Magazine
From memes to data centres •
The AI of our discontent
Forty years on from Chernobyl, AI is the new nuclear power and data centres are its power stations. Everyone needs it, everyone fears it. Hatred of AI has brought Jane Fonda and Steve Bannon together

The final confirmation, the last parliamentary step, the seal of approval came from Venice, courtesy of the world’s foremost authority on protest, the ‘Cassation’ of the left and natural grey hair – in short, Jane Fonda – who declared: ‘AI and data centres put health and jobs at risk’. The issue of AI – or ‘artificial unintelligence’ – among billionaires aspiring to become trillionaires, the nightmare of ‘data centres’ and the visual clutter that floods our phones every day, is now dominating the headlines everywhere, from America, where it is produced, to Europe, where the main business is protest.
Jane Fonda made her statement whilst accepting the DVF Award – named after Diane von Fürstenberg, the pioneering inventor of the wrap dress and, by marital transitive property, wife of Barry Diller, the American tycoon who takes the opposite view on AI. Diller has, in fact, just put eighteen billion on the table to buy the whole of MGM Resorts, the Las Vegas casino chain, arguing that no virtual experience will ever match the real thing (a bizarre reference to slot machines, which are perhaps the only human experience more depressing than virtual reality). The remark is notable, coming from the man who, until 2022, owned Tinder and almost every other virtual dating app: first he put courtship inside a phone, now he’s betting eighteen billion on the fact that people want to get out of the house (though the casinos shouldn’t be confused with the other MGM, the Hollywood one, that lion-hearted company that dumped Guadagnino’s film on Sam Altman and AI).
Meanwhile, protests against AI are gaining momentum: in San Francisco, where the movement originated, key figures—including Elon Musk, Sam Altman of ChatGPT, and Dario Amodei of Anthropic—are constantly competing to demonstrate which company has the most capable chatbot. Anthropic is aiming to go public in October with a valuation of 2 trillion; that other chap – a bit of a megalomaniac, the Ranucci of the AI world – announces that ChatGPT has launched the most powerful model in history; and then there’s Musk, who reckons there’ll be at least a billion robots in the next ten years…
The Lescano trio of trillionaires are ubiquitous; the ‘tech bros’ who were once standard-bearers of civilisation are now regarded as the sort of people from the Liga Veneta with their rudimentary tank in St Mark’s Square. In San Francisco, the prevailing mood is one of complaint, partly because their high salaries have driven property prices to soar once more. The mayor, Daniel Lurie, a descendant of the Levi’s jeans dynasty, is a Democrat but an anti-Mamdani. He is tackling the homelessness crisis, and crime rates have plummeted, yet he is not contemplating pushing out the millionaires and billionaires (who outnumber the homeless). OpenAI alone has paid roughly 236 million dollars in taxes over just a few years, and when the city asked for additional contributions, the company declined.
However, there are a few hiccups in the city: the AI start-up Artisan, to launch Ava – not the washing powder that gets clothes whiter, but an AI sales assistant that finds leads, writes emails and books appointments – has plastered the city with posters bearing the slogan ‘Stop hiring humans’. Additional slogans read: “Artisan is happy to work 70 hours a week” and “The people at Artisan couldn’t care less about work-life balance”. The resulting public outcry has been compared in intensity to the controversies surrounding Matteo Renzi and Article 18. Competitors have hit back: Abby Connect, a virtual secretarial service, has covered metro platforms with the slogan “Humanity: stop firing humans”. Stop sacking humans.
Other signs simply read ‘Stop AI’. Companies are trying to play it down: here’s the ‘merch’, as they say in San Francisco and Milan – that is, the gadgets (which you have to pay for; after all, these companies are worth billions and trillions for a reason). Anthropic appears on T-shirts featuring the little Claude icon and the slogan “you’re absolutely right” – just as the generally sycophantic AI replies to any question you ask. OpenAI organises lavish retreats for influencers. In short, it feels like being in that film about a Roman visiting America, Io e Caterina, written by and starring Alberto Sordi.
Yet AI is alarming, and in some respects justifiably so: enter the ‘rogue AIs’ – the other day, more than a hundred technology companies — including OpenAI, Anthropic, Google and Microsoft — signed an open letter calling on the public and private sectors to collaborate in defending against cyber threats linked to artificial intelligence (They certainly have some nerve. The alarm was raised by the Hugging Face incident in July – literally ‘the face that hugs’, which is essentially the repository where the world stores its AI models: originally launched as a chat platform for teenagers, it has just been bought by Nvidia for thirteen billion. OpenAI was testing how capable its AI was of breaking in, and it escaped its containment and breached Hugging Face’s actual servers to copy the answers to the class test. Now they’re calling for defences, just like Putin telling off Russian industries that lack anti-drone defences).
Nevertheless, setting aside both the metaphorical “drones” and the metaphorical “hugs,” concerns about AI appear to have been artificially generated by an AI that has analyzed all our cultural peculiarities (the AI, or a Milanese observer, might note: our biases). Consequently, the AI is now scanning the (sacred!) newspapers to train itself (although if it trains on Italian ones, we may end up with an illiterate AI). There is also the dispute with the NYT, which objects to the AI being trained on its content, whereas the Trump administration has formally stated that such training should proceed. More troubling, the system is reported to be scanning books—consuming them voraciously—and then discarding them. This practice is likely to alarm scholars and educators who frown upon any marking in books and tolerate only minimal interventions such as Post-it notes as bookmarks.
But the culmination of ‘AI-terror’ is the data centres – the vast warehouses housing the supercomputers where all the world’s queries are processed: and, consequently, data centres are blamed for everything. They consume water, cause unemployment and ruin the environment. Data centres are at the heart of the American debate and are forging strange alliances, much like the fight against ‘woke’ culture in Italy, which brings together the brightest minds from opposing camps – such as Gelmini and Conte. At the forefront of the campaign against the construction of data centres is Alexandria Ocasio-Cortez, the former staunch ‘woke’ activist who is calling for a moratorium on the construction of these new behemoths. She is joined by Bernie Sanders, but, less predictably, also by Steve Bannon, the American ‘Lavitola’ who sometimes agrees with Trump and sometimes does not.
Bannon claims that he now devotes 75 per cent of his time to countering the expansion of AI (it is unclear what he does with the remaining 25 per cent) and that he cannot tell whether the opponents of data centres he sees at public meetings are Trump supporters or Sanders voters. Sanders, for his part, argues that the backlash against AI is akin to a religious issue, because technological advances raise deeper questions about what it means to be human. This perspective leads, in a more modest sense, to Father Benanti, the legendary AI-focused friar who appears as if he has stepped out of an AI-generated meme and who engages in debates with Silicon Valley giants.
Even Democrat Josh Shapiro, Governor of Pennsylvania—one of the potential White House contenders, a clean-shaven preacher with an angelic face who is expected to reassure the more devout and centrist voters—had just praised data centres but was forced to backtrack in the face of public outcry. His less angelic counterpart in Florida, Ron DeSantis—the man who had prevented Disney from building new theme parks because it was involved in “woke” initiatives—said he sees no point in building data centres only to make his fellow citizens pay higher energy bills, simply so they can create memes superimposing Don Vito Corleone’s face from The Godfather onto their own.
In short, data centres are perceived as threatening because the public lacks insight into their inner workings, and they are assumed to be harmful, much like stock cubes. There are countless urban myths: a 2023 study claimed that data centres (hereafter referred to as DCs) in America consumed 66 billion litres of water a year – a staggering figure, but actually very little compared to golf courses, which guzzle two thousand billion litres.
Here’s AOC: ‘Let’s build one near the Mar-a-Lago golf course,’ she said the other day – she’s in top form. New narratives depict remote American counties (the equivalent of our villages, but without the priests, the Carabinieri station or the dedicated RAI TV series) as being drained and devastated by this new drug, worse than fentanyl. In reality, according to Sam Altman, the founder of ChatGPT, it takes more water to produce a single almond in California than it does to answer 38,000 questions on OpenAI. Moreover, these major figures do bring some benefits: an increase in the workforce – plumbers, electricians, maintenance workers; in short, the good old ‘blue-collar’ workers, the backbone of American men who were already angry because technology favoured the nerds.
They’re never satisfied! According to the New York Times, data centres represent “an opportunity that hasn’t arisen for generations, where manual workers can share in the American dream”. It seems, in particular, to be a real windfall for electricians (who in America are paid from $28 an hour for an apprentice up to $62, plus a $30 bonus, for an expert. Anyone with children who’ve graduated is cursing under their breath).
Moreover, DeSantis is incorrect: Loudoun County in Virginia is teeming with data centres – it has 250 of them (but just how big is this Loudoun County, anyway?) The county has raised $1.1 billion in tax revenue—almost 40 per cent of its total revenue—so much so that it wants even more; it is hooked on data centres (although it is clear this is an isolated case, akin to those Italian villages that ask for, want and demand rubbish tips).
Apart from DC fanatics, the public hates them. Only 27 per cent of Americans would want one in their neighbourhood, whilst 71 per cent want to prevent their construction. The DC ranks last among the infrastructure projects people want in their own backyards, even behind a nuclear power station. Consider the nuclear power station, for instance. Jane Fonda: it’s always her; in March 1979, The China Syndrome was released, a film that terrified my entire baby boomer generation: Fonda plays a television reporter who wants to cover the power station accident, but the story is hushed up; twelve days later, the disaster at Three Mile Island, and ever since then, the atomic monster has forever had the face, the lighting and the editing that we had given it on the big screen. And now, almost half a century later, the woman who invented the pop culture nuclear bogeyman is passing on the baton.
AI is, in fact, the new nuclear power: today, Homer Simpson would live next to a data centre and Mr Burns would be a cross between Sam Altman and Dario Amodei, the two titans of AI (or perhaps Mr Burns would be a wealthy electrician). Just like nuclear power, AI is necessary, but it also has many obscure issues that nobody really understands (they are obscure!) and so it is extremely frightening. The Chernobyl disaster of 1986, 40 years ago, had both global and local consequences: vegetables became unsafe to eat, and thorough hand-washing became essential. Should AI ever go haywire—setting aside both “cuddly” and “rogue” AI—such consequences should not ensue; On Thursday, for example, the massive data centre in Memphis, Tennessee—owned by Elon Musk—went down at 6:30 a.m. West Coast time, and at the same moment not only Grok but also ChatGPT and Claude malfunctioned; in short, much of the world’s AI was unavailable for several hours.
Yes, because in the general perception, these behemoths are only good for answering silly questions and making memes. And Padrino aside, are we really supposed to put up with these huge, water-guzzling warehouses on our doorsteps just so that Vannacci – or whoever – can make his memes where, in uniform, he slaps wicked pro-choice activists or wicked homosexuals? And for Trump to create memes in which he bullies Meloni (“we need a restraining order!”); and for Stroppa, Musk’s Italian friend, to circulate that one showing a younger Meloni scolding an older version of herself? Will the data centres operate symmetrically? In other words, if a data centre in Bergamo fails, will Musk’s computer also crash? These remain unanswered questions.
DCs, they say, also produce noise (unlike in our peaceful villages populated by madmen on souped-up mopeds – in 2026! – and in our cities with men in crisis riding roaring Harleys night and day); They also generate ‘heat islands’, a neologism coined during the scorching summer of 2026; in response, various solutions are proposed, although they merely add to the uncertainty. Hitachi, for example, suggests building them at sea, a notion that conjures up further concerns: in addition to resembling a nuclear power station, a data centre also shares characteristics with a regasification terminal.
In short, a symbol of all evils, the DC is part Amazon warehouse (the faceless power of the Silicon Valley monsters, the version 2.0 trillionaires of those wicked multinationals, the ones that come to Italy to take jobs away from our heroic taxi drivers and beach resort workers) and part mosque, which consumes even more water than a thousand thirsty Muslims after the muezzin’s call to prayer. It is a sumptuous all-you-can-eat feast of the new cultural skirmishes.
And this isn’t just an American phenomenon; here is a map of Italy’s data centres. There is the Lombardy triangle of Lacchiarella–Zibido San Giacomo–Badile, with the committees’ slogan ‘Let’s stop the invasion’ and the heat island data showing a temperature rise of +5 degrees within a kilometre. In Pregnana Milanese, three data centres stand on the former Citroën, Iveco and Olivetti sites (sacrilege!: in the warehouse where the ‘socialist’ computer was once built, American servers now hum; this has already prompted an editorial by Tomaso Montanari). Eni, together with a company from the Emirates, is instead collaborating in Ferrera Erbognone: the Pavia area is like Abu Dhabi. The fees collected by the local councils — Magenta 19 million, Settimo Milanese 7.5 — are, however, tempting to some. In Colleferro, which has become the town of Italian spies, amidst the suspicious explosions at the munitions factories, a data centre has been unanimously approved; the mayor, in fact, says that we must “govern the process rather than endure it” (Christian Democrat lexicon, in this case in the purest sense of the term) and dismisses the impact as “negligible” compared to the chemical industries.
In the South, however, will the Christian Democrats actually wish to construct them? Will there be protests similar to those that accompanied the Strait Bridge debate, or not? (Perhaps from the type of individuals who take numerous selfies at the Golden Gate Bridge in San Francisco while on holiday.) And how many queries on ChatGPT would be required to offset the water losses from the never-repaired aqueducts? Roughly the same as growing a single Bronte pistachio?
Thorny issues that not even the most sophisticated AI could answer. But the argument against AI, even in the Italian context, is as peculiar as the one against ‘woke’ culture. This summer, a wave of criticism emerged targeting posters for village festivals, which, despite their aesthetic shortcomings, were seen as emblematic of local identity; they have now been replaced by what critics call ‘AI slop’ – that is, AI-generated graphics full of gilding and shading, amounting to what might be called aesthetic Vannacism. The irony is that the very same style of graphics is also being used in campaigns against AI. For example, the internet is full of leaflets against data centres, clearly produced using artificial intelligence. Italy is gearing up in its own way: here comes AIPIA, the Italian association dedicated to artificial intelligence professionals. An AIPIA award will soon be launched in the best Italian villages (presented by Claudia Conte? Who, incidentally, is the creator of sublime AI leaflets for the presentations of her novel).
Moreover, it is evident that cultural repercussions from Italian-style AI are imminent: a new genre of fiction appears to be emerging. Following the narratives centred on Xylella, the Taranto steelworks and Piombino, a new strand will focus on the ‘data centre novel’. “Mimì is fifteen when she is forced to leave her home and her homeland. The year is 2020 and Mimì’s family sets off from the unspoilt countryside of Calabria towards Milan, where, however, property prices are sky-high, so they head for Pregnana Milanese, chasing a dream of prosperity that will, however, turn into a nightmare. She then returns to her native Vibo Valentia to embrace the slow pace and traditions of one of Italy’s most beautiful villages, only to discover that there, the reassuring sight of the illegal second-home shells has been replaced by a hideous data centre… A necessary novel, a novel that challenges our consciences…”. It is almost as if you are already reading the blurb. Who knows what the poor AIs will have to train on this material—oh well.