By Richard Warne, senior portfolio manager, Copia Capital
Infrastructure is not always the first area of the market that comes to mind when people think about where to invest their money. However, with the Iran war and the AI listings now lining up, markets are entering new and potentially complicated territory, which makes it a good time to look at the parts of the market that are driven by long-term demand rather than the day’s headlines.
Infrastructure exposure offers a way into structural demand that is less sensitive to short-term market swings, with the prospect of stable returns over time. Energy security, real assets, regional development, data centres and digitisation all sit under that heading and each can earn its place in a diversified portfolio.
Opportunities in infrastructure
The Iran war has shown economies of all sizes they need to move faster on energy security, and the transition to cleaner energies is interlinked with how they will do it. Infrastructure investments related to this area are well-placed to benefit.
The numbers behind this are large. Based on analysis by the International Renewable Energy Agency (IRENA) and the International Energy Agency (IEA) reports, an estimated $1trn a year will be spent on energy security and transition globally between now and 2050. Much of this is driven by AI’s need for data centres. Financing of $58bn has been committed across 42 data centres so far this year, up from $34bn across 34 deals at the same point in 2025, according to Dealogic.
See also: WisdomTree launches AI infrastructure ETF
That raises fair questions about how far data centres are fuelling infrastructure valuations and what happens to demand if there is a correction in AI stocks. Two things are worth considering here.
First, not everything trades on the same fundamentals. Real assets such as buildings, energy storage, roads and railways are structural by nature. Though not immune to stress, over the long term these projects can deliver stable income through rent, tolls and other recurring revenue. Property, utilities and the infrastructure associated with essential services usually carry inflation protections too, which supports cash flow when valuations come under pressure.
There is evidence of commitment to these assets, with analysis of the IRENA and IEA reports forecasting spending on infrastructure within emerging markets alone to be at around $1trn a year between now and 2030.
That research found the building sector, which includes data centres, was responsible for nearly 45% of the total growth in electricity demand in 2025.
Analysis also puts spending on digitisation at $1trn a year currently, and it is still rising. Even if data centre spending eases, the infrastructure it has put in place has an underlying value – railway and telecoms infrastructure built during their respective bubbles are still in use long after the share prices corrected.
Second, the money already committed to AI infrastructure is largely spent and unlikely to change. The US leads on sheer scale, but will be reluctant to cede ground to China, which is developing and building infrastructure at a similar rate. Although corrections in AI valuations may happen at some point, spending commitments continue to rise above where most expected them to go and there is little sign of that slowing.
The idea of having an element of public ownership in AI companies has drawn interest from a diverse group including veteran US democratic socialist Bernie Sanders, OpenAI CEO Sam Altman, and Donald Trump. This may not be the eventual route taken, but it nevertheless highlights a broad appetite to keep AI innovation going.
IPO implications and the longer view
The IPO plans for AI are a clear signal of how much institutional capital is committed to this build-out. SpaceX has already completed the largest IPO on record, tied heavily to AI and computing infrastructure. The mechanics behind these listings are revealing. Index providers and underwriters have adjusted their usual rules to get the flotations away, from faster index inclusion to larger retail allocations than normal.
That a company which made a $5bn loss last year is now among the largest listings in history says more about the current appetite for AI than about the underlying health of US equities. Anthropic and OpenAI, the makers of Claude and ChatGPT, are expected to pursue sizeable listings this year or next as well.
See also: Fund in 5ive: The WS Nomura Global Infrastructure Securities Fund
If these companies list at, and maintain, high valuations, AI will become an even larger and more concentrated part of US equity indices than it already is. Even before the Iran war and these IPOs, the markets were looking more susceptible to periods of volatility, and a heavier weighting to a single theme will make diversification harder to come by.
Infrastructure will not insulate a portfolio completely, and parts of it are exposed to the same AI demand that could correct. What it offers is a different set of return drivers that do not all depend on the same handful of AI companies.
