The uni-context and the capital cycle

The core job of an investor, above all else, is to keep trying to understand the world better. Annual reports, conversations and data contribute to that process. So can history, fiction or a philosophical idea from another discipline. Reading widely and listening carefully are part of the work - recognising patterns, making connections, joining the dots on a global basis and noticing when our existing frameworks no longer explain the facts.

Most articles we read are chisels in service of this process. They sharpen an existing idea, correct a detail or gradually improve the framework. Occasionally, however, we come across a hammer - an idea powerful enough to break open an existing framework and force us to reconstruct it.

Derek Thompson’s conversation with the philosopher Agnes Callard about the “uni-context” was, for me, one such piece. See link here - A Philosopher’s One-Word Theory to Explain Why the World Feels So Weird

Callard’s starting point is that human beings historically lived within multiple local contexts. A home, a classroom, a place of worship, a market and a bar each had their own norms. How we behaved depended partly on where we were, whom we were with and the role we were performing.

The uni-context emerges when these separate environments are drawn into a single field of observation, judgment and comparison. Everything becomes visible beside everything else. Activities once understood on their own terms must now compete on common measures.

What can be measured becomes easier to compare. What can be compared becomes easier to rank. What can be ranked attracts attention and capital. Once everyone operates within the same evaluative field, behaviour begins to converge.

Technology provides a familiar example. The smartphone became a marker of identity, social position, a device & a escape into this Uni-context world. Social media connected people, but also gave their appetite for attention and status a global stage. The technology created a common context, human nature shaped what people did within it.

When an advantage becomes the game

Markets may be the most powerful uni-context of all. Benchmarks, screens and thematic funds place different companies within common categories. Consultants and risk systems encourage institutions to use the same language. Social media synchronises the narrative.

The evolution of sports offers the clearest parallel. Moneyball began as an analytical advantage. Once every team adopted similar data and optimisation methods, the advantage became the common language of the game. Teams increasingly recruited and played in the same way.

The same paradox operates in investing. Differentiation is valuable, yet the ecosystem continually pushes investors towards conformity. Fund managers know that being conventionally wrong is usually more survivable than being unconventionally wrong.

The uni-context therefore does not eliminate differentiation. It makes genuine differentiation harder - and therefore potentially more valuable.

Where, then, should an investor look for it? In a world increasingly drawn to reducing every problem to a mathematical formula, the contrarian response is to examine what resists that reduction. What this means in practise is that for example while analyzing a stock, pay particular attention to the judgment and integrity of management, the culture of a business and the trust it earns. What is easiest to measure is not necessarily what matters most. (See note What Counts Cannot Be Counted”)

The same forces that make differentiation harder also help explain why capital increasingly travels in crowds. Once an opportunity acquires an accepted label - China, commodities, the internet, clean energy or artificial intelligence—it becomes easier to compare, explain, benchmark, index and fund.

The narrative does not merely describe the flow of capital. It coordinates it.

The uni-context meets the capital cycle

This is where Callard’s idea connects with the capital-cycle framework associated with Marathon Asset Management.

The capital cycle framework in a nut shell - High returns attract capital. Capital creates new capacity. New capacity increases competition. Competition eventually reduces returns. Low returns reverse the process - capital leaves, capacity contracts and the foundations of recovery are created.

My supposition is that the uni-context does not repeal this process. It accelerates it.

A successful theme can attract public equity, venture capital, private equity, corporate capital expenditure, credit, government support and retail flows all at the same time. The snowball gathers size partly because everyone can see it gathering size.

The uni-context explains why capital converges. The capital cycle explains why that convergence may eventually destroy the scarcity which attracted it.

Popularity is therefore not proof that an investment thesis is wrong. It is a reason to investigate how much of the supply response is already being financed.

Being contrarian is not always synonymous with being right. As Newton wrote about shorting manias ‘the market can stay irrational long beyond your ability to stay solvent’. For a fiduciary, it can be career-limiting & also keep an investor out of genuine scarcity profits while demand exceeds supply.

Memory semiconductors provide a useful example. AI infrastructure has created a sharp increase in demand for high-bandwidth memory, while supply is constrained by fabrication capacity and yields. Strong demand meeting inelastic supply can drive prices far beyond normal industry economics, with profits growing much faster than revenues.

The eventual supply response is predictable and a certainty. Its timing is not.

A shortage does not need to last forever to create enormous shareholder value. Two or three years of scarcity can generate enough free cash flow to transform a balance sheet and justify a substantial revaluation. An investor who predicts normalisation correctly but two years too early may be analytically right and economically wrong.

But scarcity alone does not make an investment attractive. Price determines how much scarcity we need—and for how long. A valuation that already allows for normalisation can survive a supply response. One that requires exceptional returns to persist leaves little room for new capacity or adaptation.

Nor is building more capacity the only way scarcity ends. High prices and physical constraints change behaviour - customers redesign products, substitute inputs and learn to use less of what has become scarce. For example, as we speak, AI model developers are changing the recipe too. Memory-efficient architectures and other techniques seek to produce more useful output from less memory and computation. The supply response may arrive through a new fabrication plant or through a model that needs less of what the plant produces.

The difficult question is not whether every snowball eventually stops growing. It is how much hill remains while the snowball is still accelerating.

An invitation to dream

Demand is usually the easiest part of a boom to explain.

In its 2007 investor presentations, the largest Australian mining company made a compelling case for Chinese urbanisation and long-term demand for metals. I recall a slide comparing the density of London’s Underground network with the infrastructure then in place in major Chinese cities. The subtext was an invitation to dream - imagine how much copper and iron ore would be required if every large Chinese city built anything resembling London’s network.

The image was intuitively correct and, in hindsight, almost completely useless for timing an investment. It told investors nothing about when the cities would build, what prices they could afford, how construction would be financed or how much new commodity supply would arrive first.

That is the seduction of demand forecasting. A long-term demand thesis can be right while providing remarkably little help in deciding what to buy, at what price or for how long.

The mistake during the mining boom was not believing in Chinese growth. It was extrapolating temporary scarcity economics too far into the future. High prices financed new mines, railways and ports. By the time the new capacity arrived, demand growth had slowed and some of the industry’s assumptions no longer held.

The beautiful reflexivity is that the crowd’s enthusiasm helped finance the supply that changed the economics.

The questions we can answer

Investors are naturally drawn to the demand question because it offers a grand story. Every company will use AI. Every employee will have an assistant. Every country will want sovereign computing capacity. Falling costs will create applications that do not yet exist.

Much of this may prove correct. But a forecast of eventual consumption does not tell us the price or return on the capacity being built today.

Supply, pricing and finance leave more immediate evidence. Factories are being constructed. Equipment is being ordered. Lead times change. Discounts appear. Contract terms soften. Guarantees multiply.

Demand tells us why the opportunity might become large. Supply, price and finance help tell us where we are in the cycle.

The investment question is whether the imbalance that produced today’s returns is strengthening or weakening.

When belief becomes financing

AI resembles both the mining boom and the telecommunications boom. Like mining, it requires an enormous physical build-out with genuine bottlenecks and long lead times. Like telecommunications, it combines rapid technological change with an ecosystem in which participants increasingly finance one another.

Nor will the cycle turn everywhere at once. Excess capacity could hurt infrastructure owners while lowering costs for cloud platforms and application companies. The end of scarcity in one part of the chain may create opportunity in another. We need to ask where the bottleneck moves and where the returns move with it.

Chip suppliers invest in customers. Customers make advance payments to infrastructure providers. Data-centre projects can be supported by long-dated leases and guarantees.

None of this proves that demand is artificial. But it makes independent demand harder to distinguish from demand created or accelerated by the availability of finance.

Belief is not merely psychological during a boom. It becomes a source of funding. One participant raises money and spends it with another, which records revenue and may finance further activity elsewhere. Each transaction may be real, yet the whole system can become increasingly dependent on the same forecast.

Accounting offers its own clock for the cycle. My simple rule of thumb is that when I spend more time tracing cash flows, testing whether demand is arm’s-length and uncovering contingent obligations - and when the footnotes produce more questions for management than the product does - for me it is a signal that we are well past the early innings.

Early in a cycle, the challenge is to understand the opportunity and the bottleneck. Later, it is to establish who is financing whom, which obligations survive if growth slows and whether separate sources of demand depend on the same pool of capital. Network swaps and the capitalisation of network costs complicated the interpretation of some telecom accounts near the end of that boom. The lesson is to recognise when the analytical questions change.

In my view, the practical funding limit is the free cash flow that the largest strategic participants are willing to commit, plus the debt that bond markets, private credit and infrastructure investors are prepared to absorb. That cash flow also has other claims on it - dividends, buybacks, acquisitions, debt obligations and investment outside AI.

Capital may appear unlimited when belief is strong. But is it ever?

Many booms do not end because the central idea is disproved. They end when the cost of financing the capacity required by the idea rises above the return that capacity can earn.

The first sign of a turn may therefore not be that people stop believing in AI. It may be that belief becomes more expensive to finance. Each marginal dollar may require a wider spread, stronger covenant, longer customer commitment, supplier investment or additional guarantee. The promise can remain intact while the terms on which it is financed begin to change.

Right future, wrong price

Neither the uni-context nor the capital cycle tells us that the crowd must be wrong.

AI may transform the economy. Compute consumption may grow for decades. Today’s infrastructure may be wholly inadequate for tomorrow’s applications.

But technological truth, corporate profitability and investment return are related, not identical.

Being right about the destination does not tell us the cost of getting there, who will earn an adequate return for financing the journey or how much competing capacity will have been built before we arrive.

Reflexive contrarianism is therefore as dangerous as unquestioning optimism. The contrarian may see the eventual supply response and miss years of extraordinary scarcity profits. The optimist may understand the ultimate demand and mistake temporary scarcity returns for permanent economics.

The real task is to judge how long scarcity can persist, how much of that duration the price already assumes and what evidence would change our view. Markets may anticipate the turn before it appears in utilisation or reported profits. Capacity commitments, customer behaviour and financing terms therefore matter—and the less room a valuation leaves for disappointment, the more carefully we should size the position.

The most dangerous moment may not arrive when people stop believing in AI. It may arrive when the promise remains intact but the economics have changed.

The crowd can be right about the future and wrong about the price.