Is this the next opportunity in AI?

The report surveyed 333 single family offices across 30 countries, each with an average net worth of US$1.6 billion. It is among the most comprehensive data sets available on how ultra-high-net-worth capital is actually deployed. And what it reveals, on AI at least, is a gap between stated conviction and portfolio construction.

Where the returns are

The issue is not that family offices lack AI exposure entirely. Most have it, through listed equities. The Magnificent Seven and their equivalents have ensured that any portfolio holding US large-cap stocks carries meaningful AI beta. That’s the easy part.

Christophe Aba, International Head of Investments and Advice at J.P. Morgan Private Bank, noted that to fully capture the AI opportunity, investors need to look beyond mega-cap leaders toward the enablers driving the supply chain, from semiconductors and power infrastructure to networking and cooling systems.

He adds that private market exposure is equally important, with the top 10 AI companies already valued at around US$1.5 trillion, underscoring that much of AI’s future value is still being created outside public markets.

That’s the crux. AI as a public market theme has already re-rated. AI as a private market opportunity is still compounding.

The infrastructure blind spot

The infrastructure finding is the most striking in the report. Seventy-nine per cent of global family offices allocate nothing to infrastructure, despite the asset class serving as the physical backbone of the AI buildout through data centres, power grids and connectivity networks.

Hyperscaler capital expenditure, the spending by Microsoft, Amazon, Google and Meta on AI infrastructure, is forecast to exceed US$600 billion in 2026 –  a 36 per cent increase year-on-year.

The irony is that infrastructure , including AI infrastructure, carries characteristics that family offices typically prize: long-dated cash flows, inflation linkage and tangible collateral. The very families most concerned about inflation, the report shows, allocate nearly 60 per cent of their portfolios to alternatives.

On the venture and growth equity side, the numbers are similarly striking. Fifty-seven per cent of respondents have no exposure to either asset class. Among those who do, the average portfolio allocation sits at 3.3 per cent.

This matters because the application layer of AI, the companies building on top of foundation models to disrupt industries, is largely a private market phenomenon. The businesses most likely to generate outsized returns from AI deployment over the next decade are not yet listed. They are being financed through venture and growth rounds that most family offices are structurally absent from.

J.P. Morgan’s own analysts were direct on this point, noting that the most rapid wealth creation in an innovation cycle typically occurs in private markets, in the application layer, in the companies that will disrupt existing businesses or reshape labour markets.

Why the gap exists

The mismatch is not irrational. It reflects structural friction.

Venture capital and growth equity require manager access, long lock-up periods, and the capacity to absorb J-curve drag over several years. Infrastructure funds carry similar constraints. For family offices, particularly newer ones or those with concentrated liquidity needs, these barriers are real.

The report also shows that 31 per cent of family offices hold 10 per cent or more of assets in cash. In a falling rate environment, that cash drag compounds the opportunity cost of sitting outside private markets entirely.

There is also a familiarity dynamic at work. Some 80 per cent of global family offices outsource at least some portfolio management, with the functions most commonly outsourced being US public equities, fixed income and private equity reflect where external manager infrastructure is most mature. Venture and infrastructure outsourcing is less developed, and the due diligence burden of assessing individual managers in those categories is substantial.

Convergence ahead

The report signals that change is coming, albeit slowly. Private equity leads planned allocation increases over the next 12 to 18 months, with 37 per cent of respondents intending to raise exposure. Growth equity and venture capital are both in the top 10 planned increases globally.

Infrastructure sits at 24 per cent of planned increases globally, higher among international offices. The recognition is there – the execution is lagging.

For asset consultants advising family offices, the implication is clear. The AI conversation in most client portfolios is still a listed equities conversation. Moving it into private markets, and into the infrastructure that makes AI physically possible, requires a deliberate reframing of where in the AI supply chain the most durable returns are likely to accrue.

The investors and family offices, that close this gap over the next two to three years are likely to look back at 2026 as the window when the allocation mattered most.