IMF: Growth Is Splitting Between AI Winners and Energy-Exposed Economies

The IMF now expects the world economy to grow 3.1% in 2026 and 3.4% in 2027, while global inflation reaches 4.7% this year as energy and food costs stay high. Those aggregates are less important for location decisions than the divide beneath them: economies benefiting from the AI investment boom or favourable terms of trade are outperforming, while energy importers and lower-income countries face weaker growth, higher costs and less room to absorb another shock.

This is not a conventional developed-versus-emerging-market split. The favourable side includes countries embedded in the AI capital cycle, whether through data centres, semiconductors, power equipment or the services that finance and operate them. It also includes commodity exporters whose terms of trade have improved. The exposed side includes locations dependent on imported energy, especially where currency pressure, food inflation and high sovereign borrowing costs already constrain policy.

For investment agencies, the immediate winner is the location that can convert macro resilience into a credible operating case. Reliable power at a predictable price now affects far more than energy-intensive industry. It determines whether a jurisdiction can support data centres, advanced manufacturing and cold-chain logistics without passing volatility through to investors. Access to capital matters in the same way: a government facing high debt-service costs may still announce incentives or infrastructure, but companies will discount promises that lack funded delivery plans.

The losers are locations still selling yesterday's cost model. A low nominal wage is less persuasive if imported fuel raises transport and electricity costs, food inflation erodes real incomes, and fiscal stress delays grid or site works. Equally, an economy can post strong headline growth from an AI investment surge while capturing little beyond construction unless it has the skills, suppliers and market access to retain higher-value activity.

Agency leaders should therefore rerun their priority-sector propositions against four questions. How much of the operating cost depends on imported energy? Is grid capacity contracted or merely planned? Can the government fund its side of the project under current borrowing conditions? And does the location participate in the investable parts of the AI value chain, or merely consume imported technology? These tests will separate robust projects from announcements that work only under a benign energy price or financing assumption.

The strategic response is not to abandon exposed markets. It is to price the exposure honestly and build around it. Energy-importing economies can improve their position through diversified generation, storage, interconnection and procurement arrangements. Lower-income locations can narrow their target lists to projects with resilient export demand, realistic infrastructure requirements and private financing. AI-adjacent economies should measure domestic supplier spending, skilled employment and exportable services rather than count computing capacity as value capture.

Use the IMF's October country data to stress-test target sectors and live projects before 2027 budgets and incentive packages harden. Recalculate power, logistics, wage and public-finance assumptions under higher energy and food inflation, and distinguish contracted infrastructure from policy intent. The locations that can show funded delivery and controlled exposure will gain; those relying on a global growth headline will misread the market.