Why the next AI trade may be built in concrete, power and land

Artificial intelligence is usually discussed in the language of software, chips, and market concentration. For advisers, however, the more interesting question may be what happens when a digital revolution starts making very physical demands on the real world.

Every large AI model needs somewhere to run. It needs servers, cooling, power, fibre, land, specialist buildings, and resilient tenants willing to commit to long leases. That makes AI more than a technology theme. It is also becoming a real assets theme.

How the listed property universe has changed

With implications that reach beyond the obvious mega-cap technology stocks and into the evolving universe of global listed real estate. That shift matters because property markets are not static.

Over the past few decades, global REIT indices have moved well beyond the old shorthand of office towers, shopping centres, and logistics sheds. Dexus describes this as a shift from traditional office and retail dominance towards sectors including industrial, self-storage, healthcare, data centres, and mobile towers. Technology-driven real estate now makes up a much larger part of the opportunity set.

For advisers, that may require a mental reset. Advisers still too often treat listed property as a yield sleeve or a simple proxy for interest rates. But the deeper story increasingly centres on the cash flows attached to structural demand. In the case of AI, those cash flows attach to buildings that look unglamorous but have become essential to the digital economy.

Dexus Global REIT Fund portfolio manager David Kruth has framed the broader opportunity in precisely those terms. He points out that data centres and senior housing are expected to outgrow the broader real estate market.

“This is where we believe investors with an international focus can secure a relatively attractive and growing return in the years ahead.”

Dexus anticipates data centres and seniors housing will enjoy 6 to 7 per cent annual growth over the next five years, compared with around 3 to 4 per cent for the overall real estate market.

Digital growth needs physical assets

The data centre story is the cleanest example of the new property logic. It is not simply that AI demand is strong. It is that the supply response is difficult.

As Kruth explains, “Data Centre REITs own and operate facilities that house servers and networking equipment for cloud computing, web hosting, streaming, AI workloads and enterprise IT infrastructure.” These are not generic buildings. They sit at the intersection of real estate, digital infrastructure and energy, and Dexus notes that the fund has direct exposure to data centres across North America, Europe and Asia.

That global opportunity is not without volatility. The emergence of more efficient AI models, including DeepSeek, triggered a repricing of AI-exposed investments as investors questioned whether infrastructure spending would need to reach the same scale.

Dexus believes those fears have proved overdone. Hyperscaler tenants are expected to keep lifting capital expenditure as cloud businesses expand and public AI adoption gathers pace.

The physical bottleneck is power. Dexus notes that while demand for data centres is growing quickly, supply is constrained, “although more so by power than access to capital”. That line is important for advisers because it explains why this is not just another growth story vulnerable to instant competition. The constraint is not merely financing. It is access to energy, suitable land, and infrastructure approvals.

This is where property fundamentals reassert themselves inside a technology narrative. Strong demand is useful, but strong demand combined with tight supply and low vacancy is what creates rental tension.

Dexus points to established specialist REITs such as Digital Realty, Equinix, and Keppel Data Centre REIT as operators positioned to help satisfy that demand, while also noting that mobile tower REITs may enjoy a new wave of leasing requirements as AI-infused applications increase data requirements on mobile devices.

The office question is more nuanced

The other side of the AI story is the office. Here, the thesis becomes less elegant and perhaps more useful for advisers.

In the March quarter, the Dexus Global REIT Fund underperformed its benchmark by 50 basis points. The manager attributed that largely to its overweight position in office REITs. Dexus noted that office sentiment has been clouded by concerns over how AI development will affect space demand, with the market “witnessing lower hiring following the penetration of AI” and ultimately questioning the long-term case for office occupancy.

That does not mean the office sector is finished. It does mean old assumptions need to be discarded. AI may reduce demand for some kinds of labour and therefore some kinds of office space, while increasing demand for other forms of work, collaboration, and innovation. The blunt version of the argument, that AI kills the office, is too easy. The more useful version is that AI increases dispersion.

Premium assets in markets linked to technology, research, healthcare, finance, and education may not behave like secondary stock in weaker locations. Buildings that can attract high-quality tenants, support new ways of working, and provide genuine amenity may still have a role. Buildings that were already marginal may struggle further.

For advisers, that means office exposure needs to be interrogated rather than dismissed. The question is not whether a portfolio owns office, but what kind of office, in which market, at what valuation, with what tenant base and balance sheet support.

Why the average REIT may not be enough

This is the point where active management becomes crucial. AI is not lifting all property assets equally. It is strengthening some cash flows, challenging others, and increasing the gap between winners and losers.

The March quarter showed that clearly. Data centre REITs were the strongest global REIT sub-sector, returning 12.96 per cent, supported by AI-driven demand and limited supply. Assisted living REITs also gained, while residential REITs were the biggest detractor due to a large supply and decelerating rents.

At the portfolio level, Dexus benefited from North American retail stock selection. Its underweight to Digital Realty detracted, however, as the global data centre provider caught the full force of the AI tailwind.

The lesson is not that advisers should chase the latest performing REIT sector. It is that the composition of listed real estate matters more than it used to.

Dexus makes this argument through the lens of income growth. “When investing in REITs over the long term, net operating income, rental revenue minus operating expenses, is a key driver of total returns,” Kruth said. “Consistent, growing NOI fuels both.”

In Dexus’s view, gains from higher rent and occupancy are likely to see senior housing, data centres, and mobile towers deliver significantly higher earnings growth than other REIT sectors.

That is a more thoughtful way to understand the AI trade. It is not a speculative punt on technological enthusiasm. It is a question of which real estate assets can convert digital demand into rent, occupancy, and ultimately net operating income.

Where income growth is actually coming from

The Dexus Global REIT Fund is one expression of that idea. The fund prioritises REITs with resilient, growing income streams underpinned by robust balance sheets. It targets growth at a reasonable price through active, high-conviction positioning. The broader objective is to generate sustainable income, maintain real capital value over a five-to-seven-year period, and provide lower-than-market volatility relative to its benchmark.

For advisers, the broader point is that listed property is becoming a more sophisticated allocation than the old yield conversation allowed.

AI may be born in code, but it is being housed in physical assets. The opportunity is to identify which assets are becoming more essential as the digital economy expands, and which are being left to defend yesterday’s assumptions.