Choking on the Cloud

The Ecocidal Revelations of "AI"

It may have gone out of style for a few decades, but ecocide is back. Yes, we’ve learned to recycle; yes, we’re saving for electric cars and installing solar panels on our roofs; yes, we reuse our bottles and store food in glass whenever we can. It’s hotter these days, but we’ve scaled back the AC. These habits are the fruition of a half-century of warnings, hard work, research, and legislation. But the “AI revolution” is here, and in just a few hours of talking to an algorithm on Instagram you can undo it all – and not even notice.

If you haven’t read the news, you should. In October of 2023, Scientific American announced that, if the “current trends in AI capacity and adoption” continue, the manufacturer NVIDIA could ship “1.5 million AI server units per year by 2027. These 1.5 million servers, running at full capacity, would consume at least 85.4 terawatt-hours of electricity annually – more than what many small countries use in a year.” This increase in electricity, according to a recent article in Jacobin, is “rapidly outgrowing the electric grid while keeping dirty sources of power, like coal plants, operating.” But those companies “have a long track record of arranging for special, discounted rates for their massive power consumption – which means in many cases, ratepayers like you are subsidizing data centers’ undisclosed energy use.” Like most people who don’t pay for what they use, this renders tech companies unaccountable and irresponsible with the power grid. Jacobin identifies a particularly egregious example, though it’s hardly the only offender: “Despite its goals to become carbon-free by 2030, Microsoft’s emissions jumped by 30 percent in 2023, thanks to its recent investments in AI.” Just this month, Google announced that it failed to meet its environmental goals for the same reason.

Any elementary school student can tell you that increased power usage transfers more carbon from the ground to the sky, where it continues to raise the overall temperature of the planet. But these data centers, it turns out, also “suck up substantial amounts of water to cool their servers, and are often located in places where land is cheap – like deserts. Only a few operators report their water usage, even though a fifth of servers draw water ‘from moderately to highly stressed watersheds.’ One paper estimates that globally, the demand for water for data centers could be half that of the United Kingdom within the next several years.” This has invited a lot of states to reconsider their stance on building and subsidizing large data centers, exporting the problem instead: “Latin America, for example, is now seeing a surge in data center development, including near drought-stricken Mexico City, which is hurtling toward a day in the near future when its taps run dry.” Jacobin goes on to cite a paper by Boxi Wu, a graduate research student at Oxford. A glimpse at the “entire supply chain” of the tech industry, “including the rare earth minerals its infrastructure requires and the electronic waste being generated by rapidly advancing chip technology,” indicates that the countries who benefit from this consumerist frivolity – primarily the US and EU – are only reproducing “global economic and political power balances” that are “linked to past colonial dynamics,” as these labor practices and “pollution-intensive mineral mining” take place in “China, Africa, and Latin America.” Citizens of the DRC have recently spoken out against this, equating the tech industry’s unsustainable and exploitative demands on their nation’s cobalt mines with “modern-day slavery.”

To the rumors that “AI” is “efficient” and can analyze large sets of data to reduce emissions, data scientist Alex de Vries offers a clear-eyed corrective. “Every time a major new technology makes a process more efficient, it actually leads to more people demanding whatever is being produced,” he told Scientific American. “Efficiency boosts demand, so boosting efficiency is not really saving energy in the end.” As an example in a separate article, Scientific American offers the scenario of fast fashion:

When an eerily specific ad pops up on your Instagram or Facebook news feed, advertising algorithms are the wizard behind the curtain… Targeted ads push a steady rotation of cheap, mass-produced clothes to consumers, who buy the outfits only to replace them as soon as a new trend arrives. That creates a higher demand for fast-fashion companies, and already the fashion industry is collectively estimated to produce up to eight percent of global emissions. Fast fashion produces yet more emissions from shipping and causes more discarded clothes to pile up in landfills.

This opacity is becoming an industry standard. Because tech companies refuse to share information with researchers, Jacobin reports, “it remains surprisingly difficult to put an exact number on just how much power and water AI might use. Yet that demand is soaring as the technology is tacked on to everything from your iPhone’s operating system to how your car insurance company calculates your rates.” Disclosing information, these companies claim, “harms their competitive advantage.” This lack of transparency should be familiar to anyone who’s even heard of an oil company – the industry that “big tech” seems most intent on emulating in its race to extract as many resources as possible before legislators get in their way. This is exacerbated by a new rush of lobbyists intent on warding away regulation:

A report by nonprofit watchdog Public Citizen found that the number of lobbyists on AI issues mushroomed in 2023, increasing by 120 percent. The vast majority worked for corporate interests, including sixty lobbyists employed by Microsoft, while Amazon hired an additional thirty-five. Last year, Amazon, Google’s parent company Alphabet, Meta, and Microsoft each spent more than $10 million on lobbying for various interests.

This kind of lobbying, of course, may be irrelevant now that an activist Supreme Court has, in striking down 1984’s Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., effectively removed the expertise of agencies from regulatory decision-making and transferred them to a judicial system increasingly under corporate sponsorship and control.

“AI,” so called, isn’t all that new – not that you’d know it. What’s new is the way those in media have decided to talk about it; and with this new information available about its environmental consequences, people are talking. If there is an “AI revolution,” its primary social aspect seems to be this shift in consciousness: With these “generative algorithms,” the tech industry is losing more of its credibility as the self-described utopian center of libertarian philanthropy.

Things were already getting rough. While on the rise since the nineties, the immense visibility of right-wing extremism after 2016 has eaten away at social media’s mass appeal, as well as its long list of false promises. Then, in 2020 and 2021, the social devastations of the pandemic ran an axe through the middle of the internet: You’d either had enough of looking at people on a screen, or the internet was now where you wanted to live (those for whom “touch grass” was coined as a panacea). Now, in 2024, the environmental and labor catastrophes that these corporations have brought to mainstream consciousness have thrown this entire industry of “disruption” into question. The planet, after all, has already been disrupted by reckless corporations that have no meaningful oversight, which makes Silicon Valley’s over-investment in these “learning models” a bluff too big not to call.

It’s more than a single hand, however. The whole game is up. Just as Trump’s use of Twitter to foment American fascism betrayed, in retrospect, the entire idea of social media as a libertarian project to turn everyone who uses it into an isolated, de-politicized consumer, the tech industry’s blitzkrieg of generative algorithms betrays the socio-commercial internet itself as a kind of environmental trojan horse. Despite its benefits, the internet has always been an ecocidal and imperialist project – an invisible extension, as Wu says, of colonialism. Prior to the widespread usage of these generative algorithms – primarily as distraction, in the same way that memes are distracting, but also as a low-quality replacement for copywriters, photographers, and illustrators – “the average internet user’s digital activity generated 229 kilograms of carbon dioxide a year,” according to Jacobin, making the current global internet usage, “about 40 percent of the per capita carbon budget needed to keep global warming under 1.5 degrees Celsius.”

We know now that every time you ask an algorithm to generate an image or tell you a joke, you’re harming the environment, enriching tech billionaires, and destroying someone’s ability to make a living. But what these ecocidal consequences of “AI” have revealed is that, all along, every time we’ve saved a photo to the cloud or uploaded a new TikTok or bought a new iPhone, we’ve been wasting fresh water, adding more carbon to the atmosphere, endangering workers overseas, and enriching the corporations who’ve made it fun for us to destroy our planet and put ourselves and our peers out of work. While the scale has changed, the internet has never been “clean” because even a photograph snapped on a cell phone has to exist, and it has to be stored somewhere. Even though you can’t feel them the way you once felt a Polaroid or a Kodak, those 17,000 photos on your phone are real, as are the metals mined to produce them, the water used to cool the servers that store them, and the corporations continuing to enrich themselves on this data storage. Every time you post a screenshot of some person’s tweet or conversation; every time you make a meme about the president’s age; every time you record a video unboxing some new product you’ve had delivered to your apartment; and every time you write a post on Substack, what you are doing is real. The internet is and has always been real life, and real life is matter, not energy – decidedly finite. It costs to create, and if those costs are hidden from those pushing the buttons and uploading the images, they will be immeasurably higher for everyone else.¹

Like the oil giants who promised that plastics would be recyclable – only to dump them in our oceans and in our bodies – tech companies have learned to obscure the reality behind the fantasy.  That these corporations have spent billions over decades convincing us that all of this is happening “in the cloud” attests to their immense talent for marketing. “AI” is one of the most brilliant PR campaigns of this century, as every single tech company now thinks it’s necessary to have a little widget run by an algorithm tucked away in the corner of its interface, ready to answer whatever question you may have – inaccurately. Meanwhile, “the cloud” is an ethereal, whimsical name for the vast data farms draining the water table and keeping coal plants pumping smoke into the atmosphere despite record year after record year of high temperatures, droughts, storms, fires, and floods.

---

As with any technology, these learning algorithms are neutral. The internet is neutral. Even social media, as a concept, is neutral. What is not and can never be neutral is who owns and administrates these technologies – and who pays for their development.

One of David Graeber’s many laments in The Utopia of Rules is the “broken promise” of the flying car. Raised on the sci-fi of the sixties and seventies, and born at the tail end of a stellar run of technological advancement – most visibly represented by the Space Race, but more practically by the microwave and the birth control pill – the year 2000, Graeber says, was supposed to have felt, for those who’d waited for it, like the future. Instead, in the twenty-first century, most of us are worse off. Despite all the promises of flying cars and colonies on Mars, he writes, “what technological progress we have seen since the seventies has largely been in information technologies – that is, technologies of simulation.”

Graeber is being facetious (a flying car is stupid), but he’s right to pinpoint the inward curve of Silicon Valley technology. Since the personal computer, almost all advancements in consumer technology have happened on the other side of screens. Because of this remove from the “real world,” getting consumers to buy into these advancements has required more marketing than, say, convincing someone in the 1950s to buy a dishwasher. (The dishwasher washes your dishes; but what the hell does Facebook do?) This obfuscation may be part of the point:

A case could be made that even the shift into R&D on information technologies and medicine was not so much a reorientation towards market-driven consumer imperatives, but part of an all-out effort to follow the technological humbling of the Soviet Union with total victory in the global class war: not only the imposition of absolute U.S. military dominance overseas, but the utter rout of social movements back home. The technologies that emerged were in almost every case the kind that proved most conducive to surveillance, work discipline, and social control… Information technology has allowed a financialization of capital that has driven workers ever more desperately into debt, while, at the same time, allowed employers to create new ‘flexible’ work regimes that have destroyed traditional job security and led to a massive increase in overall working hours for almost all segments of the population.

Rather than liberate consumers with household conveniences that allow for more leisure, we are told that screens are the leisure, when in fact they are nonstop work we do for free. Yanis Varoufakis calls this the work of the “digital serf” in the “fiefdom” of the cloud. Social media users (and Amazon reviewers, Spotify streamers, Netflix viewers, etc.) provide the free labor that convinces the real customers – corporations – to pay “cloud rent” to Apple, Amazon, Microsoft, and a handful of other companies he calls “technofeudalists.” The explosion of information technology has in the hands of these companies transformed leisure into free labor that maximizes the value a tiny group of people can extract for letting a corporation access data centers. The old model of consumerism – seducing people to spend money on disposable goods that harm the environment, all in order to maximize their leisure or opportunities for entertainment – seems a forgotten luxury compared to the modern reality: that we now waste immeasurably greater resources to provide a tech giant with free labor, getting little out of the deal outside a brief hit of dopamine (followed by mental illness, as is increasingly obvious – not to mention the collapse of democracy).

If unsustainable consumerism in the twentieth century was fun, in the twenty-first it seems joyless and compulsory: To to be seen; to find a job; to cultivate an audience for your art or your craft; and even, in some locales, to buy basic goods for your family – to do all of this you have to waste water and electricity. This is water and electricity that has nothing to do with the job you’re looking for, the audience you want, the groceries or household goods you’ve ordered.

What’s worse, the more starry-eyed these corporations become about generative algorithms, the more wasting resources becomes a mandatory, everyday part of one’s career.² With more and more “little email jobs” taking place entirely behind screens (often remotely, requiring even more decentralized data duplication, more video calls, more bandwidth, all of which isn’t as “green” as the lack of commute would promise), the foundation for an ecocidal workplace is already laid. This is how “AI” has integrated itself so seamlessly into daily life. The adoption of generative algorithms in the workplace is a globalized, diffused environmental disaster whose lack of spectacle is precisely its danger: It’s a simple yet exponential extension of the extant, banal, and omnipresent activity of adding data to the cloud.

Given the reality of corporate waste (anyone who’s ever worked at a large or mid-sized company knows to laugh whenever that cliché about “government bloat” pops up), the workplace embrace of “AI” risks becoming a bigger ecological catastrophe than an oil spill or a meltdown. While our employers pursue “green initiatives,” the strain they place on the environment only grows – all because they’ve been sold an idiotic fantasy.

---

I used to think “content pollution” was a moralization of a pragmatic, technical problem: that too many images and too many words had overwhelmed us, as a society, and that we were choking on all this extraneous information. It’s a notion that comes from Sontag’s On Photography, where she proposes an “ecology” of images:

The force of photographic images comes from their being material realities in their own right, richly informative deposits left in the wake of whatever emitted them, potent means for turning the tables on reality – for turning it into a shadow. Images are more real than anyone could have supposed. And just because they are an unlimited resource, one that cannot be exhausted by consumerist waste, there is all the more reason to apply the conservationist remedy.

Images shape our perceptions of the world, of each other, and even ourselves; they aren’t reality, but they affect reality. What’s more, they are real. They are also ubiquitous. In fact, the idea of image-making is perhaps as ubiquitous as as the images themselves. The average American may see thousands of images per day, but it’s likely that they’ll create a few of their own – snapping a photo on the way to work, screenshotting a conversation, even uploading a photo to social media (the upload, after all, is its own copy; you’ve just doubled the space a single photograph takes up in the world). Video content is, of course, becoming as ubiquitous. If Instagram has sanctioned photographic activity, turning everything into visual content, TikTok has sanctioned videographic activity: Everything in the world now exists to end up in a video.

The vast majority of these images and videos are sitting on servers as I write this, cooled by drinking water and made instantaneously accessible by, among more palatable energy sources like solar or wind, the literal burning of fossil fuels.³ While Sontag imagined photographic activity as an inexhaustible resource, it turns out that images themselves are consumer waste, and the devastating effects of our consumption are more hidden – more outsourced – than they’ve ever been. There’s nothing moralistic about this, it turns out; and content pollution isn’t a metaphor at all. Content is pollution, and we’re not only polluting with no consequences – no fines, no costs, no impositions, no regulations, no shame – but we’re being invited, on a minute to minute basis, to pollute more, to pollute bigger, pollute faster; we are meant to take pride in the quantity of our waste, measured as it is in likes, streams, downloads, shares, posts, and followers. And now, with the “AI revolution,” we are encouraged to pollute with incomprehensible, joyless, and artless efficiency.

This kind of technology and its haphazard adoption is, of course, endemic to the tech industry’s libertarian ideology. Libertarians aren’t against government; they just want to become the government – and that’s precisely what tech companies have done over the last thirty years. Knowing this, one could assume that the only thing these algorithm-pushers want is to consolidate all resources, all information, and all power into their own hands. The “AI revolution,” then, despite the way our media and our politicians portray it, is not about efficiency or liberation, but a banal and ecocidal coup d’état where the most precious resource on earth becomes a proprietary but potable coolant – which the technofuedalists, of course, will happily provide at a premium.

Like social media, online retail, and every other aspect of the socio-commercial internet, “AI” is not a product; it is propaganda meant to degrade labor, steal intellectual property, destabilize facts, discredit government institutions and regulatory agencies, destroy the environment, and corrode democratic society. What “AI” makes clearer than all previous iterations of this propaganda is, in its total disregard for the environment, the unwavering reality of our widespread allegiance to Silicon Valley technology: a future where you beg men like Jeff Bezos and Elon Musk for a drink of water.

---

---

Thank you for reading! In an effort to incubate a new book, I’m writing a series of essays about technology and culture. Because of this, for the first time since I’ve launched Entertainment, Weakly I am now accepting subscriptions. There is no paywalled content: everything I post, anyone can read. But if you do have the cash and you’re feeling generous, a supportive subscription at any amount is always appreciated – and helps me keep writing. Thank you so much!

1. It’s not a tit-for-tat analogy, but a single Polaroid costs the photographer approximately $2.00 to produce. Cost has always deterred waste.

2. As one small example, a former client of mine insisted on hosting our meetings over Zoom so they could record the conversation, feed it into an algorithm, and spit out a “generated” summary of the meeting – a summary that was so inaccurate it was fundamentally useless (it occasionally mistook me for the client). All that water, all that carbon, for something neither of needed or could even use.

3. Every influencer entreating us to live more sustainably, to eat less beef, to use canvas bags, could follow their own advice and delete their water-wasting, carbon-spewing account.