Advanced artificial intelligence does not intervene in capitalism merely as a technology of automation, but as an infrastructure of coordination. Its significance is not measured solely by the replacement of tasks, the productivity of individual processes, or the compression of operating costs. The most relevant point concerns the transformation of the way in which the economy observes itself, anticipates its own imbalances, and organizes distributed decisions across increasingly interdependent productive, financial, and institutional chains. The market, historically founded on synthetic signals such as prices, scarcity, demand, and supply, is now accompanied by a cognitive apparatus capable of processing correlations, forecasts, and scenarios before traditional signals become fully visible.
This transformation modifies the very nature of economic decision-making. The classical industrial enterprise reacted to the market: it detected variations in demand, corrected production, and adjusted prices, inventories, investments, and employment. The enterprise augmented by artificial intelligence instead tends to operate within a predictive temporality. It does not merely wait for consolidated data, but works on weak signals, probabilities, behavioral patterns, interactions among supply chains, demand volatility, logistical risks, and financial sensitivities. The decision no longer arises after the market, but within an anticipated simulation of it.
The OECD has described artificial intelligence as a possible general purpose technology, capable of affecting productivity, distribution, and growth, while subsequent analyses have estimated its macroeconomic effects on G7 economies over a ten-year horizon. The International Monetary Fund has linked generative AI to a structural transformation of labor markets, income distribution, and returns on capital, with greater exposure for advanced economies and cognitive occupations. This is therefore not a sectoral technology, but a transversal economic function, destined to affect the formation of value and its institutional distribution.
The theoretical issue remains that of dispersed knowledge. Friedrich Hayek showed that the fundamental economic problem does not consist in concentrating all information in a single center, but in coordinating fragmented knowledge among individuals, firms, and territories. Price, from this perspective, was a form of compressed communication: a minimal signal capable of orienting millions of decisions without the need for central command. Artificial intelligence does not eliminate this insight, but moves it onto a denser plane. Alongside the price signal, there now stand models capable of estimating when a resource will be needed, where it will be lacking, with what probability it will be demanded, and what effects it will produce on margins, procurement, reputation, credit, and systemic risk.
The novelty, therefore, does not consist in a return to centralized planning, a category inadequate for interpreting computational capitalism. What emerges is a hybrid coordination, in which the market, digital platforms, proprietary algorithms, financial systems, logistical chains, and predictive models contribute jointly to the formation of allocative decisions. The factory, the warehouse, the commercial network, treasury, and pricing no longer operate as separate functions, but as interconnected modules of the same informational architecture. Value depends not only on productive capacity, but on the capacity to read, anticipate, and synchronize flows.
Within this framework, computational power assumes an economic and institutional nature. It is not merely technical power, but the capacity to make certain relations visible before other actors do, transforming anticipation into competitive advantage. The Bank for International Settlements has situated the evolution of artificial intelligence in the financial sector within the broader history of information processing, emphasizing how advances in information-processing capabilities are redefining intermediaries, markets, and decision-making infrastructures. Competition no longer concerns only capital, labor, and access to markets, but also data quality, model speed, algorithmic governance, and control over computational infrastructures.
The economic focus therefore becomes unavoidable. AI does not merely produce efficiency, but redistributes decision-making power. Those who possess more sophisticated models can anticipate demand, modulate prices, optimize inventories, select customers, assess risks, and react to shocks more rapidly. The Federal Reserve has observed that the adoption of AI in the United States economy tends to show a significant relationship with firm size, while studies on algorithmic pricing detect a greater propensity among large firms to adopt AI pricing systems, with implications for growth, profitability, mark-ups, and sensitivity to monetary shocks.
The risk is not only monopolistic in the traditional sense. It is cognitive. When the capacity to interpret the market becomes concentrated in a few organizations endowed with data, infrastructures, and computational capital, competition may formally remain open while the effective capacity to orient oneself in the market becomes asymmetric. Opacity does not necessarily arise from concealment, but from complexity. A dynamic price, an automated credit decision, a demand forecast, or an algorithmic selection of investments may appear as technical outcomes, but they incorporate objectives, priorities, constraints, and visions of risk.
For this reason, the governance of artificial intelligence cannot be reduced to regulatory compliance or cybersecurity. It concerns the definition of the objectives embedded in systems: efficiency, margin, resilience, employment continuity, sustainability, market power, waste reduction, financial stability. Every allocation model contains an implicit institutional grammar. It establishes what must be optimized, what may be sacrificed, and what remains outside the perimeter of measurement. Technical neutrality, at this stage, is a weak form of conceptual irresponsibility.
Labor is involved in the same process. AI does not simply replace executive activities, but reorganizes the relationship between human judgment, procedure, and automated decision-making. The most exposed tasks are not only repetitive ones, but those in which knowledge, language, classification, and prediction constitute an essential part of value. The consequence is not a linear trajectory of labor elimination, but a deeper segmentation between augmented labor, monitored labor, devalued labor, and labor made invisible by the digital infrastructure that coordinates it.
The institutional question emerges precisely here. If the market becomes more capable of thinking, or at least of calculating, society must become more capable of governing the conditions of that calculation. It is not enough for firms to adopt more powerful AI. It is necessary to understand what new dependencies are produced among capital, data, energy infrastructures, proprietary models, financial systems, and public apparatuses. Money, credit, industrial policy, and competition regulation will progressively be called upon to confront markets in which the speed of reaction is not uniformly distributed.
Algorithmic capitalism will not abolish scarcity, conflict, error, and uncertainty. It will make them faster, more measurable, and at times less perceptible. The synchronization of processes does not coincide with economic harmony, just as prediction does not coincide with responsibility. Artificial intelligence can strengthen the market’s capacity to process information, but it leaves open the question most relevant to firms, institutions, and democratic systems: establishing which ends should orient an economic rationality that possesses increasingly powerful instruments for anticipating the future, without thereby automatically possessing a criterion for judging it.
Global AI Observatory