The Thermodynamics of Finance

A financial statement can become larger without a house becoming warmer, a field more productive, or a person more capable. It can also become larger because all three have happened. The number alone cannot tell us which world we are living in. That is where the thermodynamics of finance begins: not by declaring money to be energy, but by asking what relationship the money maintains with the things that require energy to exist.

Finance is an arrangement of claims, obligations, prices, and expectations. The economy those arrangements help direct has another inventory: people, land, materials, machines, energy, and time. A claim on a future product is not the product. Yet the claim can help bring the product into existence by financing work before its result is available. The distance between the promise and the performance is not automatically fraud or failure. It is also where investment happens.

The difficult question is what occupies that distance. Is there a credible process turning resources into something people can use? Or is the expectation of a higher future price doing the work that production has not yet done? Neither a rising valuation nor a skeptical headline answers this. We have to follow the relationship out of the financial statement and into the world.

Energy is conserved. Its capacity to do useful work is not preserved through every transformation: real processes generate entropy, and some energy becomes less available for further work under the surrounding conditions. An operating economy draws on energy and materials and produces outputs, waste, and heat. Those are physical processes, whether the transaction is entered in a paper ledger or an electronic one.

Money does not obey the same conservation law. As the Bank of England explains, a commercial bank making a loan normally creates a matching deposit. That creation is constrained by profitability, regulation, credit demand, monetary conditions, and the borrower's ability to repay; it is not an unlimited privilege. But it is enough to establish the distinction: a new monetary claim can be created without a matching quantity of new energy being created. ¹

Nor is a dollar a fixed quantity of useful energy. Prices depend on institutions, scarcity, bargaining, expectations, and many other conditions. An economic loss is not a measurement of entropy. Calling a market frightened, hot, or frozen does not supply its temperature in the physicist's sense. If we want a literal thermodynamic model, we owe definitions, units, a boundary, and measurements—not merely a vocabulary that sounds like science.

Nicholas Georgescu-Roegen, an economist who brought the entropy law into the study of economic processes, made the material ground difficult to ignore. What this essay takes from that work is the insistence that production cannot be described adequately as money circulating without physical consequences. What it does not take is permission to derive a market crash from the second law. The ground matters; it does not decide every structure built on it. ²

A loan brings purchasing power forward against an expectation of later repayment. That can finance a machine, an education, a home, or an enterprise before the beneficiary has accumulated the full price. The future is being used to organize the present. This is an extraordinary human invention. It also requires an answer to a mundane question: where will the repayment come from?

There is a difference between borrowing to create an income stream and borrowing because an asset is expected to become easier to sell at a higher price. The two can coexist in the same project. A productive investment can be overpriced; a speculative owner can hold an asset that does useful work. The categories are not verdicts about the moral character of everyone involved. They are questions about which process is expected to carry the obligation.

A backlog is not automatically a loan book. An announced contract is not automatically cash collected. Its significance depends on enforceability, cancellation terms, delivery requirements, customer creditworthiness, and payment timing. If a customer's payment depends on another round of fundraising, that dependency belongs in the analysis; it does not by itself establish that the customer will fail. An entire chain can look well funded when each link is counting on a different link's future success.

The physical constraint and the financial constraint meet at delivery. Financing can accelerate construction; it cannot substitute a promise for the transformer, the trained worker, or the electricity that operation requires. The promise has to cross into performance somewhere.

Money can move faster than a factory can be built. A price can change faster than a household can move, a worker can retrain, or a power system can expand. The difference in tempos matters. A system can be solvent in an estimate of long-term value and still unable to meet the payments due now. Liquidity is access to the means of paying when payment is required; solvency concerns whether assets and obligations leave sufficient value. They are related, not identical.

Brunnermeier and Pedersen distinguish market liquidity—the ability to trade without a large price impact—from funding liquidity—the ability to obtain financing. Their model shows how the two can reinforce one another under specified conditions. Falling prices can tighten funding conditions; tighter funding can force selling; forced selling can push prices down again. That is a financial mechanism, not an entropy measurement. It supplies more than the dramatic word 'meltdown' because it names a loop that can be investigated. ³

Leverage makes the question sharper. Borrowing can enlarge what an investment accomplishes and can magnify the effect of a loss on the investor's own capital. Connections can spread risk across participants, but they can also carry distress between them. Diversification is not protection from a shock shared by everything in the portfolio.

A smoother system is therefore not necessarily a more resilient one. But friction is not automatically wise either: restrictions can prevent useful investment as readily as reckless expansion. The question is which interruption keeps a particular system capable of continuing its work—and which merely protects an established participant from challenge.

Cui bono: who benefits? The companion question is who bears the harm. An investment can produce useful services, wages, and returns while also imposing costs on people who did not choose it. It can fail its owners while leaving something valuable behind. We cannot infer the complete public result from either the investor's profit or the critic's inventory of damage.

The boundary of the accounting is doing part of the thinking. A cost outside a company's accounts has not necessarily disappeared; it may have become another person's expense, a public obligation, or an environmental consequence. Whether that happens in a particular case requires evidence about prices, contracts, laws, and physical effects. 'Externalized cost' is a question to document, not a phrase that makes every project guilty.

Likewise, a financial transfer is not automatically wasted energy. Interest, insurance, and fees can pay for real services or allocate risk; they can also become expensive arrangements that produce little corresponding benefit. Physics cannot settle that distinction for us. We have to ask what the arrangement contributes, what alternatives exist, and whether the people carrying the consequences had any meaningful say.

Waste heat has a physical definition. An unfair bill requires an ethical and institutional argument. Keeping those meanings distinct does not weaken the inquiry. It prevents a scientific word from making a political decision without admitting that a decision was made.

Engineered intelligence arrives as something that can appear almost weightless: an answer, a design, a piece of code. Its production is not weightless. Training and use depend on computing equipment, electricity, buildings, networks, and the physical work that provides them. The IEA's Energy and AI report examines that electricity demand and its possible development. Its projections are scenarios about energy systems, not a verdict on the financial soundness of any company. ⁴

The useful question is not simply how much energy a model consumes. It is what a particular use accomplishes, against what alternative, at what total cost. Finding a better material, reducing a manufacturing error, or helping a person solve a previously inaccessible problem can be worth doing. Generating another unwanted promotion may not be. Counting computations cannot supply the judgment of value.

Greater efficiency can reduce resources required per task. If lower costs encourage enough additional use, total consumption can nevertheless rise. Whether that happens, and by how much, must be measured in the particular activity. A more efficient instrument and a larger total demand can coexist without either observation being false.

We can hold the excitement without surrendering the question. New materials, better buildings, and more capable tools are genuine possibilities worth pursuing. Their financing should be examined as financing; their usefulness as usefulness; their resource demands as resource demands. One successful result does not certify the entire investment cycle. One speculative financing arrangement does not invalidate every result produced by the instrument.

A financial system can contain intelligent analysts and still make a decision no individual analyst would defend alone. The relevant question is not whether someone, somewhere, noticed the weakness. It is whether the weakness could alter the commitment. An unanswered objection filed neatly in the record has not necessarily governed anything.

This is where the thermodynamic language stops and the human question becomes unavoidable. Heat does not choose the boundary through which it flows. People choose contracts, priorities, protections, and who is permitted to refuse. What looks like an unavoidable cost may partly be an avoidable arrangement about where the cost lands.

Socratic questioning is a possible improvement, not a guarantee against financial failure. What is the objective? Why that objective? Which assumption carries the repayment? Who benefits if the plan works, and who pays if it does not? What evidence would change the decision? The same questions must reach the questioner. Otherwise an examination of authority becomes another authority with a preferred answer.

No question can make uncertainty disappear. It can make an assumption visible before the obligation becomes difficult to reverse. But visibility is insufficient if nobody has the standing, time, or decision right to act on what has become visible.

The thermodynamics of finance is useful only if it returns us to something the financial description left out. In a literal inquiry, that means measured energy and material flows, useful outputs, waste, and a specified boundary. In a financial inquiry, it means obligations, maturities, cash flows, losses, and the people legally responsible for them. In a public inquiry, it means the distribution of benefits and burdens. The accounts can be compared. They cannot simply be collapsed into one number.

This essay's proposed lens would lose its value if ordinary cash-flow, resource, and distributional analysis supplied every distinction with nothing gained from the thermodynamic framing. A stronger literal claim would need to connect a defined physical quantity to a defined financial result and survive comparison with explanations that do not use it. Beautiful language is not an exemption from that work.

There is no crash prediction hidden here. An economy can waste resources without an immediate financial collapse. Financial arrangements can fail while valuable equipment and knowledge survive them. A company can be profitable while a wider community carries costs, and a public investment can be worthwhile without generating a private profit. The different results require different accounts.

The title brings us back to the relationship. Finance can organize the work by which a possible future becomes an actual one. It can also organize claims on a future that nobody has established can carry them. The task is not to fear every promise or believe every promise. It is to notice where the promise must meet the world—and keep that meeting open to examination.