Growth holds. The exposure map does not.

The IMF’s stable global headline is real, but it is not the decision. War-driven energy pressure and an AI-led investment boom are redistributing growth, inflation and financing risk across countries, sectors and companies.

Executive judgement

The July 2026 World Economic Outlook Update projects global growth of 3.0% in 2026 and 3.4% in 2027, broadly unchanged on a cumulative basis from the IMF’s April view. That apparent stability conceals a material rotation in the sources of growth and risk. Energy exporters and economies embedded in the global technology value chain receive support; energy importers, fiscally constrained states and businesses without access to technology demand absorb more of the shock. Global headline inflation is projected at 4.7% in 2026 and the IMF says the disinflation trend has stalled. A single “global growth” assumption is therefore an unsafe planning input.

The correct operational response is to replace one macro case with an exposure model. For a company, the relevant variables are not only country GDP. They include direct and indirect energy intensity, dependence on the Strait of Hormuz and other constrained routes, participation in AI hardware or data-centre investment, tariff exposure, refinancing dates, customer purchasing power and the ability to pass through costs. Two companies in the same country can face opposite outcomes because their exposure vectors differ.

Evidence

The IMF identifies two forces pulling in opposite directions: the lingering energy shock from the war in the Middle East and a technology-driven investment boom. Its July assumptions include a gradual reopening of the Strait of Hormuz beginning in mid-July, normalization toward the pre-war state by March 2027, policy and geopolitical uncertainty remaining elevated through 2027, and an AI investment cycle that moderates without an assumed exogenous productivity boost. Those are forecast assumptions, not observed facts. Any plan derived from the forecast must therefore show how results change if one or more assumptions fail.

World trade volume growth is projected to slow from 5.0% in 2025 to 3.5% in 2026 before recovering to 4.3% in 2027. The IMF attributes this pattern to front-loading, tariffs, changing production chains, trade diversion and technology-related trade. Advanced economies are projected to grow by 1.7% in 2026 and 1.8% in 2027, but the country pattern is uneven: the United States is projected at 2.3% in 2026, the euro area at 0.9%, the United Kingdom at 1.0% and Japan at 0.6%. These figures are useful as a macro baseline, not as a substitute for sector or company analysis.

The IMF’s comparison groups make the distributional mechanism visible. Energy exporters are partly cushioned through terms of trade. Energy importers experience a greater drag unless technology-related activity offsets it. The update identifies Korea, Malaysia, Taiwan Province of China and Thailand as the four leading AI-hardware exporters in its comparison. This does not mean every technology company or every resident of those economies benefits equally. It means their national production and trade structures create a different aggregate sensitivity to the AI capital cycle.

Mechanism

The first transmission channel is energy. A supply disruption raises oil and gas prices, worsens import bills, pressures exchange rates and can feed consumer prices. The second-round effect depends on wage setting, inflation expectations, competition and fiscal cushioning. Governments may temporarily absorb part of the shock through subsidies or tax changes, but that transfers the cost to public balance sheets. Firms experience the shock directly through fuel and electricity and indirectly through transport, materials, supplier pricing and reduced household purchasing power.

The second channel is technology investment. Demand for semiconductors, servers, networking, data-centre construction, cooling and power infrastructure can support exporters and investment destinations. The benefit is concentrated. An economy that imports compute but exports little of the value chain may face higher electricity and capital demand without receiving the same export lift. A company selling into the AI build-out can outperform a weak domestic cycle; a company outside that demand cluster may see no benefit at all.

The third channel is finance. Commodity volatility can change inflation expectations and delay monetary easing. Higher or more persistent rates affect borrowers at refinancing, not uniformly on the day a central bank changes policy. Businesses with floating-rate debt, short maturities or thin interest coverage are more exposed. Financial conditions can also tighten through risk premia even when policy rates do not move. The IMF reports that conditions tightened sharply in April and later eased; that path illustrates why a point-in-time rate alone is not an adequate financing indicator.

The fourth channel is trade architecture. Tariffs, sanctions, rerouting and supply-chain adaptation change lead times, working-capital requirements and compliance costs. Trade diversion can create opportunity for alternative suppliers while destroying established margin structures. The aggregate recovery in trade projected for 2027 says nothing about which routes, products or firms capture it.

Counterarguments

One counterargument is that the global economy has already absorbed the shock better than feared, so extensive downside planning may be excessive. The IMF itself notes limited evidence of broad second-round effects, inventory drawdowns, expanded production outside the Gulf, lower energy intensity and a higher renewable share. Those are genuine resilience factors. They reduce the probability of a uniform global recession and should prevent a risk process from treating the worst case as the base case.

A second counterargument is that AI investment could produce broader productivity gains than the IMF assumes. Faster adoption would improve the upside scenario. Yet investment spending and productivity are not interchangeable: capital expenditure can rise before productivity appears, and high valuations can correct if expected profits do not materialize. The July forecast explicitly does not assume an external productivity boost. A company should therefore separate contracted AI-related demand from speculative valuation effects and from longer-term productivity potential.

A third counterargument is that “risks are more balanced than in April” conflicts with warnings about war, inflation and market correction. It does not. A risk balance compares the distribution of upside and downside around a dated baseline. It can become more balanced even while uncertainty remains high. The contradiction only appears if “more balanced” is misread as “low risk.”

Scenarios

Base case: Hormuz reopens gradually, energy prices remain elevated but manageable, the AI investment cycle moderates without collapsing, and financial conditions remain supportive by historical standards. Companies should plan for uneven demand, selective margin pressure and continued divergence rather than a synchronized downturn.

Adverse energy case: renewed escalation delays normalization, commodity volatility returns, inflation expectations rise and financial conditions tighten. The first management questions are liquidity duration, energy pass-through, supplier concentration, inventory policy and customer affordability. Government support may soften the immediate impact while increasing later fiscal constraints.

Technology correction case: investors reassess AI profitability, capital expenditure slows and highly valued technology assets reprice. Suppliers with order concentration or speculative capacity additions face a sharper downside than diversified users of AI. The risk is not “AI disappears”; it is a timing and cash-flow mismatch between investment and realized returns.

Upside adaptation case: trade routes normalize faster, AI adoption broadens, productivity improves and inflation pressure fades. The winners are not automatically the firms with the loudest AI narrative. They are those able to convert technology into measurable throughput, lower unit costs, better decisions or new revenue while maintaining capital discipline.

Uncertainties

The freshness boundary is 17 July 2026. The IMF forecast uses assumptions and market information available before publication; commodity prices and conflict conditions can move faster than the forecast cycle. The evidence does not establish company-level exposure, the durability of the AI investment boom, the full fiscal cost of energy support or the speed of trade normalization. It also does not prove that aggregate technology investment will translate into broad productivity growth.

These unknowns are decision-specific. Missing route-level logistics data blocks a supply-chain conclusion. Missing debt maturities block a financing conclusion. Missing customer segmentation blocks a demand conclusion. “More research is needed” is not an actionable uncertainty statement; each gap must name the decision it prevents and the data required to close it.

Decision framework

Boards and operators should maintain a monthly exposure table with five columns: variable, current observation, threshold, business transmission and owner. At minimum, track energy cost per unit, supplier and route concentration, AI-cycle revenue share, tariff-sensitive gross margin, debt repricing schedule, customer affordability indicators and scenario liquidity. Every threshold should trigger a predefined action rather than another meeting.

Capital allocation should distinguish resilience spending from directional bets. Hedging, supplier diversification and liquidity buffers protect the downside. Capacity expansion and AI-linked investment pursue upside. Combining both in one undifferentiated “strategic investment” budget hides the risk-return decision. The base case should be approved alongside the conditions that invalidate it.

The central conclusion is disciplined rather than dramatic: global growth has not collapsed, but averages have become less useful. The decision advantage lies in mapping how the same macro forces transmit differently through a particular balance sheet, supply chain and customer base.

Sources