Marx in the Machine Room: How AI Proves That Matter Still Rules Mind
by Jaffa Levy
·
Artificial intelligence is presented as the triumph of digital “mind.” But if you want to know what truly drives it, don’t look to philosophy seminars. Look to Nvidia’s chips, hyperscale data centres, electricity contracts and the private jurisdictions built around them. The ideas are decorative. The infrastructure is the fact. And in that world, Marx not Hegel remains the sharper guide.
Hegel’s dream, rewritten as a marketing deck
AI is advertised as the next rung in humanity’s ascent: a new stage in the unfolding of mind, creativity, freedom and “alignment with human values.” It is a turnkey morality tale: thesis (progress), antithesis (fear), synthesis (managed inevitability). A tidy Hegelian arc repurposed for investor calls and summit speeches.
But this choreography hides the machinery. Spirit does not move through history by way of 700-watt GPUs or $80 billion annual data centre budgets. The idealist veneer distracts from the industrial core.
The AI story is a cathedral built over a factory floor. Listen to the hymns if you want comfort. Study the factory if you want truth.
Marx in the machine room
A Marxian lens begins with questions that actually bite: What are the productive forces? Who owns them? Who works on them? Who extracts the surplus? Once applied to the AI stack, these questions flatten every metaphysical claim in their path.
The productive forces of AI are unmistakeably physical: Nvidia’s H100 and H200 GPUs drawing hundreds of watts each, mounted in dense racks inside giant campuses consuming 20 to 100 MW apiece. The data is scraped from billions. The labour is stratified: well paid engineers above, precarious annotators and moderators below. The ownership is concentrated in a handful of firms brokered through capital-intensive cloud empires.
AI is not “thinking.” It is electricity, silicon, labour and capital arranged in a profitable order.
The general intellect, enclosed
Marx foresaw a world in which society’s collective knowledge its “general intellect” would be embedded in machinery. He did not foresee that it would be legally enclosed by private contracts. AI models are built from our language, our images, our code and our behaviours. The inputs are social. The outputs are proprietary.
You are allowed to query the model. You are not allowed to own the model. A century after enclosure of land, we now witness enclosure of cognition.
The model knows because you wrote, drew, coded, posted and lived. But the model belongs to someone else.
The chip is the factory; the data centre is the jurisdiction
In the industrial age, the factory was a building with smokestacks. Today, the factory is a chip architecture. The oversight of labour has become the oversight of data. Export controls are the new gunboats. The supply chain for high end silicon is now a geopolitical frontier.
A hyperscale campus is not merely a technical asset; it is a political unit. Whoever owns the compute decides which models enter the world and which do not. Whoever secures the power contract shapes the local grid. Whoever writes the API sets the boundaries of permissible thought.
Hegel returns inside Marx’s cage
Narratives about AI still evolve dialectically: exuberance, panic, settlement. Law follows the same pattern: the promise of rights, the fear of risk, the reassurance of governance. But the range of permissible syntheses is policed by ownership. A public compute commons, a democratically governed model stack, or collective rights over training data never make it onto the agenda. The dialectic is performed within a narrow corridor designed to keep capital comfortable.
The quiet contraction of freedom
Hegel’s deepest concern, freedom as self authorship unravels when decisions about citizens are intermediated by opaque systems, when work becomes the correction of machine output, and when language itself is shaped by what the model finds easy to predict.
Formally, the individual is free. Substantively, the individual moves within channels carved by infrastructure they do not control.
The verdict
AI does not vindicate idealism. It vindicates materialism. The system is built from matter, powered by grids, governed by ownership and trained on society’s collective labour. The real contradictions are not in the algorithms. They are in the data centres.
Hegel explains the story we are told. Marx explains the world that builds the story.
The question is not whether AI will change humanity. The question is who gets to point the machine—and who must live beneath it.
This article is analysis and opinion. It describes structural features of today’s AI industry and political economy. References to companies, states or platforms are general and illustrative; they are not allegations of unlawful conduct or wrongdoing by any specific entity.
