On March 25, Bernie Sanders and Alexandria Ocasio-Cortez introduced the Artificial Intelligence Data Center Moratorium Act . In his speech introducing the law at a press conference, Sanders explained the thinking behind the proposal , which would put a stop to the construction of new data centers until federal regulations are in place. “Today we are announcing legislation to impose a moratorium on the construction of new AI data centers until strong national safeguards are in place to ensure that AI is safe and effective,” he said. The bill lays out a regulatory framework that includes greater protections on privacy and civil rights, restrictions on AI exports and assurances that economic gains will include American workers, not increase electricity prices and protect the environment.
Without question, these are important issues, and the AIDCM has received enthusiastic praise from the American left. Jacobin’s Nicholas Liu opined , “Only federal legislation stands a chance at leashing a monster of this size.” Jim Walsh of Food & Water Watch, the first national organization to demand a moratorium, said data centers must be halted “so that states and communities have the time needed to properly consider their own futures.” While the moratorium builds on nationwide activism to further open up a space to contest Big AI, it does not think big enough. Specifically, it attempts to address AI harms, while at the same time accepting the corporatist, neocolonial framework of the AI economy and Big Tech in general. This can be seen in closer looks at the bill’s provisions, which are far too weak to stop the harm threatened by the rapid data center buildout.
Consider the AIDCM’s call for a federal review of products before they hit the market to ensure they respect “privacy and civil rights.” In theory, laws can be passed that tell Big Tech corporations and employers not to create or deploy AI that spies on workers and consumers, and to ensure their products do not discriminate. Yet as long as AI is owned and deployed by corporations, workers and citizens will always be threatened by “turnkey totalitarianism,” a situation where restrictions on totalitarian powers baked into these technologies can be turned back on should the government change its mind. If we lived in a world of equals who collectively owned and managed the technologies, it’s unlikely they would put themselves under surveillance or develop tools for discrimination. As Eric Hughest, author of “A Cypherpunk’s Manifesto” put it in 1993, “Privacy is necessary for an open society in the electronic age. … We cannot expect governments, corporations, or other large, faceless organizations to grant us privacy. … We must defend our own privacy if we expect to have any.”
Or consider the AIDCM’s call for the “economic gains” of AI to be shared with the American working class, in the form of policies to prevent job displacement and ensure the creation of union jobs at prevailing wages. Under this scenario, Big Tech corporations and their billionaire CEOs retain control of AI, while workers will be left to wage Sisyphean battles against them for better wages, job security and so on. It does not think big enough. The AIDCM’s loudest silence is on the matter of digital colonialism. As I argued with comprehensive data in my book “Digital Degrowth ,” the U.S. has unipolar dominance of the global digital ecosystem. (China isn’t even close.) The U.S. has substantial market share or dominates social media, search engines, operating systems, streaming media, email, online advertising, cloud computing, semiconductors, investment in AI data centers and more. Chinese open-weight models, advances in robotics, dominance of some tech minerals and a few products like TikTok are far from challenging American tech hegemony. As we will see below, this point is the central connection between AI and the environment.
On that issue, the AIDCM repeats the received wisdom that the direct footprint of AI data centers poses a critical threat to the environment, with reference to carbon emissions, water use, noise, thermal outputs and so on. Attention to AI’s impact on the planet is all to the good. But according to middle-of-the-road forecasts by the International Energy Agency, data centers are set to consume just 3% of global energy by 2035 (as the AI boom tapers off) and emit 350 megatons (Mt) of carbon (or 500 Mt on the high-end trajectory). Even with 70% of data center capacity used for AI by 2035 (the high-end estimate), we’re talking just under 1% of global greenhouse gas emissions from AI on the upper-bound projection. Water accounts for even less: Agriculture accounts for 70% of freshwater withdrawals and 90% of freshwater consumption (compared to 0.1% consumed by data centers). Golf courses use about as much as many large (hyperscaler) data centers, and there are 16,000 in the U.S. (compared to 5,000 or so data centers with an estimated 600-700 hyperscalers).
This does not mean there are never harms at the local level (as in Memphis , Tennessee), or that people shouldn’t call for data center moratoriums. It’s that the thinking and action leaves out the real elephant in the room: that a just transition is incompatible with the interwoven ills of capitalism, American hegemony and the overuse of the planet’s finite resources. Indeed, from a digital degrowth perspective that acknowledges the limits of global material resource use, Big Tech is central to the environment, but not because of data center footprints. It is central because the tech economy concentrates the planet’s finite wealth between and within countries and digitalizes the capitalist suicide machine. Data centers — whether built for AI or not — are neocolonial inequality machines.
Consider this: The U.S. is home to 4% of the world’s population, yet it has about a third of the wealth and almost half of the financial assets. If you believe in equality, as socialists do, then that should be 4% across the board. America also holds the largest environmental debts (carbon and ecological). Thanks to Big Tech, American billionaires, millionaires and even tech workers (earning a median of $300,000 per year) are benefiting from the corporate ownership of AI. Yet on a planet of 8 billion people, there is only $16,000 per head. There’s simply not that much to go around, and those salaries could never sustainably scale up without destroying the planet. Something has to give.
Big Tech digitalizes capitalism broadly, including some of the most destructive industries (industrial agriculture, oil exploration and production, fast fashion, consumerism, etc.). Along the way, it dumps electronic waste on the poor. Americans reportedly junk as many as 150 million phones per year, and much of its e-waste is illegally exported to poor countries. In Ghana, hundreds of thousands of tons of used electronics are dumped annually at waste sites, exposing workers to toxic chemicals. This received international attention when the press exposed e-waste dumping in Agbogbloshie where poor workers foraged through toxic e-waste to salvage minerals like gold. While the government eventually shut the scrapyard down, other sites for illegal waste have cropped up. The United States refuses to ratify the Basal Convention, which restricts the movement of hazardous waste across borders.
Big Tech also strengthens the U.S. empire by advancing military tech (including U.S. allies), and providing it control over information flows.
The Sanders- AOC proposal does not touch on any of this. Like the Green New Deal , it fails to recognize that the United States is a global minority that objectifies, extracts and exploits from the world’s people and nature more so than any other country. It fails to recognize that the United States is a global minority. To put this another way, the AIDCM and state-level moratoriums simply demand minor capitalist reforms that will at best erect speedbumps on the way to Big Tech’s buildout. If we want human rights, equality and sustainability, then we have no choice but to oppose digital capitalism and colonialism altogether. As we race full speed past 1.5C, we get closer to triggering catastrophic tipping points. There isn’t much time left.
Instead of calling for laws to make Big Tech slightly less tyrannical, we should call for its abolition. Instead of vague requests for Big Tech to share the wealth with Americans , we should call for AI that produces global equality. Instead of focusing on the direct footprints of data centers, we should be calling for digital degrowth.
If Sanders-AOC want to claim the moniker “democratic socialist,” they should push for actual socialism, which is about worker and community ownership of the means of production, universal class abolition, human rights and social equality. Instead, Sanders-AOC propose a social democratic capitalism that gives minor concessions to ordinary Americans . Instituting a moratorium — a brief pause — on Big Tech’s neocolonial land grab doesn’t work for workers, it doesn’t work for the Global South and it doesn’t work for the environment. A major transformation would resemble something like a global Digital Tech Deal as an organized plan to phase out Big Tech and socialize the digital economy. Like antitrust , a mildly reformist moratorium will give the false impression of change, while we accelerate full-speed toward an environmental catastrophe caused not by data center footprints, but by the larger reality of a world-capitalist system with the U.S. empire at its core.
The post Is the AI Industry Reformable? appeared first on Truthdig .