AI infrastructure investing: Energy, grids and data centres

Rising power demand and AI infrastructure

AI computing can be energy-intensive. Training and running large-scale models can require substantial electricity, although demand varies by model size, hardware efficiency, utilisation and usage. Some forecasts suggest data centres could account for a much larger share of electricity demand over time, but estimates vary by region, AI adoption, hardware efficiency and grid assumptions.

Several markets have faced grid constraints, permitting challenges, or local restrictions that affect new data centre projects. These pressures are not uniform, but they show why the availability, regulation, and location of power can matter as much as demand.

These shifts may create structural opportunities. Companies involved in grid expansion, renewable capacity and power efficiency may play a larger role in AI-related infrastructure, although investment outcomes will still depend on project economics and valuation. Utilities and infrastructure providers may offer exposure to growth themes, but they can still be sensitive to interest rates, regulation, capital costs and demand assumptions.

Utilities and renewables: The energy backbone

Some utilities and renewable energy companies are scaling up capacity for large power users, including data centres. Also, large technology companies have used long-term power purchase agreements to secure renewable electricity, although contract terms, additionality and economic impact vary. These contracts may support revenue visibility for power producers and help finance new renewable projects, depending on contract terms and project economics.

This can change the demand profile for some utilities and renewable developers. Demand for AI infrastructure may support revenue for some companies, but returns depend on regulation, financing costs, execution, and valuations. Investors may research exposure through renewable-energy funds, infrastructure ETFs or utilities with clean-energy assets, subject to costs, risks and availability.

Similar themes may appear in Asian markets, where renewable capacity, grid upgrades, nuclear policy, storage and industrial demand vary by country. A multi-region approach may provide exposure to different parts of the AI power grid buildout, but it will not capture all opportunities or remove risk.

Data centre REITs and digital infrastructure

Data centres are physical infrastructure used to store, process and transmit digital information. Demand from AI workloads may support further expansion in some data-centre markets. Some data centre REITs may benefit from long-term leases and strong tenant demand, but lease structures vary, and higher energy consumption does not necessarily translate into higher net revenue.

Construction costs, power access and financing conditions can affect data-centre development economics. Tenants may place a higher value on reliable energy access, connectivity and advanced cooling, although lease economics vary. Established data-centre hubs and constrained power markets may attract investor attention, but local supply, regulation and grid access can change quickly.

Overall, data centre REITs may provide exposure to tangible assets with recurring rental income, but they can be sensitive to interest rates, leverage, occupancy, tenant concentration, development costs, and energy access. REIT ETFs and infrastructure funds may offer broader exposure, subject to costs, concentration and market risk.

Cooling, power efficiency and the data-centre supply chain

The energy challenge doesn’t end with generation. Efficient delivery, backup power, cooling and temperature control are important parts of data-centre operations. Cooling can account for a significant share of data centre energy use, although the percentage varies by facility design, climate, workload and cooling technology.

Higher chip density has increased interest in liquid cooling, heat reuse and other efficiency technologies. Some industrial suppliers may benefit from demand for cooling, electrical equipment and automation, but outcomes depend on competition, margins, delivery capacity and execution. These companies may offer exposure to industrial demand linked to data-centre buildout, but they can also be cyclical.

Investors looking for broader exposure may review global infrastructure or industrial ETFs while checking their holdings, concentration, fees, and methodology. These funds may combine longer-term infrastructure themes with shorter equipment and manufacturing cycles.

Energy security and policy tailwinds

Governments increasingly treat electricity supply, grid capacity and energy security as strategic issues. In the U.S., clean-energy incentives have supported domestic expansion, although tax-credit rules and eligibility have changed and should be checked at the time of publication. Also, European energy policy has focused on reducing dependence on imported fossil fuels and supporting the energy transition, including through REPowerEU. At the same time, some Asian markets are investing in storage, renewables, grid upgrades or nuclear capacity, depending on national policy.

This alignment between energy policy and digital growth may support infrastructure investment, although policy, rates and project economics can change. Public funding, tax credits, and private capital may support parts of this trend, but policy design, eligibility and project economics can change.

Valuation and risk

Some renewable and infrastructure companies may have seen their valuations increase during periods of strong investor interest, so valuation and entry points matter. Higher interest rates can pressure valuations for capital-intensive businesses. Investors may review factors such as contracted revenue, leverage, financing costs, cash flow visibility, and regulatory exposure.

Energy price volatility can affect margins, while long-term contracts may provide some insulation, depending on their structure. Different exposures, such as utilities, REITs, and industrials, may have distinct risk drivers, but they can still decline in value.

Ways to research exposure

AI infrastructure exposure can be researched across several parts of the market:

• Energy providers. Utilities and renewable producers with exposure to power demand from data centres.
• Infrastructure operators. REITs, data-centre operators and construction firms involved in developing and managing facilities.
• Industrial enablers. Companies supplying electrical systems, cooling, backup power, automation and related equipment.

Thematic ETFs focusing on clean energy, global infrastructure or digital real estate may provide exposure to parts of this value chain, but can be concentrated, volatile and subject to fees. Some investors research infrastructure debt or green bonds for income exposure, but income is not guaranteed, and credit, duration, liquidity and issuer risks apply.

Long-term view: The physical infrastructure behind AI

AI expansion depends partly on real assets, including power generation, transmission lines, data centres and cooling systems. These assets are often capital-intensive and slow to build, which can make planning, permitting and financing important. This may make some assets strategically important, although investment value still depends on valuation, regulation, financing costs and project economics.

This part of the AI ecosystem may offer exposure to tangible infrastructure demand and long-term contracts, but it can still be volatile and sensitive to interest rates, regulation, leverage and execution risk.

Energy supply and grid capacity may influence how quickly parts of the digital economy can scale. Utilities, renewable energy companies, and data centre operators may be part of the AI infrastructure value chain, but investment outcomes remain uncertain.