The same AI spending may support several assets owned by a family

This week’s focus

That is one question worth bringing to a discussion of family assets after reading the International Monetary Fund’s World Economic Outlook, published on 13 October.

The IMF forecasts global economic growth of 3.1% in 2026 and 3.4% in 2027. These are economic growth forecasts, not forecasts of asset returns. Chapter 1 is especially useful for its explanation of two forces behind growth: energy supply shocks raise costs, while investment in AI and related hardware and infrastructure supports demand.[1][2]

Spending becomes someone else’s revenue along the supply chain

The IMF’s Global Financial Stability Report, released the same day, provides a specific example. It reviews 14 publicly disclosed AI-related deals since 2023, involving about US$300 billion in equity investment, US$650 billion in debt and US$1.4 trillion in future revenue commitments for services or chip purchases. These amounts describe different arrangements within that set of deals. They cannot be added together as a measure of the entire AI market, and future revenue commitments are not cash already received.[3]

Such arrangements connect investors, borrowers and customers. Investors provide capital, and recipient businesses in turn commit to buying related services or chips. If a recipient gets into financial difficulty, both the value of the investment and expected sales revenue may suffer. This interdependence is the risk the IMF identifies.[3]

The following hypothetical sequence illustrates how family assets might be affected. It is not an IMF forecast for any particular company.

If a technology company delays an expansion, equipment and construction orders may fall first, affecting suppliers’ revenue. A data centre still seeking tenants may take longer to sign new leases, while a property with long-term leases may see no immediate change in short-term rental income. Technology share prices may adjust to changing market expectations before those operating changes occur.

The three assets respond at different times, yet may depend on the same capital expenditure decision. Classifying them only as listed or unlisted, equities or property, would miss that common dependence.

The protection a contract provides depends on its terms

Common dependence does not mean every asset will necessarily fall together. Lease duration, the spread of customers, order cancellation terms and project borrowing all affect the size and timing of the impact.

For example, a long-term lease already being performed and a pre-letting arrangement that has not yet been signed cannot support the same assumption about income stability. Fixed rent does not remove the landlord’s risk: the tenant’s ability to perform, refinancing needs and costs still borne by the owner all affect the cash ultimately available to shareholders.

These differences are more useful than assigning every AI-related asset the same risk rating.

Energy costs may affect cash from another direction

The IMF also notes that economies’ exposure to energy prices differs according to their reliance on energy imports and their position in technology supply chains.[2]

For family capital, this means a business benefiting from AI investment may also face higher energy costs. The effect depends on the contract: whether the tenant pays for electricity or it is included in a fixed service fee, and whether and when costs can be repriced. Growing demand for data centres alone does not establish that a project’s cash returns to shareholders will rise.

G70’s view is that the outlook is useful for examining how family assets generate income, rather than for choosing a trading direction. Looking at major revenue sources alongside contract terms and debt maturities makes it easier to see which business would lose cash first if the same customers deferred spending, and which asset might need a capital injection at the same time. Working capital reserves should be estimated from those specific obligations.