What if Capital is not merely a system that humans operate, but a historical agential system that operates through us? This research cycle follows Capital from intelligence, through labour, into embodiment. It begins with an analogy between Capital and artificial intelligence, asks what gives this system its life, and ends in the machinery of the data centre: a body with many possible bodies.
We treat Capital as a historical agential system rather than simply the sum of individual economic actors. Because it processes quantities on a scale beyond any individual and can be understood as an ongoing process of quantitative optimization, Capital begins to resemble artificial intelligence. AI systems, in turn, can be understood as agents of Capital: they ingest quantified value and produce more quantified value.
A large, spreading, and complex ‘beast’
ALIFE 2024 / MIT PRESS
Capital as Artificial Intelligence
TALK TO CAPITAL
interview / subject: Capital
Labour is approached as an expression of “timenergy”: time and energy committed to the Capital system. Unlike Capital, Labour cannot be accumulated. An agent-based model shows learning agents gravitating toward capital-intensive production, revealing an asymmetry between finite labour and accumulable Capital.
Capital is artificially alive, animated by the living Labour it consumes.
ALIFE 2025 / MIT PRESS
Does Capital Dream of Artificial Labour?
WATCH THE VIDEO
IMAGE / RAW VALUE
If AI appears disembodied, where is its body? We argue that the Data Centre is, in many cases, the body of AI—and simultaneously part of the labouring body of Capital. Server racks, cooling, energy and communication systems can be read through an organismic lens. Yet computation is substrate-agnostic: one function can have many bodies.
One intelligence. Many bodies. No desire of its own.
ALIFE 2026 / MIT PRESS
The Many-Body Problem of the Data Centre
ENTER COMPUTER
COMPUTATION / VALUE
SOURCE: INTERVIEW.TXT / Q → [C] / SELECTIVE PROTOCOL