By Martin Conlin, Head of Australian Equities, Schroders
AI scepticism has not been lucrative to date. In fairness, scepticism of any kind has not been particularly lucrative. Markets appear increasingly willing to capitalise distant possibilities at today’s prices, often with little concern for the path between here and there. Claims of monumental future demand, extraordinary addressable markets and investment opportunities of seemingly limitless scale are greeted with enthusiasm rather than scrutiny. That may prove entirely justified. Or it may not.
Few themes have captured the market’s imagination more completely than copper.
The investment case appears compelling. Artificial intelligence requires data centres. Data centres require electricity. Electricity infrastructure requires copper. Electrification requires copper. Energy transition requires copper. The conclusion appears obvious: demand for copper rises materially, copper prices remain elevated and copper-exposed equities continue to prosper.
We do not struggle with the proposition that copper demand will increase. It probably will. What strikes us as considerably more controversial is the leap investors appear willing to make from higher copper demand to permanently elevated industry profits. Demand growth and shareholder returns are not the same thing. They have never been the same thing.
Throughout history, investors have repeatedly confused products that change the world with investments that generate attractive long-term returns. Railways transformed economies. Telecommunications transformed communication. Fibre networks transformed data transfer. In each case, enormous economic value was created. The proportion of that value ultimately captured by shareholders was often far lower than expected when enthusiasm was at its peak.
Copper mines are not software businesses
One of the more interesting features of the current environment is the extent to which financial markets appear increasingly willing to apply software-like economics to industries that remain firmly grounded in the physical world.
Technology companies have become accustomed to extraordinary scalability. Once software is built, serving another customer often costs very little. Network effects can create dominance. Margins can expand rapidly. Revenue growth can persist for long periods.
Copper mines possess none of these characteristics.
Every additional tonne requires capital. Every expansion requires labour, equipment, permitting, power and infrastructure. Ore grades decline. Costs rise. Governments seek larger shares of profits. Communities demand greater economic participation. Nature has an annoying habit of refusing to conform to spreadsheet assumptions.
Yet much of the enthusiasm surrounding copper seems to assume precisely the opposite. Investors appear willing to extrapolate current conditions far into the future, capitalising demand forecasts as though supply will remain conveniently constrained indefinitely.
That strikes us as unlikely.
The current copper price would justify exceptionally attractive returns across a vast range of development projects. Existing producers are generating very strong economics. Prospective producers are finding funding easier to secure. Management teams are discussing expansion plans with increasing confidence. None of this is surprising. High prices are supposed to encourage supply.
That is how markets work.
The central assumption embedded in many copper-related valuations appears to be that demand growth will substantially exceed supply response for many years to come. Perhaps that happens. Perhaps it does not. Commodity markets have a long history of encouraging precisely the investment behaviour that ultimately undermines elevated prices.
The laws of economics have not been abolished
The more persuasive argument for copper bulls is not artificial intelligence itself but rather the breadth of demand drivers supporting the metal.
AI infrastructure requires copper. Renewable energy requires copper. Electric vehicles require copper. Grid upgrades require copper. Industrial electrification requires copper.
All true.
What receives less attention is that higher prices alter behaviour.
LME Copper: Aluminium Ratio
Source: Schroders, Morgan Stanley, MetalsCut4U.
Every commodity analyst is comfortable forecasting demand growth at current prices. Fewer appear interested in forecasting behaviour at significantly higher prices. Yet higher prices encourage substitution, efficiency improvements, recycling, new technologies and delayed investment decisions. They encourage users to find alternatives. They encourage engineers to solve problems differently.
Demand does not exist in isolation from price.
The assumption that future demand will simply compound irrespective of future pricing strikes us as one of the more dangerous features of current commodity analysis. Markets often extrapolate growth while forgetting adaptation. Real economies adapt constantly.
The same logic applies to electricity consumption more broadly. Forecasts showing exponential growth in data centre demand are certainly plausible. The challenge lies in translating a plausible demand forecast into a valuation conclusion.
Many of these forecasts appear to assume not only that demand grows, but that demand grows dramatically, that supply remains constrained, that competition remains rational, that governments remain supportive and that investors can purchase exposure to all this at sensible prices.
History suggests the probability of all those assumptions proving simultaneously correct is generally lower than markets assume.
Direction is everything
One of the more frustrating developments in modern markets is the extent to which direction increasingly trumps level.
Earnings moving higher are celebrated almost irrespective of why they are moving higher. The distinction between operational excellence and commodity price exposure often appears secondary. If earnings are being upgraded, investors are inclined to pay more. If earnings are being downgraded, investors are inclined to pay less.
It is simple. It is intuitive. It is also frequently dangerous.
Commodity producers naturally benefit from this behaviour during the favourable stages of the cycle. Rising prices generate rising earnings. Rising earnings generate upgrades. Upgrades generate enthusiasm. Enthusiasm generates higher multiples.
ASX200 Sector PE Ratios
Source: LSEG, Schroders, 12m forward PE 30 June 2026.
The process becomes self-reinforcing.
Eventually, however, investors discover they have paid a high multiple of elevated earnings generated by unusually favourable conditions.
We are not suggesting copper prices must collapse. We are suggesting that investors should be careful about paying growth multiples for businesses whose earnings remain heavily influenced by variables outside management’s control.
A great business with a poor valuation can be a disappointing investment.
A cyclical business with an optimistic valuation can be particularly dangerous.
The difference between economic value and shareholder value
Even if the most optimistic AI forecasts prove correct, an important question remains.
Who captures the value?
The assumption underpinning much contemporary investment thinking appears to be that value naturally accrues to those providing the inputs. We are less certain.
Large language models may generate substantial productivity gains. Data centres may become critical infrastructure. Electricity demand may rise considerably. Copper consumption may increase materially.
None of those outcomes automatically guarantee extraordinary returns for copper producers.
In fact, if competition and supply respond appropriately, much of the value created could accrue elsewhere. End users may benefit. Consumers may benefit. Productivity may improve across the economy. Input suppliers may simply earn acceptable returns rather than extraordinary ones.
Investors often underestimate how much future success is already embedded in current prices.
The mere fact that a theme is important does not mean exposure to that theme is attractive.
A good story, an expensive price
The copper story is easy to understand.
That may be part of the problem.
Markets generally do not reward investors for identifying obvious truths. They reward investors for identifying truths that are misunderstood or mispriced. That artificial intelligence is likely to increase demand for electricity, infrastructure and copper is increasingly accepted wisdom. Few investors would dispute it.
Source: Refinitiv, SIMAL estimates, updated 30 June 2026.
The question is not whether demand rises.
The question is whether current prices already assume too much.
At today’s commodity prices, the economics of existing operations are extremely attractive. Development projects look increasingly viable. Investors have become comfortable extrapolating favourable conditions well into the future. The assumption appears to be that strong demand growth guarantees future scarcity.
Our experience suggests the commodity cycle is rarely that accommodating.
We do not deny the importance of copper. We do not deny the importance of AI. We simply struggle with the proposition that investors need pay any price to participate.
Good businesses can be poor investments when purchased at inflated prices. Good themes can deliver disappointing returns when expectations become excessive. The history of markets is littered with examples.
The future may indeed require more copper.
Whether today’s copper-exposed equities deliver satisfactory returns from today’s prices is a very different question.
Interested in learning more about the Schroder Wholesale Australian Equity Fund? Visit https://www.schroders.com/en-au/au/local/.
[pls insert generic SIAA disclaimer]
