The Confidence Recession•Chapter 4 of 8

The Capital That Still Has to Earn

Chapter 4 explains why AI utility, provider profitability and equity returns can diverge sharply — and how that divergence can…

July 26, 2026 6 min read Research Track
The Capital That Still Has to Earn

The Capital That Still Has to Earn

Updated: 26 July 2026

So far, we have described how uncertainty around straits, energy, Europe and export chains can seep into the real economy. Chapter 4 turns the perspective around: what happens on the capital side when the real world grows more cautious while financial markets, pension systems, lenders and shareholders still expect rising returns?

The short answer is this: a gap opens between the future that was promised and the future that can actually be sustained. That gap is the financial core of a confidence recession.

Why Capital Cannot Simply Wait

Capital is not neutral. It comes attached to claims: shareholders want gains and dividends, creditors want interest and repayment, banks want stable collateral, states want tax revenue, pension funds need ongoing returns and venture investors want the next valuation round. Those claims keep running even when real demand slows.

In an expanding economy, such claims can often be serviced at the same time. But when consumption, investment and world trade weaken, a distributional struggle begins over a smaller income base. That is when it becomes clear that the financial system may have promised more than the real economy can carry under new conditions.

The problem becomes sharper because financial markets trade the future. Valuations rest on expectations about future cash flows, margins, market share and productivity. When that future becomes more uncertain, doubts do not affect only individual firms. The price of time itself changes.

The financial system can discount future earnings, but it cannot permanently hallucinate them into existence.

From Valuation Story to Financing Squeeze

When the world consumes less, expected revenues and margins come under pressure. That does not have to end immediately in recession. It can begin as a valuation problem:

weaker sales outlook
→ lower earnings expectations
→ falling valuations
→ more expensive equity capital
→ tighter lending
→ less investment
→ weaker growth.

A falling share price is therefore not just a market event. It can hit the real economy when companies postpone issuance, cancel acquisitions, cut research, close sites or leave positions unfilled.

Business models whose current valuation depends heavily on distant earnings promises are especially vulnerable. In good times, that looks like a turbocharger. In a confidence recession, it becomes a lever in the other direction.

AI Utility Is Not the Same as Provider Profitability

This distinction matters especially in AI. The fact that artificial intelligence can raise productivity, change search and work processes or create new applications does not automatically mean that all providers will earn above-normal profits. Still less does it mean that current stock prices are already justified.

Between AI utility, provider profitability and equity returns stand several hurdles:

  • Investment: data centres, chips, power grids, cooling and software absorb enormous sums.
  • Price competition: if many providers offer similar models and services, margins fall.
  • User monetization: not every productive application creates immediate willingness to pay.
  • Substitution: AI can cannibalize existing revenue rather than merely add new revenue.
  • Regulation: privacy, liability, copyright, competition and data-localization rules can raise costs.
  • Social backlash: layoffs, automation pressure and dependency concerns can trigger boycotts, regulation or sovereignty moves.

An AI wave can therefore be economically useful and financially disappointing at the same time. That is not a theoretical nuance but a classic feature of technological transitions: the benefits often spread more widely than the profits remain concentrated.

The AI Investment-Payback Shock

One especially sensitive risk is the AI investment-payback shock. This means the possibility that very large investments in data centres, chips, networks and power infrastructure rise faster than the returns that can actually be monetized.

The transmission chain can look like this:

AI euphoria
→ rising investment by hyperscalers and suppliers
→ high valuations across the infrastructure chain
→ doubts about monetization or demand
→ reassessment of earnings expectations
→ pressure on capex plans, suppliers, credit channels and labour markets.

This shock does not require the technology to fail outright. Even a small gap between expected and realized return can be enough if very large sums, high multiples and tight capital-market interconnections were built up beforehand.

The more individual stocks, indices, ETFs, options markets and credit structures depend on a few winning narratives, the greater the spillover risk becomes. In this context, spillover means that a correction in one area spills into others — for example from semiconductors to utilities, from data centres to construction firms or from technology stocks into consumption and credit markets.

If Capital Has Flowed Back into the United States

In recent years, return expectations, dollar strength, technological leadership and the depth of US capital markets have attracted large sums. Part of that capital seeks safety; another part seeks above-average growth. Both motives can change, however, if expected returns fall or if political and fiscal risks rise.

What happens then to the capital that flowed into the United States? It does not simply disappear, but it can change form:

  • from growth stocks to more defensive equities,
  • from risky bonds to short-duration government paper,
  • from illiquid to more liquid holdings,
  • from foreign exposures back into home markets,
  • from market assets into cash, gold or selected commodities.

Part of the money therefore remains within the system, but shifts from a mode of “return chasing” to one of “damage control”. That very shift can raise financing costs even when a large nominal capital stock still exists.

Labour Markets, Automation and Demand

Under confidence stress, automation becomes ambiguous. Firms seek efficiency gains, lower wage costs and better scalability. But the macro question is critical: who buys the additional goods and services if part of the workforce is displaced, unsettled or weakened on the wage side?

The issue is not that every AI application destroys jobs. The issue is that the distribution of gains and losses can diverge across time and across social groups. Firms save labour costs today, while new jobs, higher wages or broader prosperity may appear only later or unevenly.

That can create a new kind of demand problem:

more automation
→ less job security or a lower labour share in parts of the market
→ weaker consumption propensity
→ weaker revenues outside a few winning sectors
→ fresh doubts about capital returns.

This is also where a possible political backlash may emerge: “Don’t buy from firms that replace people with AI.” Whether that becomes a broad movement is still open. But as an early warning signal for confidence and demand risk, it deserves attention.

The Short Answer to the Core Question

So what happens when the world consumes less while global capital needs more return than ever? Then not every claim can be satisfied at the same time. Part of the future gets repriced. Some valuations fall, some investments are postponed, credit becomes more expensive, employment planning grows more cautious and political distributional conflicts intensify.

Not every correction immediately becomes a world recession. But the larger the gap between financial claims and real earning power becomes, the more likely a phase of repricing, deleveraging, weaker demand and political reaction becomes.

Sources and Method

This chapter links financial logic, AI investment dynamics and macroeconomic transmission. The featured image is an illustration. Core source families:

Gridizer Research is an analytical publication. It does not constitute financial, legal or political advice.