Artificial intelligence could add as much as 0.5 percentage point to annual global growth over time, according to IMF Managing Director Kristalina Georgieva. Moving from 3.0% to 3.5% growth for a decade would compound into an enormous increase in output. The same investment boom is also using energy, labor and capital at a moment when governments and borrowers face higher interest costs.
In an October 7 speech in Singapore, Georgieva said AI hardware and related technology already accounted for more than one-tenth of world goods trade. She compared the investment wave with the build-outs of railroads, electricity networks and telecommunications. The 0.5-point estimate is a potential gain if adoption is managed well, not the IMF's unconditional base forecast.
The growth channel
AI can raise productivity by automating tasks, improving scientific discovery and allowing workers to produce more with the same time and capital. The benefits are likely to be uneven. Countries that supply chips, power, data centers, models and skilled labor may capture more of the initial investment, while economies outside those networks risk falling further behind.
Productivity also arrives with a lag. Infrastructure spending affects demand immediately, while the output gains depend on redesigning processes, training workers and diffusing tools across smaller firms. That timing gap is where financial risk can accumulate.
The compounding arithmetic illustrates the upside. If an economy grew at 3.5% rather than 3.0% for ten consecutive years, its output level at the end of the period would be roughly 5% higher, all else equal. That is an illustrative scenario, not an IMF forecast, because the 0.5-point figure is an estimated potential contribution and the timing, adoption rate and distribution of gains remain uncertain.
Energy turns a technology boom into an inflation question
Georgieva described a negative energy-supply shock occurring alongside the positive AI-demand shock. Data centers, manufacturing and electrification compete for power just as expensive oil and gas raise costs. If energy supply cannot respond, the investment boom can push prices and bond yields higher before productivity offsets them.
Higher yields affect both companies and governments. AI-related borrowers issue debt to fund infrastructure, competing with sovereigns for capital. Governments already have limited room: Georgieva said global public debt was on course to exceed 100% of GDP, while the long period in which interest rates sat below economic growth had ended.
That change in the interest-growth relationship matters for debt dynamics. When the effective interest rate exceeds nominal growth, governments cannot rely on economic expansion alone to stabilize the debt ratio; primary deficits become more expensive to carry. Heavy private borrowing for AI infrastructure can add to demand for capital at the same time, keeping the hurdle rate for both public investment and corporate projects elevated.
Why an earnings miss could spread
The IMF chief warned that hyperscaler leverage, concentrated AI investment and large global holdings of U.S. equities could transmit disappointment beyond individual companies. If expected AI revenue fails to cover depreciation, power and financing costs, companies may cut capital expenditure. That would affect chipmakers, server vendors, utilities, construction and credit markets at once.
This is a risk scenario, not a prediction of collapse. Strong earnings can validate the spending and support productivity gains. The policy challenge is to keep inflation expectations anchored, supervise leveraged exposures and rebuild fiscal credibility without choking off useful investment.
The distribution question also affects the macro result. If gains accrue mainly to a small set of firms and highly skilled workers, headline productivity can rise while wage and regional gaps widen. Broader adoption, worker training and competitive access to compute would determine whether the growth benefit supports consumption and tax revenue or remains concentrated in asset values.
The next evidence is not another headline capital-spending number. It is the return on that spending: AI-related revenue, customer productivity, operating cash flow, power availability and debt service. The technology can raise the global growth ceiling, but the path to those gains may first increase the economy's sensitivity to rates, energy and earnings execution.