AI-enabling goods powered almost half of first-half trade growth, but an energy shock and weak European goods exports reveal how uneven the global recovery has become.
The most important number in the World Trade Organization’s latest forecast is not merely the new projection of 3.9% growth in merchandise trade this year. It is the concentration hidden behind that surprisingly strong total. Almost half of the expansion in world goods trade during the first six months of 2026 came from products that enable artificial intelligence, according to the organisation’s October 8 assessment. The result is a world economy that looks markedly more resilient in aggregate than it feels in many factories, airports and energy-importing households.
There is a genuine recovery in cross-border shipments, and it should not be dismissed as a statistical illusion. But the sources of that recovery are unusually narrow. East and Southeast Asian manufacturing networks are supplying the chips, servers and related equipment required for an extraordinary wave of data-centre investment. At the same time, war in the Middle East has disrupted oil and gas exports, raised transport costs and undermined travel. A single global trade forecast therefore conceals two different stories: industrial acceleration around AI and a painful adjustment across energy-dependent services and regions.
The Geneva-based WTO now expects the volume of merchandise traded worldwide to increase by 3.9% in 2026, up from its March projection of 1.9%. It expects 4.1% growth in 2027, compared with a previous 2.6% estimate. These are changes in projected physical trade volumes, not revenue growth, stock-market returns or global output. The distinction matters, because unusually expensive semiconductors and volatile oil prices can push the dollar value of trade in directions different from the quantity of goods actually moving through the system. The official October Global Trade Outlook and Statistics, together with reports from Reuters and Associated Press, provides the basis for this assessment.
A forecast revision large enough to change the debate
Forecasts move routinely as new shipping, production and consumption data become available. Doubling an estimate in a matter of months is less routine. The WTO began the year anticipating that the powerful expansion in electronics exports during 2024 and 2025 would lose momentum, while higher energy costs and the disruption of shipping routes through the Persian Gulf would restrain trade. Both assumptions contained plausible economic logic. Investment booms eventually mature, and prolonged conflict normally makes international commerce more expensive.
By October, however, the first half of the year had delivered evidence against a simple slowdown narrative. World merchandise trade volume expanded by 3.5% in the first six months of 2026. Producers and buyers sourced some disrupted commodities from alternative suppliers, merchants redirected shipping through other corridors, and demand for AI equipment continued accelerating. The WTO’s revision recognises these developments rather than declaring that the geopolitical costs have disappeared. It changes the central estimate while leaving substantial uncertainty around the distribution of gains and the durability of investment.
The March forecasts remain important as a baseline. The WTO’s March analysis explicitly posed the durability of AI investment and the persistence of high oil prices as competing forces for 2026. That framework has not become obsolete. Instead, one force proved stronger than originally expected, while the other inflicted major damage on particular routes, countries and service industries without stopping aggregate goods trade. A forecast is a conditional view of available evidence, not a guarantee of future demand.
Why an artificial-intelligence boom produces physical cargo
AI is often described as a software revolution, but building the systems behind it requires a substantial international flow of manufactured goods. High-performance processors, memory devices, networking equipment, power-management components and finished servers are assembled through production chains that span multiple borders. Cooling systems, electrical equipment and other data-centre hardware add a second layer of industrial demand. A new model or software application may be distributed digitally, yet the computing capacity needed to train or serve it rests on an extensive physical supply chain.
This is why investment decisions made by cloud companies and data-centre operators can appear in customs data thousands of kilometres away. A design created in one country may require fabrication, packaging, testing, integration and final installation in several others. At each stage, goods cross borders and generate activity for suppliers, logistics providers and industrial service companies. An investor looking only at the final buyer may miss the manufacturing ecosystem capturing the earliest orders. The trade effect is real, although the multiple crossings of components mean customs flows should never be confused with the net value added by any single economy.
The WTO says trade in AI-enabling products increased by 67% year on year in the first half of 2026, after already expanding strongly in earlier years. Such products accounted for 47% of the increase in global merchandise trade during that period. The second percentage describes a contribution to growth, not a 47% share of all trade. It does not mean nearly half of all containers carry AI equipment or that 47% of global production is related to machine learning. Preserving that distinction prevents a powerful statistic from becoming a misleading description of the entire economy.
The contribution statistic and its limits
A high contribution to growth can arise when a relatively concentrated category expands exceptionally fast while other categories grow more slowly. AI hardware is doing exactly that. Equipment orders can be large and expensive, and the relevant supply chain contains repeated movements of intermediate goods. Meanwhile, shipments of some traditional manufactured products remain sluggish. The result is a large influence on the year-on-year change in trade even though ordinary consumer goods, agricultural commodities and industrial materials still form an enormous part of international commerce.
There is another important measurement issue. The 67% figure refers to trade in AI-enabling goods as reported in the WTO analysis, while the headline forecast concerns merchandise trade volume. Dollar-denominated data are sensitive to changes in product prices and composition, whereas a volume measure attempts to remove price effects. Advanced chips may become more capable and more expensive even when the number of physical units does not rise proportionately. Readers should therefore avoid converting the reported increase in AI-related trade value directly into an estimate of global real output or productivity growth.
Strong sales of hardware also say less than some investment narratives imply about the financial returns ultimately earned by AI applications. Equipment purchases register immediately in trade statistics. Revenue generated from new computing capacity may arrive later, and commercial adoption can be uneven. If customers eventually reduce spending or operators discover they have built excess capacity, the trade stimulus could fade before the full economic benefit becomes clear. That is a risk, not a conclusion that the current wave is inherently unsustainable.
Asia moves to the centre of the global goods outlook
The regional forecast reveals the magnitude of the shift. The WTO expects Asian merchandise exports to grow by 9.9% in 2026 and Asian imports by 9.5%. Those figures are far stronger than the corresponding projections for Europe and reflect an ecosystem that both produces sophisticated goods and buys the intermediate components required to manufacture them. Regional trade is not simply Asia selling completed chips to American customers. It also includes substantial exchanges among Asian economies before finished equipment reaches its destination.
Taiwan is an instructive example of that network’s momentum. Figures released on October 8 showed its September exports reaching $87.22 billion, a monthly record and 60.9% above the level a year earlier, according to Reuters. The report identified AI-related demand and the United States as major drivers. The data are a separate national indicator, not a component that should simply be added to the WTO’s global forecast, but they provide a concrete example of the concentrated demand reflected in the broader trade outlook.
South Korea, Singapore and other parts of the regional electronics ecosystem play different roles in design, production, assembly, transport and specialised materials. The precise distribution of value added varies by product and changes over time. What matters for the macroeconomic picture is the density of firms and infrastructure needed to make extremely complex components at scale. Building such a network requires long-term industrial capabilities that cannot be recreated overnight in response to a change in tariffs, political rhetoric or a new data-centre project.
The advantage comes with a concentration risk. A disruption affecting a limited set of advanced manufacturing locations, suppliers of specialised inputs or important regional shipping routes could spread quickly through the sector. Strong demand does not remove the vulnerability created when a disproportionate amount of growth depends on a narrow technological ecosystem. For governments, the lesson is to recognise both the value of participation in that ecosystem and the limits of dependence on a single set of customers or manufacturing bottlenecks.
North America’s demand is reshaping the trade map
The WTO expects North American merchandise exports to rise by 5.7% in 2026, while its imports increase by a more modest 1.4%. The disparity should not be interpreted as proof that North American buyers are marginal to the AI boom. The region remains a major centre of demand for computing infrastructure and a source of investment decisions that trigger production throughout Asia. Trade projections reflect several product categories, price dynamics and comparisons with earlier periods, not only one high-growth sector.
Some of the comparison is affected by unusually strong imports in early 2025 ahead of tariff changes. That front-loading raised the base against which later imports are measured. The WTO’s July assessment of first-quarter trade drew attention to this effect and to the strong quarter-on-quarter movement in North American imports despite a weaker annual comparison. A year-on-year decline or modest rise can therefore coexist with healthy current demand in specific capital-goods categories.
This distinction matters for interpreting trade tensions. Tariffs can redirect sourcing, encourage stockpiling and change the origin of imports without necessarily reducing the underlying need for equipment. A semiconductor may move through a different production or assembly route because of policy, but the eventual customer still needs the component. Some costs rise in the transition, and the geographic winners change. The volume of global trade can nonetheless remain substantial if final demand is strong enough to support a more complicated network.
Europe’s goods weakness is not the whole European story
Europe sits on the other side of the forecast divergence. The WTO projects a 0.1% decline in European merchandise exports in 2026 and only 0.5% import growth. These are modest figures in contrast with the rapid Asian expansion and imply that Europe’s factories are not capturing the AI-led trade impulse on the same scale. That does not establish that every European industrial business is shrinking. It does indicate that the region’s aggregate volume performance is subdued while other parts of the world benefit from a highly specific investment cycle.
Several forces may be at work, and they must not be collapsed into one explanation. European manufacturers face competition in machinery, automobiles and other products, the cost of energy remains important for industry, and some of last year’s trade figures were affected by shifts in shipments ahead of expected policy changes. The current WTO projection records the combined outcome rather than measuring the contribution of each cause. Detailed national production and order data are needed to establish which effects dominate particular sectors.
Europe’s service economy presents a striking contrast. The WTO expects the region to achieve 4.6% services export growth in 2026, the fastest projected among the large regions it reports. It expects Europe to account for more than half of the increase in global services exports. The region can therefore have weak goods-export volumes and a comparatively strong performance in cross-border services at the same time. That is one reason a single manufacturing indicator is a poor substitute for a comprehensive assessment of Europe’s external sector.
Two kinds of trade are moving in different directions
The WTO revised its estimate for world commercial services trade volume growth in 2026 down to 3.3%, from 4.8% in March. It projects a recovery to 6.4% in 2027. The deterioration in the current year’s services outlook stands beside the dramatic upward revision in goods trade. Far from contradicting each other, the two projections show how different parts of international commerce respond to the same geopolitical shock.
Transport and travel are especially exposed to route closures, higher insurance charges, fuel costs and uncertainty. Aircraft cannot always substitute a modestly longer flight path without consequence. Shipping companies can redirect vessels, but those changes consume time, fuel and equipment, alter sailing schedules and may change the cost of supplying distant markets. Firms supplying digitally delivered services face different constraints. Their export growth depends more directly on connectivity, customer demand, labour and regulation, although energy and infrastructure still matter indirectly.
The WTO reported that spending by international travellers grew by only 5% year on year in the second quarter, compared with 15% in the first. International tourist arrivals declined by 0.8% in the second quarter, according to figures cited from the United Nations tourism agency. These observations are consistent with a disruption to travel, but they do not imply that every tourist destination experienced the same contraction. Routes and regions linked to the Middle East are more directly exposed than others.
Digitally delivered services offer an important counterweight. The WTO reported computer-services export growth of 18% in the first quarter and an estimated 12% in the second, both in annual value terms. Financial-services exports grew by 14% year on year in the second quarter. Those figures describe distinct service categories and periods. They should not be blended into a single estimate for total services activity. They do, however, suggest why economies with strong software, professional and financial sectors may fare better than those heavily dependent on physical travel flows.
How the energy shock was absorbed, but not erased
The Middle East conflict has dramatically changed trade in energy from the region. According to the WTO’s October figures, Middle Eastern crude-oil exports fell by roughly 24% in the first half of 2026, while exports of liquefied natural gas declined by 47%. Those are severe regional losses that affect producers, buyers and the logistics systems that connect them. The changes also show why concentrating on global totals alone can understate the economic dislocation caused by war.
The global declines were much smaller. Shipments from other suppliers limited the contraction in world crude-oil exports to around 6% and in LNG exports to about 1%, the WTO reported. This is the practical meaning of supply-chain adaptation: producers with available capacity, traders, shipping routes and importers adjust to replace some missing flows. The substitution can prevent a global shortage from being as large as the original regional disruption might suggest. It does not recreate lost revenue for affected exporters or guarantee that buyers pay the same price.
Indeed, a market can preserve most of its trade volume at significantly higher cost. Cargoes may travel farther, buyers may compete for limited flexible supplies and transport and insurance charges can rise. Those costs affect inflation and business margins even when the quantities delivered remain relatively stable. A family paying more for energy or a chemical producer facing higher feedstock costs will not necessarily experience the aggregate resilience described by a global volume statistic. Resilience means the system keeps operating; it does not mean the shock is painless.
Fertilizer markets expose another layer of dependence
Energy is only one of the affected commodity chains. Fertilizer production and transport are closely tied to natural gas availability, shipping routes and concentrated suppliers. The WTO says global imports of nitrogenous fertilizers were only 2.8% below recent averages despite severe regional disruption, while phosphatic fertilizer imports stood 2.2% above those averages. Alternative supply sources helped the market adjust, but the figures cannot answer whether all farmers obtained the required products at affordable prices and in time for local planting seasons.
Agricultural consequences emerge with a lag. Farmers can respond to expensive fertilizer by changing application rates, switching crops or reducing purchases, while distributors may draw on inventories built earlier. The effects on yields, food prices and farm income depend on weather, credit, subsidies and local market structure as well as international trade. Therefore a modest decline in worldwide fertilizer imports is reassuring about aggregate availability but is not enough evidence to declare every food-security risk resolved.
The policy relevance extends beyond the countries producing oil and gas. Import-dependent economies face a different challenge when fuel and fertilizer markets tighten simultaneously. Higher electricity and transport prices can raise the cost of producing and distributing food just as agricultural inputs become more expensive. Governments then confront trade-offs between supporting households, protecting fiscal positions and avoiding measures that unintentionally obstruct international supply. These pressures are among the reasons the WTO continues to stress the importance of functioning trade rules during a crisis.
Containers and corridors: adaptation has a measurable footprint
Another indicator of adjustment appears in port traffic. Global container throughput increased by 3.9% during the first seven months of 2026, the WTO reported. That is an important counterpoint to the assumption that more expensive or difficult routes necessarily translate into a global collapse in cargo movements. Shippers continue to meet demand by adjusting networks, vessel deployment and ports of call. The ability to reroute flows provides flexibility when individual passages become unsafe or unreliable.
Container throughput is not identical to the volume of merchandise traded, however. A container may be handled at multiple ports, and the pattern of transshipment can change when routes are redesigned. Higher handling totals can reflect changes in network complexity as well as greater final demand. This is why the WTO combines information from customs statistics, physical volumes and transport activity rather than treating a single measure as definitive. The broader picture is stronger than any one indicator, but every metric has limits.
For firms, disruption is felt through predictability as much as through average cost. Delays complicate inventory management, manufacturing schedules and working-capital requirements. Companies may hold more stock, diversify suppliers or pay for alternative transportation to protect production continuity. Such adaptation improves the odds of keeping goods flowing but may absorb cash and managerial attention that could otherwise fund productivity investment. The added expense is part of the hidden price of resilience.
Stronger trade is not the same as faster global growth
The WTO forecasts world GDP growth of 2.6% in 2026 and 2.9% in 2027. Those numbers are far less dramatic than the near doubling of its merchandise trade growth projection for the current year. There is no contradiction. Trade measures cross-border transactions, while GDP measures the value of final goods and services produced. An intermediate component can cross borders more than once; each movement may count in trade, whereas GDP accounting removes double counting in the calculation of value added.
The distinction also highlights the concentration problem. A powerful surge in shipments of electronic components can lift trade without generating equal gains in employment or household income across all countries. Manufacturers in specialised production networks may see a rapid expansion, while consumers elsewhere experience higher fuel bills and limited real-wage growth. A trade headline can therefore coexist with restrained domestic demand or political discontent in major economies.
For policymakers, the question is not whether trade growth is ‘real’; the published figures indicate that it is. The more useful question is how much of that growth is broad-based, how efficiently it supports final consumption and investment, and whether the productivity benefits diffuse beyond the firms building or operating AI infrastructure. Those questions require labour-market, investment and national-accounts data as well as shipping statistics.
The China-US relationship remains a separate source of friction
The WTO’s broader analysis points to an increasingly altered geography of trade between the United States and China. Tariffs, export controls, industrial subsidies and investment screening can change the direct bilateral relationship even when some products continue reaching their final users through third-country production networks. A reduction in bilateral imports does not by itself measure the total reduction in the economic relationship, because intermediate inputs, services and corporate supply chains create indirect links.
The economic consequences differ by sector. In advanced semiconductors, restrictions on technology and equipment can determine where new capacity is constructed. In everyday manufactured goods, firms may shift assembly in response to tariff schedules or sourcing risk. For exporters, preserving access to major markets often requires investment in compliance, documentation and additional production capacity. Such expenditure can make the supply chain more geographically diversified while also making it more costly.
The WTO’s upgraded forecast is therefore not a verdict that trade policy friction no longer matters. It shows that trade can expand while policy makes its routes and participants more complicated. The distinction will become more important if governments respond to the AI boom with further domestic-content requirements or new controls on critical technology. Future trade growth depends not only on the appetite for computing power but also on the rules that shape how that appetite can be served.
AI investment is an opportunity and a vulnerability
The WTO expects worldwide spending on AI infrastructure to rise by at least 30% in 2026. Current market projections cited in the report suggest additional growth in capital expenditure of 10% to 20% in 2027. These figures are forecasts about investment behaviour, not confirmed outcomes. They nevertheless explain why suppliers are expanding capacity and why electronics orders have become so influential in global trade data.
The risks of such concentration are familiar from previous investment cycles, although the technologies are different. If companies build computing capacity faster than paying customer demand develops, purchasers may delay replacement cycles or reduce new orders. If power availability, cooling, grid connections or planning permissions constrain data-centre construction, equipment deliveries can become misaligned with project schedules. If financing costs rise, investors may reassess expected returns. Each path could feed back quickly into exports from the specialised economies serving the sector.
There is also an upside scenario. If AI applications generate significant productivity improvements, today’s infrastructure spending could stimulate broader business investment and demand for services in subsequent years. More firms might use computing capacity to redesign manufacturing, logistics or customer operations. But the transition from hardware spending to widely distributed productivity gains is not automatic. It requires reliable products, organisational adaptation, skilled workers and business models that can support the costs of deployment.
A growth map with pronounced winners and losers
The regional projections underline the unevenness. After Asia’s forecast 9.9% goods-export growth, the WTO expects North America to grow by 5.7%, Africa by 5.6% and South America by 3.4%. Europe is expected to slip by 0.1%, while the Commonwealth of Independent States is projected to decline by 3.9%. The Middle East faces a much sharper 17.2% contraction in merchandise exports. These are WTO regional groupings and aggregate projections, not forecasts for every constituent country or industry.
Import projections tell a related story. Asia is expected to increase goods imports by 9.5% and Africa by 8.9%, while the CIS is projected to rise by 8.8%. Europe is forecast at 0.5% and North America at 1.4%. The Middle East is expected to experience a 15.4% fall in imports. Those changes reflect distinct economic environments, including investment intensity, production structure, financing conditions and direct exposure to conflict. They do not provide a single ranking of household welfare or economic policy success.
The services figures complicate any simple division between an accelerating Asia and a stagnant Europe. Europe’s projected 4.6% growth in service exports is higher than the WTO’s 4.0% projection for Asia, with Africa at 3.1%. The Middle East, by contrast, is expected to suffer a 10.3% contraction in services exports. Economies that compete in professional, financial and digital work have a different exposure from regions reliant on energy shipments or aviation connections. Those differences deserve equal prominence with the more spectacular chip-trade statistics.
Why the distribution of gains matters politically
Governments can celebrate growing trade only up to the point at which households ask whether they have benefited. A geographically concentrated expansion can raise employment and tax receipts in manufacturing clusters without offering similar opportunities to workers elsewhere. Export growth may also enrich owners of productive assets before it produces broad wage gains. The gap between a favourable macroeconomic number and everyday experience becomes politically consequential when food, transport and energy remain expensive.
This is not an argument against technological investment or openness to trade. It is a reminder that the benefits of participation depend on skills, infrastructure, financial access and domestic institutions. An economy that supplies specialised components or advanced services can capture more value than one limited to low-margin tasks. Building higher-value capabilities takes time, and policy must balance industrial ambition against the risk of expensive interventions that do not create competitive firms.
The WTO’s director-general has emphasised that not all countries have equal access to the opportunities created by AI. That observation gives the forecast a developmental dimension. Countries without reliable power, digital networks or the capacity to attract investment may see little direct benefit from the capital-intensive boom, yet still face import-cost increases caused by the energy shock. Trade policy can influence that distribution, but it cannot substitute for the domestic investments required to participate productively.
What businesses and investors should actually monitor
Three categories of evidence will help determine whether the current forecast becomes a durable trend. The first is AI capital expenditure: announced commitments matter less than the pace at which equipment is ordered, installed and used. The second is the energy and transport environment, particularly whether alternative suppliers and routes continue meeting demand without another step-change in costs. The third is the breadth of industrial orders beyond electronics, because diversification would make the global trade recovery less dependent on a single investment cycle.
For manufacturers, indicators such as order backlogs, lead times, customer concentration and inventories can reveal whether growth is translating into sustainable production. Shipping companies will watch route reliability, vessel utilisation and the cost of rerouting. Governments will monitor customs volumes alongside prices, wages and fiscal pressures. None of these signals should be interpreted alone: an increase in nominal exports caused by higher prices is different from an increase in real shipments, and a surge in chip orders does not automatically lift domestic consumption.
The next forecast revisions will also show whether the WTO’s expectation of 4.1% merchandise-trade growth in 2027 remains credible as the comparison with an exceptionally strong AI investment year becomes more demanding. Its projection of a services recovery to 6.4% depends on conditions improving in sectors that have suffered from war-related travel and transport disruption. If the conflict persists or investment spending moderates, those numbers could change again. Forecast uncertainty should be treated as part of the story, not buried in a closing disclaimer.
Final assessment: resilience with an unusually narrow engine
The October WTO revision is encouraging evidence of adaptability. Despite a severe shock to Middle Eastern energy shipments, global trade networks have found alternative suppliers and routes, while demand for AI infrastructure has supported an unexpectedly vigorous expansion in goods trade. Those are significant economic developments, not merely optimistic interpretations of limited data. The ability to keep physical commerce functioning during geopolitical stress has immediate value for businesses and households worldwide.
Yet the headline conceals a pronounced imbalance. Asian electronics supply chains and North American technology spending are helping lift world trade at the same time that Europe records weak goods exports, the Middle East suffers devastating reductions in merchandise shipments and transport and travel face mounting costs. Europe’s relatively robust service exports show that the world is not splitting cleanly into one winning and one losing region. Different industries in the same economy can inhabit sharply different cycles.
The most consequential conclusion is therefore conditional. If AI investment broadens into productive applications and energy supply chains continue adapting, the higher trade trajectory may prove durable and support more diversified growth. If AI buyers pull back before other sectors recover, or if another energy or shipping shock overwhelms the system’s spare capacity, the same concentration that made 2026 look surprisingly strong could make the next correction unusually abrupt. The WTO has upgraded the global outlook, but it has also illuminated how much of the recovery now rests on a narrow technological foundation.
