The infrastructure gap: why family offices talk about real assets but hold almost none

Interest is rising; holdings are not

Family offices are a large pool of patient capital. Deloitte Private estimated that they managed about US$3.1 trillion in 2024, and projected US$5.4 trillion by 2030 (Deloitte, September 2024). Long-dated real assets should suit them. The surveys show they rarely own any.

J.P. Morgan Private Bank surveyed 333 single family offices in 30 countries between May and July 2025. Of these, 79% had no allocation at all to infrastructure, transportation and other real assets, and the average allocation was 0.7% (J.P. Morgan, 2026). Yet 24% planned to increase infrastructure, and 41% named it an investment theme. The report also calls access to energy and natural resources a strategic imperative.

Other surveys point the same way. UBS found 37% of 307 family offices allocating towards infrastructure, and the same share towards power and resources (UBS, 28 May 2026). BlackRock surveyed 175 single family offices in spring 2025. Of these, 30% intended to increase infrastructure in 2025–2026, and 75% felt positive about the asset class (BlackRock, 16 June 2025).

Part of the gap is how surveys count

Not every report separates infrastructure. Goldman Sachs reports private real estate and infrastructure together, at 11% of portfolios (Goldman Sachs, September 2025). Some families also own infrastructure-like assets through their operating businesses. Even allowing for this, the distance between stated intent and actual holdings is wide.

Four reasons the gap persists

Illiquidity. Families value the option to change course. In the RBC and Campden Wealth survey of North American family offices, 48% named improving liquidity as a primary objective for 2025 (RBC and Campden Wealth, 2025). J.P. Morgan notes that long lock-ups limit the take-up of evergreen funds (J.P. Morgan, 2026).

Fund returns. Andonov, Kräussl and Rauh (2021) studied infrastructure funds raised in 2002–2018. Their average public market equivalent (PME), explained below, was 0.93, against 1.05 for buyout funds. They found no significant link between infrastructure cash flows and realised inflation, although their sample contained few high-inflation years. The inflation hedge, often given as a reason to invest, has to be shown deal by deal.

Concentration and ticket size. Direct infrastructure investing is dominated by a few large institutions. In the same study, the top 20 investors accounted for 55% of direct-investment observations. The authors conclude that direct investing is “not applicable for the majority of institutional investors” (Andonov et al., 2021, p. 3893).

People. In the RBC and Campden Wealth survey, over 90% of North American family offices reported difficulty recruiting (RBC and Campden Wealth, 2025). In BlackRock’s survey, 75% reported gaps in private-market analytics and 63% in deal sourcing (BlackRock, 16 June 2025). Owning infrastructure well requires project, credit and contract skills that few small teams hold.

Three tests before choosing a route

A family office comparing routes into infrastructure needs more than a headline return. Three published measures help. The first compares returns with public markets after adjusting for risk. The second prices the cost of being locked in. The third prices the cost of owning a single asset.

Test 1. Did it beat public markets, after risk?

The public market equivalent (PME) used by Andonov et al. (2021) discounts every cash flow at the return of a public index. In our notation, with I_t the growth of the index up to time t:

$$\mathrm{PME}=\frac{\sum _t{D}_t/I_t+{\mathrm{NAV}}_T/I_T}{\sum _t{C}_t/I_t}$$

Here C_t is capital paid in, D_t is cash returned and NAV_T is the remaining value at the end of the period T. A PME above 1 means the investment beat the index. Infrastructure funds raised in 2002–2018 averaged 0.93 (Andonov et al., 2021).

The simple PME assumes the investment carries the same market risk as the index. Korteweg and Nagel (2013) relax this with a stochastic discount factor, which values each cash flow by the state of the market when it arrives:

$$M_t=\exp (at-b{r}_{m,t}),\mathrm{GPME}=\sum _t{M}_t(D_t-C_t)$$

Here r_m is the cumulative log return of the market up to time t, and a and b are estimated from public-market data alone. A generalised PME (GPME) of zero means the investment earned a fair return for its market risk. The simple PME is the special case a=0, b=1. On this risk-adjusted basis, Andonov et al. (2021) estimated a GPME of −0.257 for infrastructure funds, which they describe as comparable to venture capital.

Test 2. What does it cost to be locked in?

Illiquid assets cannot be sold when the owner chooses. Ang, Papanikolaou and Westerfield (2013) model an asset that can only be traded at random moments, arriving at an average rate λ per year. The chance of being able to sell within t years is then:

$$P(\text{sale possible within }t)=1-e^{-\lambda t}$$

In their baseline case, a risk-averse investor who can trade on average once every ten years gives up about 22% of total wealth in certainty-equivalent terms. If the same ten-year interval were known in advance, the cost would be about 2.8%. Most of the cost comes from not knowing when the exit will come. Even with trading opportunities once a year on average, the optimal share in the illiquid asset fell from 0.60 to 0.37.

For a family office, the lesson is practical. An exit route agreed before investing, such as a refinancing date, a sale right or a market for shares, reduces the cost of illiquidity more than a shorter lock-up whose end is uncertain.

Test 3. What does it cost to own one asset?

A fund spreads risk across many assets. A single asset does not. Kerins, Smith and Smith (2004) estimated the cost of capital for investors who cannot diversify. With a 4% risk-free rate and a 6% market premium, a well-diversified investor’s cost of capital for their sample was 11.4%. For an entrepreneur with 25% of total wealth in the venture, it was 40.0%. Their sample was high-technology ventures, not infrastructure, but the logic carries over. The required return on a single asset rises with the share of family wealth it represents.

Taken together, the three tests show what moving up the ladder demands. Direct ownership must beat public markets after risk, pay for its illiquidity and pay for its concentration.

Framework: the infrastructure access ladder

There are five routes into infrastructure. Each step up the ladder gives the family more control and better information. Each also requires more in-house capability and a larger, more concentrated commitment.

Rung 1. Listed infrastructure

Shares in listed utilities, transport and telecoms companies give daily liquidity and small tickets. Research on 1,400 listed infrastructure assets found low risk and low correlation with other industries, but strong company-specific risk (Boral, 2013, summarising Rothballer and Kaserer). Risk varied with sector, region, asset maturity and regulation. The family owns a share of a company, not an asset, and its returns move with equity markets.

Rung 2. Infrastructure funds

Funds are the standard route for investors without a deal team. Across private markets, J.P. Morgan found 67% of family offices using drawdown funds and 35% using evergreen funds, with long lock-ups limiting take-up (J.P. Morgan, 2026). Funds bring diversification and a professional manager. They also bring fees, blind pools and long commitments. Andonov et al. (2021) found that funds which exit deals faster tend to perform better, which sits uneasily with the patient-capital case for infrastructure.

Rung 3. Co-investment alongside a sponsor

A co-investment puts the family’s money directly into one deal led by a fund manager, usually at lower fees. In the RBC and Campden Wealth survey, co-investments made up 8% of the average private-markets portfolio, a modest but growing share (RBC and Campden Wealth, 2025). The risk is selection. Fang, Ivashina and Lerner (2015) studied the direct deals of seven large institutions. Co-investments offered by sponsors returned about 8 IRR points less than the sponsor’s own fund. Their sample was private equity, not infrastructure, but the information problem is the same.

Rung 4. Club deals in a single asset

In a club deal, several investors buy into one asset together, often with an industrial partner. Club deals made up 69% of family-office deal flow in the first half of 2025 (PwC, 2025). A recent infrastructure example shows the form. In April 2026, a consortium bought 30% of Nordic Ferry Infrastructure, which runs 59 routes with 101 vessels, from EQT Infrastructure V (EQT, 9 April 2026). The consortium joined a ferry operator, Gotlandsbolaget, with a pension fund and a foundation-owned investment group (White & Case, 13 April 2026). The reported price was €510m (Riviera, 15 April 2026).

The operator brings know-how, the pension fund brings long-term capital and the investment group brings flexibility. The weak point is governance among the partners.

Rung 5. One asset, one company

At the top of the ladder, the family invests directly in a special purpose company that owns one project or one vessel. The company signs the contracts, takes project debt and pays its lenders before its shareholders. Lenders hold a defined security package over the asset, its contracts and its income. That package also protects the equity, because every party can see how cash flows.

Fang et al. (2015) found that investors’ own solo deals outperformed co-investments, and linked the difference to information. The same study, and Andonov et al. (2021), show the cost: concentration, larger tickets and the need for specialist people. This rung works only if the family brings in the skills it lacks, before closing and after it.

What’s the bottom line?

The infrastructure gap is real, and it is not a lack of interest. Families say they want real assets. The routes on offer either lock their capital in funds whose returns have trailed buyout funds, or hand them deals selected by someone else.

The way across is not to jump to the top rung unprepared. It is to choose the rung that matches the family’s capability, and to buy the capability it lacks. For a family that wants control and information, one asset in one company, structured like a project financing, is the most transparent route. It is also the most demanding.

Summary

Surveys from 2025 and 2026 show family offices naming infrastructure as a theme while holding almost none of it. Illiquidity, fund returns, deal concentration and scarce skills explain the gap better than any lack of appetite. Five routes lead into infrastructure, from listed shares to a single asset in its own company. Each step up brings more control and information, and demands more capability. Three published tests compare them: risk-adjusted return against public markets, the cost of illiquidity and the cost of concentration. For families that want to own real assets directly, the decision is less about the asset class than about the route and the skills behind it.

This Insight is for general information only and does not constitute investment, legal or tax advice. Figures from public sources are as reported at the dates shown.