Horses helped build entire economies, yet they never understood money. They worked for oats. For centuries, this simple fact went unnoticed, simply because it didn’t matter. People managed the money, horses provided the power. The system worked because there was a bridge between the two worlds: money could buy oats, and oats provided the energy for work.

Today, a new kind of “workhorse” is emerging: artificial intelligence (AI). And like the horse, it doesn’t care about money. This observation may sound trivial, but it isn’t. It points to a profound shift that could change our thinking about money, value, and economic power.

The horse’s forgotten lesson

Before steam engines and electricity, horses were not just helpful, they were essential. In pre-industrial economic systems, a significant portion of value creation depended directly on animal power. Fields were plowed by horses, goods were transported by horse-drawn carts, and entire logistics systems relied on them. In terms of energy consumption, horses were among the most important sources of power.

In other words, a significant portion of economic output was literally generated by living beings who knew neither prices, wages, nor wealth. And yet, despite this enormous contribution, horses were not economic actors in the human sense. They had no understanding of money, no awareness of exchange, and no ability to store value. Their world was biological: food, rest, security.

The interface between these two worlds, the human monetary system and the horse’s biological system, was simple: humans used money to buy oats, hay, and water. These resources, in turn, enabled the horse to work. The horse never came into direct contact with money. It only reacted to what money could affect in the physical world. Money, in this context, was not a universal language. It was a layer of translation.

From oats to electricity

AI functions remarkably similarly. AI systems aren’t interested in salaries or savings. They don’t accumulate wealth or strive for maximum monetary return. Instead, they work with other input factors: electricity, computing power, data, and access.

While a horse converts oats into movement, AI converts electricity into decisions. Behind every AI model lies a chain of physical and technical processes. Electricity powers data centers. Data centers operate hardware, GPUs, and other specialized chips. This hardware executes algorithms. The result is output: text, predictions, code, and recommendations. This output creates economic value.

At no point in this chain does money enter the AI system itself. People use money to pay for electricity, infrastructure, and development. But the AI, just like the horse, never “sees” it. It only reacts to the resources that can be mobilized through money. In this sense, electricity is the grain of the digital age.

A subtle but profound change

For much of economic history, the existence of non-monetary producers did not fundamentally challenge the role of money. Animals and machines were important, but human labor and decision-making remained central to value creation. AI is changing this balance: For the first time, we are witnessing the emergence of a production factor that is not only non-monetary but also highly scalable, adaptable, and increasingly central to the economy. AI writes code, analyzes financial markets, optimizes logistics, generates content, and supports cross-industry decision-making processes. In doing so, it is taking over tasks that were previously the domain of human cognition. And yet, despite this growing role, AI remains fundamentally indifferent to money.

Here, the historical comparison becomes particularly clear: In pre-industrial economies, horses made a significant contribution to value creation. However, their impact was ultimately limited by biological, spatial, and geographical constraints. AI, on the other hand, operates without many of these limitations. It can be replicated, distributed, and scaled across sectors almost instantaneously. If horses augmented human muscle power, AI augments human cognition to an extent that will dwarf anything seen before. For the first time, we are not merely supplementing human labor at the periphery, but replicating and substituting core elements of human decision-making, analysis, and coordination in entire economies.

The decoupling of money and production

This creates a tension at the heart of the modern economy. Money has traditionally fulfilled three main functions: it served as a medium of exchange, a unit of account, and a store of value. But all three functions presuppose actors for whom money is important, who use it for trading, measuring, and saving. What happens when the primary producers of value do not participate in this system? We may be witnessing the beginnings of a decoupling of money and production.

In a traditional economy, these two were closely intertwined. Workers earned wages, businesses generated revenue, and prices coordinated supply and demand. Money was both the measure and the mechanism of economic activity. In an AI-driven economy, this connection becomes less direct.

AI systems don’t earn wages. They don’t manage bank accounts. They don’t set prices or negotiate contracts. Instead, they consume resources and produce goods. The coordination of these resources still occurs through monetary systems: companies pay for computing power, investors allocate capital, and markets determine prices. But the production process itself increasingly takes place outside the monetary sphere. The disruption caused by AI doesn’t end with the labor market, however: it extends to savings, because when money loses its central role, the values stored within it also lose reliability.

The economy can be divided into two levels. The first is the monetary level, where people operate. Here we find prices, wages, contracts, and financial markets. This is the familiar world of money. The second is the resource level, where production takes place. Here, the relevant variables are electricity, computing power, data availability, and infrastructure. This is the world in which AI operates.

Money connects these two levels, but is no longer their substance. It becomes an interface, a means of translating human intentions into resource allocation.

When resources behave like money

As part of this transformation, certain resources are taking on characteristics traditionally associated with money. Computing time, for example, is increasingly being traded, sold, and allocated with precision. Access to high-performance chips can determine which companies lead the way in AI development and which lag behind. In some contexts, computing power is priced, rationed, and prioritized in ways similar to financial markets .

Similarly, access to data and infrastructure is becoming a crucial factor for economic performance. These are not currencies in the traditional sense, but rather important factors of production that can be exchanged, accumulated, and utilized.

In these areas, the relevant question is not, “How much money do you have?” but rather, “What computing power can you access?” This marks a subtle but important shift. Money remains necessary, but it is no longer sufficient. Economic power increasingly depends on control over physical and technological resources.

Rethinking the nature of value

The rise of AI also forces us to rethink our understanding of value. In human economics, value is expressed symbolically: prices, wages, and financial indicators form a common language for comparing goods and services. Money allows us to translate the physical world into numbers.

This perspective is profoundly human. For a long time, this was sufficient because the majority of value creation originated from humans. But with the rise of AI, this focus is shifting. If, in the future, an increasing share of value creation is generated by systems that lack an understanding of money, a new question arises: What does money look like from the perspective of these systems? The answer is sobering: From AI’s point of view, money is neither a goal nor a measure, but at best a detour. Or, to put it more bluntly: For the systems that will generate the majority of value creation in the future, fiat money is simply irrelevant.

AI operates within its own functional system. Its results are measured by their quality: accuracy, speed, efficiency, and relevance. These are not monetary values, but rather key performance indicators.

When AI systems create value, it does so indirectly. A model that optimizes logistics, for example, generates economic benefit – not by making money, but by reducing costs, improving processes, or increasing performance. This creates a growing gap between the symbolic representation of value (money) and the actual value creation (AI-driven processes).

The new question of power

When money becomes more of an interface than the core of production, the question of economic power shifts accordingly. In a monetary system, power is often associated with capital, with the ability to invest, grant loans, and distribute financial resources. Institutions such as banks and financial markets play a central role in this.

In a resource-driven system, power is more closely tied to infrastructure. Whoever controls the energy supply, chip manufacturing, data centers, and network access gains a strategic advantage.

However, this does not mean that money becomes irrelevant. On the contrary, it remains an instrument for acquiring and coordinating resources. But it is no longer the sole or decisive factor for economic performance.

To return to the previous analogy: having money is not the same as having oats. And in a world where work depends on oats – or electricity – the latter may be more important.

A familiar pattern on a new scale

In some respects, this development is not entirely new. The distinction between monetary and non-monetary systems has always existed. Humans have always relied on animals, machines, and natural resources to create value.

What is new, however, is the scale and scope of this change: Horses were strong, but limited. They required care, had limited performance, and functioned in relatively simple systems. While their role was important, it did not fundamentally alter the economic structure. AI, on the other hand, is universally applicable, scalable, and deeply integrated into modern infrastructure. It can operate across domains, learn from data, and continuously improve. Its potential impact is far greater.

Horses once contributed significantly to economic output without ever being used for monetary purposes. AI will achieve something similar, but on a scale that dwarfs anything seen before. Consequently, the decoupling of money and production is no longer a fringe phenomenon; it is becoming a central feature of the economic landscape.

If the machines that create most of the world’s value no longer use money, the true currency of power is no longer money, but control over the systems that power them: electricity, computing power, data, and the infrastructure that connects them. The widespread introduction of autonomous, self-optimizing AIs will ultimately end the monetary system as we know it – because AIs do not care about fiat money.

Featured image by Patrick Schüffe with  AI assistance (ChatGPT).