The Infrastructure Blind Spot: Why AI's Physical Backbone Is Going Unowned

Master the Moment and Reach Your Peak with Defoes

"Seventy-nine percent of global family offices hold zero allocation to infrastructure today, even as 65% prioritise AI exposure. Capital has flowed heavily toward AI's visible layer, chips, software and platforms, while the power and connectivity beneath it remain largely unowned."

Family offices are moving decisively toward artificial intelligence as an investment theme, yet many are overlooking the physical assets that make AI possible. J.P. Morgan Private Bank's 2026 Global Family Office Report, drawing on insights from 333 family offices across 30 countries with an average net worth of $1.6 billion, found that 79% hold zero allocation to infrastructure, even as 65% intend to prioritise AI exposure. Average infrastructure exposure across respondents sits at just 70 basis points, and only 20% report any allocation at all. For a theme this dominant in client conversations, the gap is striking.

The mission driving this AI enthusiasm is clear: family offices want exposure to what many view as the defining economic shift of the decade. The obstacle is that most have concentrated on that exposure in software, chips and platforms, the visible layer of AI, while overlooking power, connectivity and logistics, the physical layer beneath it. This is the infrastructure blind spot. AI cannot function without electricity, data centres and grid capacity, yet these are precisely the assets family offices have largely left unowned.

The scale of the underlying demand explains why this gap matters. Global data centre electricity consumption reached 415 terawatt-hours in 2024, roughly 1.5% of world electricity use, and is projected to more than double to around 945 terawatt-hours by 2030, a figure that would exceed Japan's total electricity consumption today. A single large AI-focused data centre can already consume as much power as 100,000 households, with the largest facilities under construction consuming twenty times that amount. Investment is scaling to match: global spending on data centre infrastructure is projected to approach $7 trillion through 2030, with more than $5 trillion tied specifically to AI-related build-out.

This is the consequence of the blind spot. Capital concentrated in AI's software and compute layer, without corresponding exposure to the power and physical infrastructure underpinning it, leaves portfolios exposed to a structural mismatch: heavy conviction in a theme but limited ownership of the assets that theme physically depends on. Grid constraints are already a live issue, with utilities and generation strained by demand growing four times faster than overall electricity use since 2017. McKinsey separately projects $6.7 trillion in global capital deployment for data centre infrastructure through 2030, underscoring the scale of the build-out still ahead.

Addressing this gap does not require a single concentrated bet. Access can be structured across several layers: public equities in electrical, cooling and data centre operators offer liquid exposure, specialist real estate structures provide contracted cash flows tied to digital infrastructure, and private infrastructure funds or direct co-investments can extend access to development pipelines and power-rich campuses further up the capital stack. Each layer carries a different risk and liquidity profile, and manager selection becomes increasingly important the further private capital moves into direct development.

For family offices already committed to AI as a long-term theme, understanding this physical layer is a natural extension of that conviction, not a departure from it. Defoes' role is to help clients understand how infrastructure fits within a broader AI-exposed portfolio, providing the structure and context needed to evaluate access points across public, private and real asset formats. As with all real asset allocations, infrastructure investments carry distinct risks, including illiquidity, regulatory exposure and long development timelines, and decisions should always be made independently or with an appropriately regulated adviser. Access can even extend to power generation and grid-adjacent assets directly, an area some managers now treat as core rather than peripheral to AI-themed portfolios.

Disclaimer: The content provided herein is for general informational purposes only and does not constitute financial or investment advice. It is not a substitute for professional consultation. Investing involves risk, and past performance is not indicative of future results. We strongly encourage you to consult qualified experts tailored to your specific circumstances. By engaging with this material, you acknowledge and agree to these terms.