As investors increasingly frame AI as a binary trade between winners and losers, the real opportunity may lie in identifying the companies being mispriced by simplistic market narratives.
The market’s response to artificial intelligence has been anything but linear.
Twelve months ago, investor concern centred on whether AI enthusiasm had run ahead of fundamentals.
Today, that debate has flipped. The dominant question is no longer whether AI is overhyped, but whether it will prove so disruptive that large parts of the economy struggle to keep up.
That shift in narrative has been reflected clearly in market behaviour. In early 2026, sectors tied to physical production — energy, materials and industrials — have materially outperformed, while software and other capital-light industries have lagged.
The market, in effect, has begun to price AI as a binary outcome: a force that creates clear winners and losers, with limited middle ground.
For long-term investors, that framing is too simplistic.
AI is unlikely to reshape the economy in a uniform or predictable way. Its impact will be uneven, non-linear and highly company specific.
In that sense, AI is less a single investment theme and more a dispersion engine — one that is already creating meaningful gaps between perception and reality.
The re-rating of the “physical economy”
One of the clearest expressions of the current narrative is the re-rating of companies perceived to be relatively insulated from AI disruption.
Industries tied to real assets and physical production — sometimes described as “AI immune” — have benefited from a sharp shift in investor preference. The logic is straightforward: AI cannot manufacture jet engines, lay copper wiring or replicate the in-person experience of many services businesses.
At the same time, a range of cyclical tailwinds have reinforced the move. Industrial companies are emerging from a prolonged period of cost discipline, defence spending is rising globally, and demand across areas such as aerospace and infrastructure has strengthened.
Taken together, this has supported a rotation into more capital-intensive sectors.
But while the underlying drivers are real, the market’s interpretation may be drifting toward overconfidence. The idea of “AI immunity” risks becoming too absolute. Few businesses are entirely untouched by technological change, particularly one with the breadth of AI. Even sectors built on physical production are likely to experience shifts in cost structures, supply chains and competitive dynamics over time.
In that context, the more relevant question is not whether a business is immune, but how its economics evolve as AI adoption broadens.
When disruption is over-discounted
At the other end of the spectrum, the market has taken a far more pessimistic view of companies perceived to be directly exposed to AI disruption.
Software, consulting and information-based businesses have come under pressure amid concerns that generative AI will commoditise large parts of their value proposition. The emergence of increasingly capable AI tools has reinforced the perception that many existing business models face structural decline.
There will almost certainly be cases where that proves correct.
However, the breadth of the sell-off suggests the market may be extrapolating too aggressively. Within these sectors, business models vary significantly. Some companies operate as systems of record, deeply embedded in customer workflows, with switching costs that are difficult to replicate. Others are already incorporating AI into their own product offerings, potentially strengthening rather than eroding their competitive position.
Moreover, the complexity of enterprise decision-making is unlikely to diminish in an AI-driven environment. If anything, the proliferation of tools, data sources and implementation choices may increase demand for guidance, integration and oversight.
That creates the possibility that parts of the market have been indiscriminately discounted — the proverbial “babies thrown out with the bathwater”.
The crowded logic of “picks and shovels”
A third area attracting significant attention is the infrastructure underpinning the AI buildout.
The scale of planned investment is unprecedented. Major technology platforms are expected to commit hundreds of billions of dollars to data centres, semiconductors and associated infrastructure — an outlay that, in aggregate, represents a meaningful share of economic output.
This has naturally drawn investors toward the so-called “picks and shovels” of the AI economy: chipmakers, equipment providers, power generation and cooling systems.
The logic is compelling. Regardless of which applications ultimately dominate, the infrastructure enabling AI development and deployment is essential.
Yet this is also where the investment case is most widely understood.
History suggests that when a thematic narrative becomes broadly accepted, the associated opportunities can become crowded. That does not negate the long-term growth potential of these businesses, but it does raise the bar for future returns. In some cases, the market may already be pricing a substantial portion of the expected upside.
From theme to selectivity
Taken together, these dynamics point to a more nuanced reality than current market positioning implies.
AI is unlikely to produce a simple divide between winners and losers. Instead, it is accelerating dispersion — not just across sectors, but within them. Even in areas that have performed strongly, outcomes have varied widely at the individual company level.
For investors, that has two implications.
First, sector-level views are becoming less reliable. Labelling an industry as either “beneficiary” or “victim” of AI risks overlooking the diversity of business models within it.
Second, the importance of bottom-up analysis is increasing. Understanding how individual companies generate revenue, interact with customers and allocate capital will be critical in assessing how they are likely to respond to AI-driven change.
In that sense, the opportunity set is less about identifying a single AI trade and more about navigating a shifting landscape of relative value.
A different way to think about AI exposure
For financial advisers and their clients, this suggests a reframing of how AI exposure is approached within portfolios.
Rather than asking which sectors are most exposed to AI, a more productive line of inquiry may be:
- Which businesses can integrate AI to enhance existing advantages?
- Which are mispriced due to overly simplistic disruption narratives?
- And which are being valued on the assumption of outcomes that may already be well understood?
These are inherently more granular questions, but they better reflect the nature of the transition underway.
AI remains in the early stages of both development and adoption. Its economic impact is likely to unfold over years, not quarters, and in ways that are difficult to model with precision.
What is already clear, however, is that it is reshaping how markets differentiate between companies.
For investors willing to look beyond the prevailing narrative, that may prove to be the more durable opportunity.
By Matt Reynolds, investment director at Capital Group
