From Finance to AI and Back
There is a moment when an instrumental relation becomes structural. For decades finance used technology as a tool: first computers to automate calculations, then the internet to distribute markets, and later algorithms to optimise execution. In each phase technology operated inside finance, and finance remained the system that contained and directed it.
With artificial intelligence this relation has inverted. Finance no longer uses technology as a tool. Technology, in the form of AI, now reshapes the categories through which finance understands itself, while financial capital fuels the development of AI. Financial capital and cognitive infrastructure are co-producing one another in a circular process in which neither contains the other.
The mechanism is visible in the numbers. Sovereign wealth funds in the Gulf, major American asset managers, and Asian technology investment vehicles are acquiring significant positions in foundational models, data centres, and the energy infrastructure that powers computation. These are investments in infrastructure that produces the models used to manage portfolios, assess risk, and identify investment opportunities. The loop is closed.
Soros (1987) described reflexivity as the capacity of financial markets to be transformed by the expectations they generate. Prices do not reflect reality; they construct it. Participants form expectations about future prices, expectations influence behaviour, behaviour alters prices, and new prices reshape expectations. The circle reinforces itself until it breaks.
The finance-AI loop exhibits the same reflexive structure at a deeper level. Expectations about foundational models attract capital. Capital accelerates model development. More powerful models become more useful for capital allocation and risk management. Their usefulness attracts further capital. The circle no longer closes on the price of a single asset but on the cognitive infrastructure of the economic system.
This co-production has a consequence that is not yet fully understood. When finance and AI co-produce one another, the categories used to analyse one become insufficient to understand the other. A financial analyst who does not understand foundational models cannot evaluate cognitive systemic risk. A technical specialist who does not understand allocative power cannot see where their models are deployed or what effects they produce. The competence required to read this transition is transversal.
The transversal technician is the structural response to a transition that has created a domain in which no existing discipline is sufficient on its own.