There is something interesting about the global economy these days. It has spent quite some time absorbing one shock after another, while many of the assumptions that once looked reasonably stable have continued to shift.

The IMF expects global growth of 3.1% this year, alongside headline inflation of 4.7%. Those figures matter, but they only capture part of what has been happening underneath. Markets and institutions have been adjusting continuously to an environment that looks increasingly different from the one they entered only a few years ago.

We naturally focus first on what moves. Prices change, volatility rises, a constraint appears or expectations shift. Those effects are visible and usually easier to measure.

What follows is harder to see. The initial change starts reshaping the choices around it, and those choices can take the system somewhere different. Some adjustments fade as conditions normalise. Others leave behind something that was not there before.

That is where the story becomes more interesting.

Let's start with geopolitics

The first effects can appear in commodities, trade routes, currencies or market volatility, but financial markets soon have to interpret what they mean for inflation, growth and monetary policy. A disruption that begins outside financial markets can therefore change the assumptions on which assets are being valued.

Recent BIS research on emerging-market sovereign risk offers one measure of this process. Fredy Gamboa and José Vicente Romero find that higher geopolitical risk raises both sovereign CDS and bond spreads, with anticipated threats producing larger and more persistent effects than realised events in their sample. Markets can therefore react to the possibility of disruption before the disruption itself arrives.

The consequences are rarely distributed evenly. Assets that previously traded on broadly similar assumptions can begin to diverge as markets reassess where disruption is most likely to matter. What begins as a geopolitical event can therefore create greater dispersion across markets rather than simply pushing everything in the same direction.

That changes portfolio construction as well. Investors may reconsider how much risk they want to carry, where they want to carry it and how much protection or liquidity they need around those positions. The question becomes less about whether markets are simply “risk-on” or “risk-off” and more about which risks investors still want to own, and on what terms.

The wider market environment is important too. When relatively low-risk assets offer very little return, investors have a stronger incentive to look elsewhere. When that changes, the hurdle for taking additional risk rises across asset classes. The same geopolitical uncertainty can therefore have a different market effect depending on the opportunity set investors are starting from.

Over time, geopolitical risk can move from being treated mainly as an exceptional scenario to becoming part of the assumptions behind everyday investment decisions. It can influence how markets judge resilience and vulnerability even when no immediate disruption is taking place.

By the time another disruption arrives, markets may therefore already be viewing the world through a different set of assumptions.

The next shock lands on a market that has already moved.

Then we can add AI to the mix

And we realise that not all structural change begins with risk becoming more expensive. Sometimes it begins because an important capability becomes dramatically cheaper.

Artificial intelligence is usually framed as a productivity story. Tasks can be completed faster, software produced more efficiently and research or analysis carried out at lower cost. If the same output requires fewer inputs, productivity improves.

But lower cost does not necessarily mean the same amount of activity will simply be produced more efficiently. William Stanley Jevons observed a version of this more than 160 years ago. Greater steam-engine efficiency reduced the coal required for a given amount of work, but also made steam power economical across more uses. He called it a “confusion of ideas” to assume that more economical use must necessarily reduce overall consumption.

AI is clearly not coal, and the analogy should not be treated as a law. The useful question is what becomes possible once the cost of a capability falls far enough. If sophisticated analysis becomes dramatically cheaper, we may apply it to decisions that previously did not justify the time or expense. The same could happen with software that was once too specialised or short-lived to make economic sense.

That possibility is already producing economic consequences before the eventual productivity gains are clear. The BIS describes the AI boom as a large and increasingly debt-financed investment surge that is supporting trade and equity markets while producing significant wealth and terms-of-trade effects across countries.

Expectations about future capability are also creating real investment in semiconductors, data centres, networks and power infrastructure. BIS research expects debt and private credit to play a larger role as investment needs move beyond what operating cash flows alone can comfortably support.

What began as a technology story has therefore become a physical investment story. If AI produces very large productivity gains, the capacity created today may support forms of activity that barely exist yet. If expectations prove too optimistic, some investments will disappoint, but the system will not simply rewind. Some of the infrastructure and capabilities created along the way may continue to matter beyond the investment cycle that produced them.

Technology cycles do not wait for certainty. Capital moves while the answer is still being discovered.

Should we also zoom out to 2008?

We cannot yet know how much of the current AI cycle will endure, but recent history gives us a clearer example of what can happen when adaptation has enough time to become part of the system itself.

The Global Financial Crisis of 2007–09 exposed fundamental weaknesses in bank capital, liquidity and risk management. The response changed banking materially. Capital requirements increased, liquidity standards were introduced and supervision became more demanding. Banks became more resilient to many of the vulnerabilities revealed during the crisis.

The financial system that developed afterwards, however, was not simply the old one with safer banks. Non-bank financial intermediation grew substantially in importance. By the end of 2024, non-bank financial institutions accounted for 51% of total financial assets across the jurisdictions covered by the Financial Stability Board’s monitoring exercise. There is no single reason for that growth. Institutional savings, deeper capital markets, technology, asset valuations and demand for alternative financing all played a role.

What the post-crisis period makes easier to see is how activity responds when the economics of one route change. A 2026 IMF study finds that when macroprudential policy tightens credit through bank subsidiaries, banking groups with affiliated non-bank financial institutions shift some lending towards those entities. On average, the increase in non-bank lending offsets more than half of the contraction in bank lending caused by the tightening.

The activity does not necessarily disappear because one channel becomes more constrained. Some of it moves.

If that happens repeatedly and at sufficient scale, markets start developing around it. Investors build expertise, dedicated pools of capital grow and borrowers become familiar with different sources of financing. Relationships that once looked peripheral become established enough to support further activity.

Over time, adaptation becomes architecture.

That does not automatically make the system stronger. As activity moves, risk can move with it, and new connections can create vulnerabilities of their own. Future shocks are then transmitted through a financial system with different channels, relationships and concentrations of risk.

The post-crisis experience gives us something the earlier examples cannot yet: enough distance to see what remained. Prices can normalise and attention move elsewhere, while some of the capabilities, markets and relationships that developed during the adjustment remain.

So what comes next?

The immediate effect of economic change can only tell us so much.

What matters over time is whether the response creates something durable enough to influence what happens afterwards. Many adjustments are temporary: repricing can reverse, investment cycles turn and policies change. Others alter the choices available afterwards.

That can mean a risk that is priced differently, a capability that becomes economically viable, or a financing route that develops enough depth to become part of the system.

The progression from change to adaptation is useful for thinking about this, even if reality never follows such a neat sequence. Markets can react in minutes, institutions may take years to adjust and structural changes often become obvious only in hindsight. By then, the event that set them in motion can feel surprisingly remote.

The global economy in 2026 sits somewhere in the middle of several such adjustments. We can see much of what moved first, but not yet which responses will endure.

That is ultimately what makes knock-on effects worth watching: not simply what changes first, but what that change leaves behind.