The Antagonist

On extraction, reciprocity, and what we're really fighting

Sometimes when I've been developing this project, my co-director Patricia has asked me: what is the antagonist of our film? It can't just be all optimistic. And my answer has always been a bit vague.

On some level, this is a film about the rise of AI and the massive worldwide construction of data centres, which is accelerating at unprecedented pace. The environmental damage is enormous. Data centres consume huge amounts of water — both for cooling and, in the case of semiconductor fabrication, for making the ultra-pure water required to clean silicon wafers. TSMC alone uses over 150,000 tonnes of water per day at its facilities in Taiwan. During the 2021 drought, the worst in over fifty years, the government rationed water for farms and households while dispatching fleets of trucks to keep the chip factories running. In Spain, data centres proposed for drought-prone regions of Aragón are being fast-tracked through planning processes with minimal deliberation. In Memphis, Tennessee, xAI installed dozens of gas turbines to power its supercomputer without permits, in a neighbourhood already suffering from decades of pollution. Across whole regions, this expansion is contributing to drought, water shortages, energy outages.

So it might seem obvious that the antagonists are the big tech companies ruthlessly driving this destruction. But I'm not sure that feels right to me. I think the real enemy is something more diffuse — the idea of singularity, in the broadest sense. The process by which everything is turned into capital. The tech companies are as powerless to this force as the rest of us, in a way. They are expressions of it, not its authors. There is a certain irony in the fact that we spend so much time worrying about the arrival of an uncontrollable superintelligence, when the extractive economic system already behaves like one — optimising relentlessly, expanding without regard for the communities and ecosystems in its path, reshaping the world according to a logic that no individual directs or fully understands.

That logic sees the river as the sum of its parts: resources to be measured, diverted, consumed for economic value. This simplicity is powerful — it builds semiconductor factories and data centres and supply chains of extraordinary efficiency — but it simply misses the vast majority of the river's value, the value that cannot be captured by these metrics. The river as habitat, as cultural memory, as a being with whom communities have maintained relationships for thousands of years, as a system whose complexity we are only beginning to understand — none of this shows up on a balance sheet. And the river itself is processing information at extraordinary sophistication — integrating geology, hydrology, biology, and climate across timescales far beyond anything we can consciously track. So you have two superintelligences at work: the ecosystem, ancient and reciprocal and sustaining; and the extractive market, recent and optimising and depleting. The antagonist of our film is the collision between them.

There is, worldwide, a growing recognition of the need to build — or to remember — other modes of relating to our environment, and other ways of recognising value. The Rights of Nature movement is one expression of this. It's an attempt, learning from Indigenous cultures and their modes of stewardship, to bring some of that reciprocity into our legal and economic structures. Rivers have been granted legal personhood in New Zealand, Colombia, Ecuador, Quebec, and most recently by the Colorado River Indian Tribes. In the UK, Fang-Jui Chang, a researcher at Dark Matter Labs and one of the key figures in our film, has been appointed as one of the first official River Representatives for the River Roding. These are efforts to create space, within the existing system, for the river to be heard.

A powerful example of a different orientation comes from Taiwan. Fang told me about an Indigenous Paiwan ritual: when the Paiwan take stones from the riverbed to build their houses, they first ask the river for permission, and promise to give the stones back when they're done.

This simple act of asking permission feels profoundly different from the way our society generally treats natural systems. It is especially stark when set against the way the tech companies are going about "permissioning" — in heavy quotation marks — their new constructions. There is little deliberation and little in the way of reciprocity toward the people, the plants, the animals who already live where these data centres are proposed. In Spain, as we learned through our research with the activist group Tu Nube Seca Mi Río, the tech companies often sway the political process by providing expert consultation for free to the politicians who have the power to grant them planning permission. The data centres in Aragón have been fast-tracked well beyond the usual processes, avoiding much of the normal machinery of deliberation, consent, and consensus-building.

The Tayal people of Hsinchu County in Taiwan live inside a version of this same dynamic. Their ancestral territories surround the reservoirs and rivers that feed the Hsinchu Science Park, home to TSMC and the heart of the global semiconductor industry. Research in Jianshi Township has documented how Tayal systems for governing water persist as living practice — rooted in communal custodianship, in a worldview where water itself has agency in the governance arrangement. These systems are not failing. They are being overridden by a system that operates on entirely different assumptions about what water is for.

And there is something else I learned in Taiwan that stays with me. The Cepo' Pangcah — an Amis community in eastern Taiwan — have a song called milikir, which means something like "shooting the rapids." It was sung when travelling by raft along the Siuguluan River. The song has no fixed lyrics, just vocables and improvised words — naming the landscape, encouraging the crew, joking to keep fear at bay. One woman sang it from a zodiac to remind me of the shared labour that the river had always organised. An artist sang it during a typhoon, pleading with the river for his life. A songwriter recorded it for an album, carrying the ancestral relationship with the river into a recording studio and from there to listeners who may never have stood on its banks. The song is a technology of relationship — it carries the river into the singer and the singer into the river, and it insists that the river remains central to the Pangcah world even as everything around it changes.

This extractive model — the one that treats rivers as plumbing and landscapes as sites — goes far beyond the case of data centres and water usage. It is the same model driving the fundamental climate collapse and ecological breakdown we are seeing across the world. The flattening of complex living systems into resources to be optimised is not a side effect of the way we do things. It is the way we do things.

I don't know if the river AI that we try to imagine in our film really gets us all the way to a solution. But thinking about tools that could help us, at scale, build a relationship with the ecosystem — a relationship built on listening and communication rather than extraction — feels like a gesture in the right direction. The Paiwan ask the river before taking its stones. The Pangcah sing to the river as they travel its waters. Maybe the question for us is what our version of asking and singing looks like, built with the tools we have now, adequate to the scale of the damage we've done.

Back to 'Beneath A Hundred Streams'

Back to Writing Index