HighworthcitizenguyBy the High Worth Citizen Editorial Team
Data centres consumed 415 terawatt-hours of electricity in 2024, roughly 1.5% of global supply. The International Energy Agency projects that figure will reach 945 TWh by 2030 — approximately the current annual electricity consumption of Japan. Private wealth has noticed. In McKinsey’s January 2026 survey of around 300 global institutional investors, 51% said they intend to increase infrastructure allocations over the next three years, the highest proportion recorded for any asset class, with family offices among the sharpest movers. Energy infrastructure has become a wealth preservation trade dressed as a technology one.
Key Takeaways
- The IEA projects data centre electricity demand will more than double to 945 TWh by 2030, from 415 TWh in 2024.
- McKinsey found 51% of institutional investors plan to raise infrastructure allocations over three years — the highest of any asset class.
- UBS’s Global Family Office Report 2026 surveyed 307 family offices averaging US$2.7bn net worth; 60% plan strategic allocation changes, favouring private credit and infrastructure.
- AI is the leading thematic exposure in the UBS survey at 65%, spanning data centres, semiconductors and adjacent power and electrification bets.
- The appeal is duration and inflation linkage, not growth — which is why the exposure suits multi-generational capital.
Why the Demand Signal Is Unusually Legible
Most thematic investment cases require a forecast. This one largely requires arithmetic. As of early 2025, 23 gigawatts of US data centre capacity was live, with a further 48 gigawatts under construction or formally committed — capacity that is contracted, financed and physically underway rather than projected. The IEA expects US data centres to account for almost half of all electricity demand growth in the country through 2030.
That load has to be served. The IEA’s modelling has renewable generation growing by more than 450 TWh and natural gas power by 175 TWh to meet data centre demand by 2035. The investable consequence sits less in the hyperscalers than in what stands behind them: generation assets, transmission and grid connection capacity, substations, transformers, cooling, and the long-dated contracted revenue attached to each.
What Family Offices Are Actually Buying
The UBS Global Family Office Report 2026, covering 307 family offices with an average net worth of US$2.7 billion, found 60% planning to change strategic asset allocation over the following twelve months, with private credit and infrastructure the clear favourites. Infrastructure allocations are expected to roughly double, while gold moves from 2% to 3% — two expressions of the same defensive instinct.
AI is the single largest thematic exposure in that survey at 65%, but the composition matters. It spans data centres, software and semiconductors with adjacent positions in power, electrification and AI-enabled healthcare. Family offices are increasingly taking the power-and-electrification leg rather than the equity leg, on the reasoning that a grid connection with a twenty-year contracted offtake is exposed to compute demand without being exposed to which model provider wins.
This also reflects a structural advantage. Family office capital has no redemption calendar. Illiquidity that constrains a pension fund or an open-ended vehicle is, for permanent capital, the source of the premium — the same logic that has driven the reallocation toward private equity secondaries among HNWIs.
What This Means for HNWIs
Treat energy infrastructure as a wealth preservation allocation rather than a growth one. Core infrastructure typically delivers contracted, inflation-linked cash flows over fifteen to twenty-five years. That duration profile matches multi-generational liabilities — succession funding, philanthropic commitments, distributions to a widening family tree — considerably better than equities do.
Distinguish sharply between the layers of the stack. Contracted generation and regulated transmission sit at the defensive end. Merchant power exposure and speculative data centre development sit at the opposite end and behave like venture risk with an infrastructure label. Both are marketed under the same heading, and the difference between them is the difference between preserving capital and underwriting a construction cycle.
Finally, access is the binding constraint. The assets with the strongest risk-adjusted characteristics are largely allocated through direct co-investment alongside specialist managers, which favours family offices with genuine in-house diligence capacity or a credible club network. Families without that should assume they are seeing the tier of opportunity that clears after the direct co-investors have taken their allocation.
Regional Comparison
The United States offers the largest pipeline and the deepest transaction volume, but connection queues in the leading data centre corridors now run for years and increasingly constrain deployment more than capital does. Continental Europe provides stronger regulatory predictability and clearer inflation indexation on regulated assets, at lower headline returns. The Gulf — particularly the UAE and Saudi Arabia — is deploying sovereign-scale capital into AI and power simultaneously, which creates co-investment access for resident family offices that is difficult to obtain from outside the region. For HNWIs already structuring around UAE tax residency, that proximity is a real and underappreciated component of the case.
Risks and Considerations
The principal risk is that demand forecasts embed assumptions about AI compute intensity that efficiency gains could undercut; a materially more efficient model architecture would leave contracted power in surplus. Construction and equipment risk is acute — transformer and turbine lead times have extended substantially, and fixed-price contracts have grown harder to secure.
Valuation discipline is the more immediate concern. Capital is arriving faster than quality assets are being created, and entry multiples on core infrastructure have compressed expected returns. Investors should also weigh political risk around electricity pricing, since rising consumer bills attributed to data centre load are already drawing regulatory attention in several US states.
The Bottom Line
Family offices are not making a technology bet. They are buying long-duration, inflation-linked, contracted cash flows that happen to be driven by compute demand — and doing so with capital that can hold the illiquidity. The discipline required is structural: own the contracted layer, avoid the merchant and development layer unless the risk is explicitly being underwritten, and refuse to pay growth multiples for what is fundamentally a preservation asset.
This article is for informational purposes only and does not constitute legal, tax, financial, or migration advice. HNWIs and family offices should consult qualified professionals in the relevant jurisdiction before making decisions based on the information presented.