Brian Merchant’s piece on the data centre rebellion on his Blood in the Machine substack is worth reading in full and carefully. It’s a fascinating and nuanced look at who profits from the arrival of data centres in small communities and where US lawmakers and politicians are going with policy.
It also positions itself politically as further to the left than the more shall we say liberal views of the Jacobin when it comes to Democratising AI or treating it as a public utility. These are arguments for making do and moving on with the problem, which are at once pragmatic but don’t get to the heart of the personal matter around AI. What do individuals really get out of where we are in the second quarter of the twenty-first century? What’s in it for us?
Politics, policy and profits versus what happens at a local level are very different things. And while Merchant claims that his post is not a snippy take-down, it is a way of criticising the ivory tower and often alarming rhetoric in the more centrist think pieces, while also being somewhat distanced from the actual reality of AI in the here and now.
Data Centres, Always There
Data centres are not a new problem, but we are actually closer to treating AI as a utility than any of us suspects. The data centres in your area are already in place or planned. The villages that are being removed to make way for power stations, such as Moerdijk in the Netherlands. Things are happening with or without our say-so. It’s just happening, but we’re also complicit in our demands.
Global data centre capacity grew fivefold between 2005 and 2025. Double-digit annual growth happened every year from 2018 onwards, and 2019, before the AI boom, was actually the fastest growth year on record. The communities now fighting Oracle and OpenAI infrastructure projects were already facing this before ChatGPT existed. The energy consumption, the water use, the noise and the land, all of that was coming regardless. AI has made it faster, more visible, and more politically charged. It hasn’t created the underlying dynamic.
So, while I applaud Merchant as someone who has been to the meetings and as a journalist and writer who obviously cares passionately about being on the right side of history, his arguments, I believe, need to go further. What’s in it for us? We already use it. Is this in any way related to a social class thing, or is all of this justification simply driven out of fear of losing the finer, more human things in life? And is that what we really care about?
AI is a Distraction, Progress is Real
This matters because framing an argument as an AI fight gives both sides a convenient exit. Opponents can position themselves as anti-AI rather than anti-extraction – the industry term for stealing IP to train AI on it. This is a much harder position to sustain as AI becomes more embedded in daily life. Proponents can argue that better AI governance solves the problem, but the problem is actually structural. Infrastructure costs are borne locally while profits accumulate elsewhere, and that’s true whether the servers are running language models or just storing your Netflix queue. However, there is also a significant upside for everyone (globally) in that AI solves problems we had to pay people to solve.
And while we can worry about where it’s all going, real commercial work is now catching on to AI’s actual power. In small businesses, we can get instant feedback on our contracts, our accounts and our marketing efforts. We are already cutting other people out of our lives through technology. Real people are losing work because they can’t compete with an AI model in your local (or not so local) data centre.
Individually, we are already ensuring that AI’s impact benefits us. It’s already democratised in some ways, if you compare it to the internet boom or any other technology-led social change – it’s understood and used by those who have the time, money and need for it first.
Where are we going?
Before we get too comfortable with Merchant’s argument, it’s worth asking who created the demand in the first place. The data centre in your community exists because you use the cloud, because your business runs on AWS, because you asked an AI something this morning.
I’m not anti AI. I work in software. I’ve spent thirty years inside these systems. I use these tools. But I am deeply sceptical of the idea that the people building this infrastructure have the interests of the communities hosting it anywhere near the top of their agenda. The flip side being that anyone who works in the information business, which tends to be most businesses these days, fundamentally has a vested interest in more AI. It’s its own economic lever.
Merchant’s conclusion is the right one: the data centre fight is where some of the most promising democratic engagement with technology is actually happening, and the left should be in those city council chambers rather than debating whether the movement is ideologically pure enough. The question of who bears the cost and who captures the benefit isn’t an AI question.
It’s the question at the heart of Human Software, too. Not about whether the technology works and for whom, but what it costs the people who keep it running and what it costs not to run it.