There is a reason why Big Tech has been so super-secretive about its water use. It’s because it was getting it so cheap that it wasn’t even worth measuring it. Water was this invisible externality, and because it was invisible to the public, Big Tech could keep telling the story of how it was becoming more and more efficient, more and more green, more and more “in the Cloud.” The cloud is on the ground. Big Tech abuses and misuses water at an alarming rate, treating it as essentially a free resource, and to keep on doing that, it was essential that the public should not know what was happening.

“Water consumption in data centers is super embarrassing,” a data center designer stated as far back as 2009. “It just doesn’t feel responsible.” Survey after survey showed that about 60% of data centers saw “no business justification for collecting water usage data.” Think about that. They were getting the water so cheap they didn’t even bother metering it. In a Microsoft data center in San Antonio, Texas, it was found that the actual cost of water should have been 11 times higher than what the company was paying. In drought-stricken Holland, a Microsoft data center slurped 84 million liters of drinking water in one year, when the local authority said the facility would only need 12 to 20 million liters.

Data centers know they have been hugely abusive of their water use and thus are desperate to hide usage figures from the general public. “There were actually water documents tracking how much this data-center campus was using,” journalist Karen Ho explained about a Microsoft data center. “But when the city came back with documents about that, everything was blacked out. They said that it was proprietary to Microsoft and therefore they couldn’t provide that information.” Public water use is proprietary to Microsoft. Sounds about right.

“The reason there’s not a lot of transparency, simply put, I think most companies don’t have a good story here,” stated Kyle Myers, a senior manager at a data center company. Data centers have a choice. They can either consume less water and use more electricity. Or they can use less energy and consume more water. “Water is super cheap,” Myers said. “And so people make the financial decision that it makes sense to consume water.”

In 2023, Bluefield Research estimated that, on a global basis, data centers were using more than 360 billion liters of water a year, including water used in energy generation. It predicted this figure would rise to over 600 billion liters a year by 2030. However, China Water Risk estimated that in 2024 China’s data centers alone used about 1.3 trillion liters of water, which is the equivalent of what 26 million people need. This amount would grow to more than 3 trillion liters a year by 2030, due to the explosive growth of AI. Which figures are correct? We don’t know.

In so many ways, data center water use is more intensive than the way an ordinary person uses water, as Shaolei Ren explained to Reece Rogers for Wired:

“The water that is available for people to use is very limited. It’s just the fresh surface water and groundwater. Those data centers, they’re just evaporating water into the air. They’re different from normal, residential users. When we get the water from the utility, and then we discharge the water back to the sewage immediately, we are just withdrawing water—we’re not consuming water. A data center takes the water from this utility, and they evaporate the water into the sky, into the atmosphere, where it may not return to the earth’s surface until a year later.”

Data centers love to be efficient. That’s why they love deserts, because dry air reduces the risk of damage and corrosion to their sensitive servers and other electrical equipment, and thus helps them to run more cost effectively and efficiently. As Steven Gonzalez explained to me:

“If you have access to cheap fresh water, deserts are a great place for data centers because they are so dry—and computers hate moisture and high humidity. That’s why there are so many data centers in Arizona. It’s almost like the goldrush. It’s a water-rush. All these companies are clustering to get this cheap water. But it’s doomed. We see how communities are struggling to pay their water bills, while data centers and other industries are getting water at a much cheaper rate. There are farmers who are directly competing with data centers to grow food. Indigenous communities are also having difficulties accessing water. The draining of the Colorado river is affecting the migration patterns of salmon and other fish, which are really important to their lifecycles.”

Of course, water evaporation in a desert environment is going to be very intense. Nicolas Dubé of Hewlett Packard Enterprise, described it as a criminal activity.

“Some hyperscalers, I’m not going to name them, built large datacenters in Arizona, New Mexico, and very dry countries. You build datacenters there, and if you use evaporative cooling, you’re going to have spectacular PUE (Power Use Efficiency). However, you’re going to consume a resource that’s way more important to that community than optimizing for a few percent of the energy consumption. I think that’s criminal. I think they should be jailed for doing that.”

Big Tech capitalist pursuit of efficiency and lowest costs is not simply ruinous to local water and to freshwater supply in general. “There is more moisture in a warming sky, a 7% increase for every degree Celsius of warming,” Kate Marvel wrote in The Climate Book. Data center evaporative cooling sends even more water up into the atmosphere, thus accelerating global warming and super storms. Historically, this water was used by plants and animals, stored in soils or underground in aquifers. Now, it’s up in the skies, accelerating droughts and super storms, where it can rain more in one day than it usually rains in a year. Droughts and floods are terrible twins.

Data gets hot. Computer servers get hot. They need to be cooled. There is a direct correlation between the amount of energy a server or data center is using and the amount of water required to keep it cool. “On average, between one to nine liters of water are evaporated during the cooling process of one kilowatt hour (kWh), measuring the amount of energy needed to power one machine that consumes 1,000 watts for one hour,” Maxim Melamedov wrote for Data Center Dynamics in 2024.

With AI, this digital thirst is surging. Our data is drinking more water than we do. Why does data need to drink so much water? Because, despite there being a global freshwater crisis, water is sold to Big Tech at super-cheap rates. Thus, it can be up to 10 times cheaper for Big Tech to choose water as a cooling agent versus other options.

If you look at the back of some fridge freezers you will find piping running back and forth. A server rack can have the same type of intricate and extensive piping, down which water—or other liquids—flow. As the servers hum, their heat is transferred through these pipes into the water. When the water reaches a certain temperature, it moves along the piping out of the server room and into what is called a cooling tower. There, the pipes are sprayed by other water to cool them down. So, there are two sets of water involved, one that is in the pipes and one that is in the cooling tower, and they don’t mix. As the water is sprayed on the pipes in the cooling tower, much of it evaporates. This is what is called evaporate cooling. As the water evaporates, it leaves behind a waste made up of minerals, metals and salt. Historically, data center water systems had been known as a source for Legionnaires’ disease. To avoid such infectious risks, chlorine and bromine-based chemicals and disinfectants are added. Bromine is highly toxic for living systems, targeting the nervous system and brain. Even with chemical treatment, the water in these pipes cannot be used endlessly, as data center expert Steven Gonzalez explained to me:

“As water is being warmed and flowing through these data centers, microorganisms flourish in these conditions. That is one reason why data centers turn to drinking water because that water has already to some degree been treated, so there is less of a risk of these microbial blooms happening. For the same microbial reason, the water can’t be endlessly recycled. It has to be dumped or returned to the sewers because even with reverse-osmosis filters and other techniques, these microbes will flourish.”

It is important to understand that we are only scratching the surface of data center demand for water, energy and other materials. A tsunami would not sate the big mouth of Big Tech. And be certain that, as water gets scarcer, Big Tech will be first in line to drink its fill. We are only at the beginning of the age of Big Tech’s hunger for atoms.

A data center moving into a community is like a prison setting up. Only worse. Super-high, aggressive security; ugly warehouse buildings. A prison will bring a decent quantity of jobs. Data centers bring hardly any jobs. What’s more, a data center will consume massively more water and electricity than a prison, while also causing far more noise and toxic e-waste.

Data centers are the new mines. They’re there to mine data, to mine electricity, to mine water, to mine us, to make us slaves to the data, data slaves, imprisoned and exploited by our own data. We are the new ore that the old imperialists and colonizers have come to extract from and then discard. For most data is not for the common good. It is for the sale of goods. Particularly for the sale of goods we don’t need and that are bad for the environment, for from these types of goods and services are the maximum profits made. Data centers are the hubs of surveillance-capitalist planned-obsolescence overconsumption.

Data is physical. For every byte that exists, energy, water and materials are required. Data exists on a machine in a building. A typical data center might be around 9,000 square meters (about 100,000 square feet). That’s about the size of a large supermarket. A data center tends to get classified as “hyper-scale” when it has more than 5,000 computer servers and is more than 10,000 square meters. At about 1.6 million square meters, the CITADEL, in Nevada—the driest state in the USA—is one of the largest data centers. It’s about 175 supermarkets in size. Very physical.

Inside these buildings, we mainly find what are called server racks or server cabinets. These are generally metals cabinets that are used to store the computer servers. A typical server rack is 42U (U being the measurement of 1 rack unit). The rack is about 200 cm (78.5 in) high, 48 cm (19 in) wide, and 80 cm (36 in) or 107 cm (42in) deep, making it about the size of a large family fridge freezer. In a non-AI rack, you might find about 30 servers.

Before the 2020s, each computer server might be consuming about 200 watts of electricity, with the total watts for the rack being about 5 kW or 5,000 watts. This began to change rapidly as the 2020s progressed, with some non-AI servers consuming as much as 700 watts, with the average being around 400 watts. So, an early to mid-2020s non-AI server rack with 30 servers, might be consuming about 12,000 watts and potentially as many as 21,000 watts.

AI radically changed those calculations. AI servers themselves can be four to five times taller than a traditional server, containing as many as eight graphic processor units (GPU). These are the intense processors that made NVIDIA one of the most valuable companies in the world. Each GPU can demand between 700 and 1,000 watts. Assuming an 800 watts average, an AI server could be demanding about 6,400 watts per server. Meaning each AI GPU server was using what a whole rack used to use in 2020.

You might get about eight AI servers in a rack, meaning that an AI rack could easily have a demand of 50,000 watts or more. That’s 10 times what a 2020 non-AI server rack would have demanded. That fridge freezer that’s about the same size as a server rack consumes about 200 watts, so that AI server rack is demanding 250 times or more energy than the fridge freezer. In the good old days—that’s pre-2020—a decent-sized data center might be classified as being five megawatt, which meant it had about 1,000 server racks. By the middle of the 2020s, data centers were not talking megawatts anymore. It was now gigawatts. One gigawatts is a billion watts. That’s enough electricity for about two million people. Some of super-mega data centers were demanding up to three gigawatts. There were plans for certain data centers to have their very own nuclear power plant. It’s not sustainable.

Chips and storage would get cheaper, faster and more powerful forever, they said. No limits. Every couple of years, Santa would arrive with twice the power at half the price. They called it Moore’s Law. For what seemed like an eternity—from the 1960s—chips were getting faster and smaller and cheaper. Then in the 2010s, things began to slow and the limits of physics came into sight. This changed a lot. For one thing, it meant even more data centers would be needed based on data growth because if the servers weren’t getting that much more powerful and cheap every year or so, you were going to need more of them for all that exploding data.

They called it Dennard Scaling. As the technology generations advanced, chips would halve in size and in power consumption requirements. You could thus double the number of transistors while maintaining the same power consumption. From about 2005, Denard Scaling stopped scaling quite like it used to. It seems things can only get so small before running up against those basic laws of physics. When the walls in the chip architecture get to a certain level of thinness, stuff begins to leak through. Inefficiencies and constraints emerge.

They called it Koomey’s Law. It proclaimed that “at a fixed computing load, the amount of battery you need will fall by a factor of two roughly every eighteen months.” Great news for battery-driven devices getting smaller and more powerful. Except that after 2000, this law too began to show stress. It was negatively affected, of course, by the slowing of Moore’s Law and Denard Scaling. By those awkward laws of physics.

Toxic optimism is the fevered faith of Silicon Valley, as many still claimed that Moore’s Law was the law of the land. Acceleration forever. Impossible is nothing. Fake it till you make it. There is always a breakthrough waiting around the corner if you throw enough money, materials, energy, water, hype and dodgy practices at it. So, the tech faithful believed like the faithful at a Las Vegas roulette table believed. The frenzy of progress, with things doubling and halving every couple of years, the sheer rush of innovation that was churning up so much material and spitting out so much waste. No time for wisdom, for the slightest reflection or doubt. Caught up in the frenzy of intelligent design. Just doing it. Dragged along by the relentless speed of things.

It all ends as toxic waste. Almost all chips are discarded long before they ever even wear out. Silicon chips are designed from a single-use perspective focused on maximizing speed and functionality for a specific purpose at the lowest possible cost. Chips are expected to have a very short life and nobody thinks about value after that life has expired. Refurbishment, reuse or recycling are not considerations. The old chip is dead. Dump it. Burn it. Bury it. Long live the new chip. Driven by cost reduction concerns, generation after generation, the value of the materials in a chip declines. There is no money in recycling silicon. It is the ultimate throwaway toxic waste.

What goes around does eventually come around. The outsourcing of harm and the pursuit of the very cheapest labor was too wildly successful for Silicon Valley. By the 2020s, China was rising and nearly all the best chips were made in Taiwan. Security and dominance fears grew and a major push began to bring back chip manufacturing to the drought-stricken deserts of Arizona and the lush jungles of nearby “friendly” countries, such as Costa Rica.

“Who wants to carry bunches of bananas rather than go to work in a microprocessor factory producing the world’s most in-demand technology?” Costa Rica president Rodrigo Chaves asked a group of local teenagers. How could any ambitious teenager resist that sort of offer, especially in a country struggling with high unemployment? Who wants bananas anyway when you can have chips? Business-friendly Chaves had been relentless in pushing what he termed “economic reactivation” as his government rolled back environmental protections and refused to ratify the Escazú Agreement, a treaty seeking to protect environmentalists in the most dangerous continent on earth for such activists. “It’s paradoxical that a country as emblematic as Costa Rica, recognized worldwide for its environmental commitment, is only interested in the economic benefits that a company like Intel can generate for its gross domestic product,” Pablo Gámez Cersosimo, a Costa Rican researcher specializing in technology and biodiversity, told me. “In Costa Rica, there is an absence of critical thinking, of national-level debate, about the environmental impact derived from the semiconductor industry.”

The first time I experienced the Valley was around 1999, as one of those gung-ho tech bro entrepreneurs eager to make a fast buck. Three years earlier, I had published a report for the Irish government called Ireland: the digital age, the internet. Nua, a company I had co-founded, had in 1996 won the Best Overall World Wide Web Business Achievement Award from the European Union. It had become one of the first companies in Europe to win significant Web contracts in the United States. When we exhibited in the Javits Center in New York in 1999, it felt like everyone knew our name because of a very popular Internet trends blog we published. We had designed a social media platform called Local Ireland that had received millions in funding from Ireland’s national telecom company. Nua would soon be valued at $250 million by some of Wall Street’s finest investment banks, only to crash, sizzle and burn in the dotcom bust of 2001. I was in Silicon Valey to sell silicon dreams.

This was Mecca for someone like me. I had built up this image of a fantastical, wondrous place, something out of a fairytale, a place with a pathway to a better and more prosperous world, a shining, broad, majestic valley, a pure and clean place great for the environment. I arrived in San Francisco airport and got a taxi south towards San Jose. The relentless traffic was the first thing that hit me. The traffic jams, though, simply fed my expectation. Soon, it would be soon. I began to see low box warehouse-type building after low box warehouse-type building, like a continuous series of Walmarts and Ikeas. I was waiting and wondering when the real Valley would begin.

According to tech futurists like Gregory Stock, the only function biodiversity had was to entertain humans. “There is an immense roster of species,” he noted, “that neither affect nor interest the vast majority of humankind.” This was in 1999 and these tech bro views have only accelerated and hardened since then as, unsurprisingly, biodiversity has collapsed all around us. Not because of some accidents. Because of Valley design and culture. Because of the triumph of technology, chemicals and plastics, of engineers, chemists and scientists. The Valley is not finished yet, by any means. It has big dreams of extending its surveillance capitalism world, while crushing any competition from our environment. By the 2010s, the “best minds” of what had now become the Valley of Pimps and Pushers were relentlessly focused on designing for engagement and addiction. By the 2020s, they were openly calling for ethnic cleansing and the building of a new and more secure Valley away from the working-class riffraff and the homeless, where the tech elite could live out their fantasies of world domination in quiet and comfort, surrounded by big screens of super clean nature.

I asked Aaron Sachs in 2023 if experience had led him to change his thinking since he published his article back in 1999. “I don’t have much to add except to recommend the writings of Lewis Mumford,” he replied. “I recently published a book partly about Mumford, and one of the subtexts was simply his warning that in this country the conversation about HOW we ought to use technology has almost always come too late—long after the technology has been hailed as the next great gift to humankind. Just part of the love affair with capitalistic innovation and consumerism…”

Even in the greenwashed Valley there were limits. By the 1970s, the concentration of toxic dump sites, of polluted water and soil, was concentrating minds. More and more people were getting sick, the US Environmental Protection Agency had been established in 1970, and regulations were tightening. There is opportunity in adversity, they say. Some bright spark wondered why bring these riffraff Mexican and Filipino illiterate women to the vaunted Valley, when you could outsource the harm to their dirtbag country? There, you could keep dumping toxins straight into the soil, air and waterways—in even larger quantities—and nobody would complain, so desperate were they for progress, modernity, development. You could exploit workers too in a way that was no longer possible in a namby-pamby government-interfering USA that was losing its raw capitalist bleeding edge. Didn’t everywhere desperately want to be Silicon—Silicon Isle, Silicon Glen, Silicon Delta, Silicon Heartland, Silicon Forest, Silicon Savannah, Silicon Oasis, Silicon Desert? The bright green silicon machine was marching and conquering everywhere it went. It would now be “Designed in The Valley”, with the manufacturing, worker abuse and pollution disappearing even further into the cloud of poor countries scrambling for crumbs of the future.

It worked, and like a magic trick, the digital warmonger was born and boomed as something greener, something softer. However, impressions cause reality distortions. By keeping things low and out of sight, urban sprawl spread faster than anywhere else in the United States. Those low-lying ‘campuses’ grew everywhere as the Valley experienced almost exponential growth, snarling traffic, putting intense pressure on housing, particularly that for ordinary working-class people. For those poor, stressed, unhealthy female migrant workers, who were still so unfortunately needed in this brave bright green, clean tech world. “By 1970,” writer and philosopher, Aaron Sachs wrote, “San Jose had only 3.2 hectares of open space per 1,000 residents, half of which consisted of school playgrounds, compared to 14.2 hectares per 1,000 people in San Francisco and 28.7 per 1,000 in Washington, DC.” The greenwashing of the Valley was in full flow.

The reality distortions would persist and grow. Behind the scenes of the Valley film set lay a region full of stressed people stuck in traffic, struggling to get by. Poisoned land and water, and house prices were crazy. Long, long commutes, particularly for the poorest workers. Rising homelessness, and starker and starker income divides, as the average working wage declined while the tech bros cluttered the stars with their satellites, while reaping the richest of dividends. By the 2020s, life for those outside the elite was a reality struggle in the Valley. Many were saying they’d leave if they could.

The working class. They couldn’t get rid of them. While manufacturing was declining in much of the US from the 1960s onwards, it was growing rapidly in the Valley. Female factory workers. “In 1970, 70 percent of the production workers in the electronics industry were women, about half of whom were minorities, mostly Mexican-Americans and Filipino-Americans,” Sachs wrote. Cheap and disposable. Headaches, miscarriages and cancers were high due to inhaling the hydrocarbon solvents used to clean semiconductors. Luckily, these poor women were easy to discard without worrying too much about lawsuits. This culture of extreme worker disposability would become a core characteristic of this new Valley. Even the white programmers were not immune. It was made brutally clear to programmers that over forty was over the hill and out of the Valley. Few got rich quick. Many died trying. Making toxic products in a macho toxic, no-worker-rights culture would be refined and later marketed as the liberating bright green “gig economy.” Meanwhile, the tech executives were ever wary. One of their favorite pastimes—yoga and meditation included—were weekend retreats focused on how to bust and crush unions.

Silicon Valley “is fundamentally misleading and ahistorical in its approach,” Aaron Sachs wrote. It doesn’t have a past that it hasn’t buried in some underground leaking chemical tank, and its present and future are constantly being reinvented by bright green marketers and branders, who excel at telling compelling stories of fantasy convenience, innovation and efficiency. Hardcore accelerated innovation is the religion and every traffic-snarled road leads to some vaunted progress. “Perhaps most dangerous is the seemingly concerted attempt of high-tech boosters to inspire scorn for the actual, physical world,” Sachs wrote in 1999. He mentioned futurists like Gregory Stock, who celebrated “comfortable indoor environments.” These tech bros deliberately set out to weaken and ultimately eliminate “the emotional links between humans and the ‘natural’ environment.” Why? Because the physical environment is the key competitor of technology. The more time you spend online, away from Nature, the more money they make and power over you they acquire.

“A lot of that design was about deliberately placing industrial infrastructure out of sight,” scientist Josh Lepawsky explained to me. “Literally putting it underground. Things like chemical storage tanks needed to store the chemicals for the manufacturing process. So, it was a deliberate urban design process, and I think it has been with us since at least the 1950s. Why does it all matter? One of the ways that it matters is that it is very useful for the marketing and the industrial interests out of which digital technologies emerge, that they can trade on these images of being light, green.

Think of all of the metaphors that go with the digital technologies we use, like ‘The Cloud’, for example. This myth of digital as ethereal can be very useful as a way to divert attention from the many classic problems that come along with industrial production, that is the use of energy and materials and the pollution that pretty much always results.” Put the chemical storage tanks and pipes underground. Think about that for a moment. Underground, these tanks were much harder to maintain. They were much more likely to leak, and the chemicals they leaked were some of the most toxic known to man. It didn’t matter. The health of people didn’t matter. The soil and water didn’t matter. What mattered was the branding.

Silicon Valley had other secrets. Like the overall computer industry, it was born from war. Its first design challenge was how to kill more efficiently. The cold and warm wars nurtured the digital seeds, with the Second World War giving birth to the semiconductor industry. It grew quickly, fed by military contracts. Early invention and innovation was focused on improving guidance systems for missiles to make them more effective killing machines. Sputnik and the arms race were the seeds from which the Internet would blossom as a result of generous military research grants, with the objective of creating a network that would be robust enough to withstand multiple nuclear strikes.

In the nascent Valley, the bright green new world order needed to “attract a better class of workers,” as Stanford University business manager, Alf E. Brandin, stated in 1956. They wanted engineers and thinkers, marketers and branders, and fewer of those unkempt working-class riffraff, those illiterate and disposable female migrant workers coming from Mexico and the Philippines to steal the jobs nobody else wanted to do. Or at least that was the impression Alf E. Brandin and his modern marketers wanted to present, and as we all know, when it comes to marketing, impressions count more than reality. Thus, the Stanford Industrial Park got renamed the Stanford Research Park. Companies that located there “had to follow strict building codes, which included ‘complete concealment’ of things like smokestacks, generators, transformers, ducts, storage tanks, and air conditioning equipment,” writer and philosopher Aaron Sachs explained. The buildings were kept low lying, hugging the ground, blending into the landscape, becoming the landscape.

It wasn’t always known as the Valley of Pimps and Pushers. Once upon a time, they called it the Valley of Heart’s Delight. From far and near, families would come on Springtime pilgrimages to participate in the famed “blossom tours” in Santa Clara Valley, California. “Miles and miles of fragrant orchards, spreading in a vista of never-ending loveliness under sunny Spring skies, hold promises of rich treats to come,” a promotional video from the 1940s stated. Up the road, in San Francisco Bay, they found rich harvests of the freshest oysters, while the soil of nearby San Jose was famous for its fertility, overflowing with prunes, apricots, cherries and apples.

Those times would pass. The valley of fruits would become the valley of silicon, while the man who used an apple for his company’s logo would keep a private orchard next to his house to reminisce on Nature and those pleasant times gone past. The soil and water would be soaked in a multitude of chemicals and heavy metals used in the making of silicon chips and other electronics. The e-waste sites would proliferate. Chemicals, ethers and gases used in the chip “clean rooms” would cause all sorts of health issues, particularly for the reproductive health of the mainly migrant female workforce. The high-tech sewage dumped freely in public drains, the leaky underground chemical tanks, laced the environment with cadmium, nickel, and lead, the nitrogen and phosphorous runoff from Green Revolution agriculture. It would all add up. There would be and a rash of deaths from food poisoning. Thus, the surge in oyster restaurants to satisfy the refined tastes of the tech bros of San Francisco could not meet their needs from local Bay-fished oysters anymore. For thousands of years, the local Indigenous had grown healthy and strong eating this abundant seafood. No more. Santa Clara County would move from being known as the Valley of Heart’s Delight to the Valley of Superfund Sites, as it attained the notorious reputation of having more toxic dumpsites than any other county in the USA. Luckily, the Valley had excellent marketing and branding. The pain and suffering of the female migrant workforce would remain well hidden, as would all the other stories of environmental degradation, well covered up by the brilliant shine of the ethereal Valley’s clean and bright green Big Tech brands.

As early as the 1950s, Big Tech began to master the fine arts of greenwashing. It would take an actual green valley and turn it toxic brown while branding it bright green. This new Valley would reflect a clean break from the dirty and polluting smokestack industries of the past. It was to be a digital world, insubstantial , almost invisible, light as a cloud. Its planners eagerly embraced the architecture of seemingly open spaces, campuses and parks, low-rise university-style buildings, soft, green landscaping, evoking sustainability and renewability. Underneath the slick branding, the chemicals bubbled in carefully hidden underground tanks, the gases rose, and the heavy metals stirred. The working environments of the female migrant workers were cruelly filled with cancer-causing gases.