemicals Are Released or UsedCooling Fluids and Refrigerants: Facilities use specialized fluorinated gases (such as hydrofluorocarbons or HFCs) and emerging liquid/immersion cooling fluids (like hydrofluorolefins or specific developmental products like Opteon 2P50) to manage high-density server heat. These chemicals can leak into the air, or seep into wastewater and local groundwater during maintenance or equipment failure.Semiconductors and Hardware: The millions of microchips and server components inside a data center rely heavily on PFAS during their manufacturing wafer-fabrication processes (including photoacid generators).E-Waste Disposal: Because high-powered AI servers degrade and must be replaced every few years, discarding old microchips and hardware creates massive amounts of electronic waste that can leach toxic residues into local environments.
Derek Vanderwerf This claim correctly highlights that PFAS chemicals exist across electronics manufacturing and data center hardware, but it overstates their presence in mainstream data center cooling while underestimating the aggressive global regulatory phase-outs currently restructuring the industry.
Fact-Checking the Claims
1. “Facilities use specialized fluorinated liquids (PFAS/HFCs/HFOs) that leak or seep into wastewater”
Status: Partially True, but heavily misleading regarding adoption rates.
The Reality: The vast majority of liquid-cooled data centers do not use PFAS fluids.
2. “Semiconductors and hardware rely heavily on PFAS during manufacturing”
Status: True.
The Reality: This is an industry-wide electronics issue, not one unique to data centers.
3. “E-Waste leaches toxic residues when servers are discarded”
Status: Misleading regarding operational lifecycle.
The Reality: While server components contain trace amounts of fluorinated polymers in circuit board coatings and wiring, enterprise data center hardware is rarely thrown into open landfills. High-density AI servers (equipped with high-value gold, copper, silver, and rare earth elements) are tightly managed through enterprise IT Asset Disposition (ITAD) programs. Old servers are refurbished, harvested for parts, or sent to specialized electronics recyclers that extract precious metals under industrial environmental controls.
Tomi Engdahl says:
On-site reclamation—such as filtering cooling tower blowdown, capturing surface runoff, or recycling industrial wastewater— lowers a facility’s reliance on local drinking water. However, it does not magically eliminate environmental trade-offs.
Reduces Potable Water Reliance: Advanced on-site treatment (like reverse osmosis, ultrafiltration, and chemical softening) allows facilities to process non-potable municipal effluent or recycle their own discharge. This can cut a site’s freshwater intake from public drinking supplies by up to 70–90%.
The Unsolved Physics & Environmental Trade-offs
Evaporative Loss Is Permanent (Thermodynamics):
In open-loop evaporative cooling towers, the primary way heat is dissipated on older data centers is by turning liquid water into steam/vapor.
The “Blowdown” and Concentrated Brine Issue:
As water recycles, pure H_2O evaporates, leaving behind concentrated salts, heavy metals, and minerals (Total Dissolved Solids).
The Indirect Water-Energy Nexus:
On-site water purification systems (like Reverse Osmosis pumps, UV sterilizers, and filtration skids) require significant electrical power to run.
Summary
On-site water reclamation is a critical mitigation strategy that drastically reduces a facility’s footprint on municipal drinking water supplies. However, it is not a complete silver bullet; total water elimination is only achieved by transitioning away from evaporative systems entirely and adopting closed-loop dry cooling or non-potable seawater heat exchangers.
Tomi Engdahl says:
If you are reading this, thank the data centers you are depending on, for always being there.
If you think data centers are only for AI, think again.
If you are employed, your employers is dependent on multiple data centers.
Your grocery store is dependent on data centers. Along with every single business, educational, healthcare, entertainment and public utility service you depend on.
Your local, county, state and federal government are all dependent on data centers.
We are as much dependent on data centers as we are dependent on our phones and electricity itself.
The United Stated is over 3m square miles. Data Centers today occupy 300 square miles. There has got to be room and proper locations for data centers.
Unless you are completely living off the grid, you are dependent upon data centers every day.
Tomi Engdahl says:
”Rivers in U.S.A. are drying up because of data centers”
Status: False / Overstated.
The Reality: U.S. data centers do not single-handedly dry up rivers. Direct water consumption by all U.S. data centers combined accounts for roughly 0.01% to 0.02% of total national freshwater withdrawals.
What actually depletes U.S. rivers: Major river systems facing severe strain (such as the Colorado River or the Rio Grande) are depleted primarily by agricultural irrigation (which uses ~70%–80% of regional water), followed by public municipal use and heavy manufacturing.
Tomi Engdahl says:
https://www.facebook.com/share/p/1K4NvF9C7D/
That quick ‘thank you’ email you just sent has an environmental cost.
Fortunately, simple and easy-to-implement digital habits can help slash our collective carbon footprint.
Behind every standard calendar invite, quick confirmation, and newsletter in your inbox lies an invisible web of energy-hungry infrastructure. While hitting “send” feels completely weightless, each digital message you dispatch has a measurable carbon footprint.
Every single email relies on a vast physical system of devices, high-speed networks, and massive data centers that run 24/7. Keeping these giant server facilities cooled and powered requires an immense amount of electricity—much of which is still generated by burning fossil fuels.
According to climate researchers, a standard email emits about 0.3 grams of carbon dioxide equivalent, while messages with large attachments or heavy images can balloon up to 50 grams.
Although a single email’s impact seems trivial, the collective weight of billions of daily messages adds up rapidly. Simple behavioral shifts, like resisting the urge to click “Reply All” on mass threads or periodically clearing out unread clutter from your archive, can immediately lower our personal digital footprints. By being more intentional with our digital habits, we can collectively ease the strain on the global energy grid.
source: Berners-Lee, M. (2020). How Bad Are Bananas?: The Carbon Footprint of Everything. Greystone Books.
Tomi Engdahl says:
The claim made in that post—that sending a simple “thank you” email creates a significant carbon footprint, and stopping them would save tens of thousands of tonnes of CO₂—stems from a widely circulated 2019 campaign, but it grossly overstates the environmental impact.
Tomi Engdahl says:
https://www.facebook.com/share/p/19pzsn56DS/
Finnish engineers are capturing waste heat from data centers to warm homes in Helsinki.
A new district heating system in Helsinki uses large heat pumps to recover waste heat from data centers and upgrade it to eighty degrees Celsius for distribution. The system connects to the existing network, replacing natural gas boilers and cutting emissions.
VTT Technical Research Centre of Finland designed the heat pump cascade using carbon dioxide as refrigerant. Data centers produce large amounts of low-grade heat that is normally vented to the atmosphere; capturing it provides a stable, year-round heat source.
The project supplies heat to several thousand apartments and commercial buildings. Helsinki aims to become carbon neutral by 2030, and waste heat recovery is a key strategy to decarbonize its district heating, which currently relies heavily on fossil fuels.
Finland’s cold climate and growing data center industry create ideal conditions for this approach. Replicating the system in other cities could turn digital infrastructure into a clean heat source, reducing emissions and energy imports.
Source : VTT Technical Research Centre of Finland, Smart Energy Systems, 2025
Tomi Engdahl says:
Just a friendly reminder the highest water usage is in reactors for electricity production. The problem isn’t the amount of water the data center uses on site. It’s the water that will be used to cool those reactors because of the extremely high consumption of electricity to run the data centers.
Tomi Engdahl says:
Armando Gene Acosta there’s many problems with ai data centers but the indirect water usage is a problem. Even closed loop zero water data centers use tons of water to generate the power needed for these data centers so unless they are powered by 100% renewable energy then they will constantly use tons of water and united states is against renewable energy yet somehow say natural gas is clean which it isn’t
Tomi Engdahl says:
So if they put the data centre next to a power station, the data centre can pre-heat the water for the station, making the power station consume less energy itself, then the condensed used water could be used in the data centre again- in a loop. More efficient and less energy use overall. Except vested interests won’t let that happen….
Martin Morris it’s a little bit more complicated than that. You won’t see temperatures above 110° in a data center, and so preheating water for a power plant isn’t super useful. Also you still need to cool the water to about 50° to be useful to a data center. It would be adding a lot of extra steps and potential failure points that doesn’t actually work for either application. A better solution in cold places would be to use the warm water throughout a city and have radiators in apartments and homes to get free warming/cooling in the winter months.
Sean Krake
“A better solution in cold places would be to use the warm water throughout a city and have radiators in apartments and homes to get free warming/cooling in the winter months.”
This is actually implemented in Finland and Sweden at several cities. I have lived in house that used data center heat for heating (significant part of heat came from data centers). I have also visited a site where that waste heat from data center was captured and processed.
Tomi Engdahl says:
Claim seen
almost all data centers use closed loop cooling which uses water but not very much. the data center water use thing is the myth that won’t die.
This claim contains a mix of accurate technological trends and significant oversimplifications regarding the global data center fleet.
”Almost all data centers use closed loop cooling”:
False. Historically and currently, a majority of large-scale, hyperscale data centers still rely on evaporative cooling towers (open-loop systems).
Closed-loop cooling is the fast-growing target standard for new high-density AI builds because operators want to eliminate public friction and secure building permits. However, describing the water issue as a total “myth” ignores the thousands of active evaporative facilities currently drawing significant local freshwater supplies.
Tomi Engdahl says:
Seen comment:
As someone who has worked on the construction of data centers in the HVAC capacity I can attest to the fact that the claim of excessive water consumption is exaggerated. The centers that I have worked on rely on closed looped cooling systems meaning the water is recycled. However even in such a system there is need of make-up water and always the chance of catastrophic failure and the need to drain and refill the system. The genie is out of the bottle and AI will continue to expand regardless of public opinion. What we should be focused on is regulating this tool ahead of the curve and not as an afterthought like was done with internet security.
Tomi Engdahl says:
Comment seen:
James Asher you’re actually wrong there. It costs more to operate an evaporative cooling system because of the cost of the water and chemicals and cooling towers. That’s why 75% of the data center industry is cooked by air cooled chillers or direct expansion systems already.
We SHOULD legislate that no new evaporative cooling systems can be installed for data centers.
The bigger water issue is that data centers require power, which right now is primarily boiling water to spin turbines, and cooling that steam using evaporative cooling towers, so we need to be more serious about finding a better solution to our energy crisis, which is consuming 10,000x more water than data centers are.
Tomi Engdahl says:
Claims:
Y’all know the very AI that is causing the planet to die can tell you how to make the necessary impliments to make data centers go dabigboom right?
Emrick Azor but can they tell tech bros that they need to take their defunct product back to R&D instead of killing everyone with it?
Rosie Wright Nah, that’s why it’s gotta be a war. With the world’s cruelest handicap, AI haters VS the disgusting slops who sit in front of their computer talking to the AI cuz they have no friends? We’d win in a day.
Tomi Engdahl says:
https://www.facebook.com/share/p/1DpDGY3v94/
The server that stores your photos also warms a Finnish home. In Finland, a data center operated by Yandex (now rebranded) uses waste heat to warm 5,000 homes in the city of Mäntsälä. The heat is captured from the servers and fed into the local district heating network. The data center is powered by 100% renewable wind and hydro electricity. This is the circular energy economy in action: the same energy that powers the internet is used twice — once for computation, once for warmth. The project has been so successful that the city has attracted additional data centers, creating a cluster of heat‑recovering facilities. Finland’s cold climate and advanced district heating networks make it an ideal location. The heat from your video stream is now part of the city’s heating system. The cloud is not just storing data; it’s warming homes. The future of data centers is as much about heat as about bytes. The server rack is a radiator, and the city is the beneficiary. #datacenterheat #finlandenergy #districtheating #circulareconomy #cleancloud
Tomi Engdahl says:
Key Contrasts in Approach
1. Energy Sourcing: Green Geography vs. “Bring Your Own Power”
The European Models: Sweden and Finland rely on their natural geographies (abundant hydro, nuclear, and wind). The Netherlands works within a tightly regulated grid ecosystem where green power tracking is legally enforced.
xAI: Because training frontier AI models (like Grok) requires unprecedented, immediate blocks of power (scaling rapidly toward gigawatt-level demand), traditional grids couldn’t move fast enough. This forced xAI into an unconventional, controversial “off-grid/behind-the-meter” approach—deploying large fleets of natural gas turbines and Tesla Megapacks across state lines in Mississippi/Tennessee to bypass local utility bottlenecks.
2. Environmental Philosophy: Circular Economy vs. Raw Compute Velocity
The European Models: Sustainability is treated as a systemic loop. Heat isn’t just dissipated; it is weaponized as a public utility to heat homes, greenhouses, and municipal water. Construction materials and PUE (Power Usage Effectiveness) are heavily optimized for ecological harmony.
xAI: The overarching directive is maximum computational throughput and speed-to-market. Environmental costs (such as emissions from gas turbines, air quality index impacts in surrounding communities, and massive water/cooling resource draws) have taken a back seat to out-scaling competitors in the AI hardware race, triggering intense pushback from local groups and federal regulators.
Summary Takeaway
Sweden, Finland, & The Netherlands represent sustainable integration—where data centers must adapt to the ecological, spatial, and social boundaries of the region.
xAI represents unconstrained brute-force expansion—prioritizing raw AI training velocity over local grid limitations, resulting in a heavy reliance on fossil-fuel generation and friction with environmental regulations.
Tomi Engdahl says:
https://www.facebook.com/share/p/1DR1aGEVa1/
Norway’s underwater data center now runs on tides alone.
A subsea data center in a Norwegian fjord is now powered entirely by tidal energy turbines. The turbines generate electricity from predictable tidal currents, providing a reliable, renewable power source. The data center uses cold ocean water for cooling, reducing total energy consumption by forty percent.
The project combines two Norwegian strengths: offshore engineering and renewable energy. The tidal turbines are installed on the seabed near the data center, minimizing transmission losses. The system operates year-round, unaffected by weather.
Underwater data centers are attractive because they eliminate cooling costs and can be deployed near coastal cities. This project demonstrates a fully self-sufficient, zero-emission design. It could be replicated in other locations with strong tidal resources.
Researchers at the Norwegian University of Science and Technology are monitoring performance and environmental impacts. The data center has operated for one year without a single outage. This is a blueprint for sustainable digital infrastructure.
Source : Norwegian University of Science and Technology Tidal Data Center Report, 2025
Tomi Engdahl says:
https://www.facebook.com/share/p/1EaRxe3w2X/
Iceland powers massive data centers entirely with geothermal heat instead of fossil fuels
Data center operators have increasingly chosen Iceland as a hosting location specifically because the country’s electricity grid runs almost entirely on renewable geothermal and hydropower generation, allowing energy-intensive computing operations to run with a dramatically smaller carbon footprint than facilities located elsewhere. Iceland’s naturally cold climate also reduces cooling costs significantly.
Because geothermal power remains constant regardless of weather conditions, unlike solar or wind, data centers in Iceland benefit from remarkably stable, predictable electricity supply that helps prevent costly computing disruptions during peak demand periods throughout the year. The country’s abundant renewable capacity relative to its small population also means expanding data center demand hasn’t strained the broader national grid.
Officials say Iceland’s unique combination of clean baseload power and natural cooling continues attracting new data center investment despite the country’s relatively remote geographic location.
Source: Icelandic National Energy Authority, 2026
Tomi Engdahl says:
https://www.facebook.com/share/p/19Y4u12oht/
The internet’s energy appetite is being cooled by the largest freshwater system on Earth. A new data center in Michigan uses cold water from Lake Michigan to cool its servers, cutting cooling energy use by 60% compared to traditional air conditioning. The system pumps lake water through heat exchangers, then returns it at a slightly warmer temperature — no evaporation, no chemicals, no cooling towers. The data center also runs on 100% renewable energy from Michigan wind farms. This is what clean computing looks like: the same lakes that built the Midwest’s industrial economy now cool its digital one. Data centers consume 4% of U.S. electricity, and that’s doubling by 2030 thanks to AI. Lake cooling is a simple, natural solution that most data centers ignore because it requires being near a body of water. Michigan, with its 11,000 inland lakes and Great Lakes coastline, is positioning itself as the green data center capital of America. The cloud is getting bigger, but in Michigan, it’s getting cooler and cleaner. Tag someone who works in tech. Why it matters: The internet uses 4% of U.S. power — lake cooling cuts that in half. #datacenter #lakecooling #michigan #energyefficiency #cleancomputing
Tomi Engdahl says:
AstroPhilesz
You’re framing this as a simple environmental tradeoff, but there’s a much bigger strategic issue hiding underneath it.
If America decides that the answer to AI’s resource demands is to restrict, delay, or surrender domestic computing capacity, we are not making technology disappear. We are simply making ourselves less competitive while countries like China continue building the infrastructure.
That is the part people need to understand.
China is not going to look at America’s water consumption and say, “Well, I guess we should stop developing AI.” They are going to build the data centers, secure the energy, develop more efficient cooling, manufacture the chips, and expand their computing capacity. If the United States voluntarily limits its own infrastructure because building it has costs, we hand them a strategic advantage without them having to take it from us.
The answer isn’t “build data centers regardless of the consequences.” The answer is build smarter.
Use reclaimed and recycled water. Build closed loop cooling systems. Put facilities where water and power can support them. Invest in nuclear, renewables and grid capacity. Develop more efficient chips and cooling technology. Require responsible resource management.
That’s leadership.
Saying “this uses too many resources, therefore we shouldn’t build it” is not environmental leadership if the predictable result is that China builds the infrastructure instead. That’s unilateral disarmament in the technological competition of the 21st century.
Freshwater matters. So does technological sovereignty.
And if we’re serious about competing with China, we don’t surrender the computing infrastructure that determines who controls the next generation of artificial intelligence, advanced manufacturing, scientific research and national security.
You don’t win a technological arms race by voluntarily leaving the arena. You win by building the capability to do it better, cleaner and more efficiently than your competitors.
Tomi Engdahl says:
We are in a closed system. Hydrogen does not have enough velocity to escape the gravitational force of earth. Water isn’t going anywhere. Stop the fear mongering
Tomi Engdahl says:
Yes, freshwater is disappearing globally at alarming rates. Recent studies indicate that Earth’s continents have experienced unprecedented freshwater loss since 2002, driven by factors such as climate change, unsustainable groundwater use, and extreme droughts.
The world loses approximately 324 billion cubic meters of freshwater annually, enough to supply about 280 million people.
1
75% of the global population lives in countries that have been losing freshwater for at least 22 years.
1
The rate at which dry areas are getting drier is now outpacing the rate at which wet areas are getting wetter, reversing long-standing hydrological patterns.
1
This situation poses severe implications for water security, agriculture, and global stability.
2
The ongoing depletion of freshwater resources is a critical issue that requires urgent attention and action.
Tomi Engdahl says:
https://www.facebook.com/share/p/19nn1KutuA/
Data centers could consume 644 billion litres of water annually by 2030 just to stay cool
A new analysis from Rystad Energy estimates that data centers consumed about 222 billion litres of water directly for cooling in 2025, a figure that could rise to nearly 644 billion litres annually by 2030 if computing capacity expands without major water efficiency improvements. Many facilities use evaporative cooling systems where water absorbs heat and is ultimately lost to evaporation.
Rystad estimates moderate efficiency improvements could lower the 2030 figure to around 543 billion litres, while aggressive water-saving measures could bring it down to about 388 billion litres, though technologies like dry cooling that reduce water use often require more electricity instead. Location matters enormously too, with AWS reporting water-use figures ranging from just 0.02 litres per kWh in Stockholm to 2.85 litres per kWh in Jakarta.
Rystad projects that regions already experiencing high water stress could account for 34 percent of global direct data-center water consumption by 2030, with areas in India among those most exposed.
Source: Rystad Energy, 2026
Tomi Engdahl says:
https://www.facebook.com/share/p/1F7ZTTuQE7/
Singapore is cooling servers with the sea itself.
A floating data center off Singapore’s coast now operates entirely on tidal energy. The facility uses underwater turbines to generate electricity from ocean currents. Seawater cooling eliminates air conditioning, reducing energy consumption by fifty percent. The data center emits zero carbon.
The structure is moored in coastal waters, connected to shore by high-speed fiber optic cables. It houses servers for cloud computing and AI applications. The cold seawater is pumped through heat exchangers, maintaining optimal temperatures.
Singapore faces land scarcity and high energy costs. Floating data centers solve both problems, using ocean space and renewable energy. The project is a collaboration between the government and technology companies.
The facility has operated for one year without a single outage. It demonstrates a model for sustainable digital infrastructure in coastal cities worldwide. Singapore is leading the way in green technology.
Source : National University of Singapore and Singapore Data Center Report, 2025
Tomi Engdahl says:
Every single time you use your phone, make a post, take a picture, send a text, email, Facebook post, medical records, dental records, TV streaming, ALL social media, booking airline flights on line, home security camera, checking/savings accounts, ATM transactions, digital home phone, streaming music, on line payments, car GPS systems, credit card charges, on line
payments, restaurant reservations, hotel reservations, airline on line tickets. all financial transactions, legal documents, home data cameras, and more ALL USE DATA CENTERS.
Tomi Engdahl says:
See the facts: https://www.cs.mu.edu/~papers/StevenGoodman/DataCenters/DataCenterInfrastructure-ADecisionReference.pdf
Tomi Engdahl says:
https://www.facebook.com/share/p/1DhGGqYwq3/
The internet’s most secure data center is inside a mountain. In the Swiss Alps, a data center has been built inside a former military bunker, deep within a mountain. The natural coolth of the rock keeps the servers at optimal temperature with minimal mechanical cooling, and the facility is powered by 100% Swiss hydroelectricity. The mountain provides physical security, protection from electromagnetic pulses, and a constant temperature of 10°C. The data center serves financial institutions and government agencies that require the highest levels of security and sustainability. Switzerland’s abundant water power makes it one of the greenest places to compute. The server that stores your most sensitive data is literally under a mountain, powered by water, and cooled by stone. The future of secure computing is underground, renewable, and Swiss. The Alps are not just a playground; they’re a data center. The mountain that has stood for millions of years is now home to the internet’s most valuable data. #swissdatacenter #hydroelectric #mountaincooling #securecomputing #cleantech
Tomi Engdahl says:
https://www.facebook.com/share/p/1ActAx8qdp/
The data center that runs on pure Arizona sun. A new data center in Phoenix, Arizona, has been designed to operate entirely on solar power during the day and battery storage at night. The facility has a dedicated 50 MW solar farm and a 200 MWh battery system, making it independent of the fossil‑fuel grid. Arizona’s abundant sunshine — over 300 sunny days per year — makes this possible. The data center also uses evaporative cooling, which is effective in the dry desert climate. This is a model for sustainable computing in sunny regions, proving that even the most power‑hungry facilities can run on clean energy. The same sun that bakes the desert is now powering the servers that run your apps. The battery that stores the sun for night use is the key to 24/7 renewable operation. The data center is no longer a burden on the grid; it’s a self‑sufficient energy island. The future of computing in the Sun Belt is solar‑powered. #solardatacenter #arizonasolar #batterystorage #cleancloud #offgridcomputing