The headline numbers for the first half of 2025 appear reassuring at first glance: the US-dollar value of global merchandise trade rose about 6.0 per cent year-on-year in H1 2025, while trade volumes (the average of exports and imports) increased by roughly 4.9 per cent.
That stronger-than-expected performance forced major forecasters to revise their 2025 outlooks upward but the apparent strength hides a complex, uneven picture that should give policymakers and business leaders pause rather than comfort.
Two structural forces drove much of the H1 bounce. First, geopolitical and policy uncertainty, especially high-profile tariff announcements, prompted import front-loading in some large markets, most visibly in North America. Firms accelerated purchases early in the year to avoid higher levies, creating a temporary spike in import volumes that is now reversing in the later quarters as inventories are drawn down.
The second, and more consequential from a medium-term perspective, was a concentrated investment cycle in AI-related goods: semiconductors, servers, finished computing equipment, and telecommunications hardware. Approximately 100 AI-related items increased by more than 20 per cent in H1 2025 and accounted for nearly half of global trade growth during that period. Those two dynamics, tactical stockpiling and strategic digital investment, explain much of the H1 surprise.
Disaggregating the picture by sector and region sharpens the policy implications. Sectorally, office and telecom equipment, electronics, and AI-infrastructure-related goods posted the strongest gains, with gains broadly in the high teens to low twenties across classifications. Chemicals rose more modestly, at around 10 per cent, while other machinery recorded single-digit growth. Traditional labour-intensive manufactures such as apparel and clothing saw only limited increases of roughly 7.0 per cent.
In contrast, fuels and mining products declined by about 5.0 per cent in value terms, continuing a downward trend evident since 2024 amid softer commodity prices and weaker demand. The concentration of growth in a narrow set of high-value, high-technology categories means headline trade growth is far less broad-based than aggregate figures suggest.
Geography also matters. Asia led export growth by a wide margin, recording double-digit expansion in the first half of the year. This reflected both its central role as the supply base for AI-related components and its strengthening intraregional demand. South–South trade outperformed the global average as well, supported by diversified demand across Latin America, Africa and parts of Asia. North America’s first-half surge was atypical, driven largely by import front-loading rather than sustained cyclical strength. Europe remained comparatively flat, constrained by weaker investment cycles and subdued domestic demand. Overall, growth in the first half of 2025 was geographically skewed toward Asia and emerging markets rather than signalling a synchronised global recovery.
The second critical lesson concerns tariffs and timing. Higher and broader tariffs announced during 2024–2025 did not halt trade growth in early 2025; indeed, they indirectly boosted it by encouraging front-loading. However, this is a timing effect rather than a durable offset to protectionism. WTO economists caution that the full drag from elevated tariffs is likely to emerge in 2026 as inventories unwind and trade adjusts to higher prices and policy constraints.
Reflecting this sequencing, the WTO revised its 2025 merchandise trade volume forecast upward to 2.4 per cent but sharply downgraded the 2026 outlook to just 0.5 per cent. That shift matters because it transforms a short-term rebound into weaker underlying momentum.
For firms and governments alike, the implications are clear. Companies expanding supply chains or capacity to capture the AI wave must manage inventory normalisation without eroding margins while ensuring long-term demand justifies capital outlays. Governments that read early-2025 resilience as validation of protectionism may find that tariffs function as a delayed tax on trade, masking their contractionary impact until it is too late.
Looking ahead, three medium-term trends should guide forecasting and strategy. First, the AI investment wave is unlikely to fade entirely. The shift toward cloud services, edge computing, and large language model deployments implies a multi-year upgrade cycle for digital infrastructure. WTO estimates imply AI-related trade could remain a structural growth engine even after the H1 spike abates: these goods now account for a meaningful share of trade and for a disproportionately large share of growth. For countries and firms able to climb the AI value chain through semiconductor manufacturing, advanced assembly, or data-centre hosting, there is a genuine export opportunity.
Second, energy and commodity dynamics move in opposite directions. Fuels and mining products are likely to remain under pressure in the near term, both from weaker prices and from longer-term structural shifts. Commodities producers should therefore expect continued volatility; policymakers must hedge budgets and avoid overreliance on short-term commodity windfalls. Industrial policy in commodity-dependent economies should prioritise diversification and added-value processing to capture more stable margins.
Third, protectionist tit-for-tat and the use of tariffs as strategic instruments risk creating bifurcated supply chains. Early signs already show firms re-routing supply lines, investing in friend-shoring and accelerating regionalisation. While diversification can be positive, it raises costs, complicates logistics and fragments global production networks. Countries that maintain open, predictable trade regimes and invest in trade facilitation will be better placed to attract investment seeking scale and predictability. Conversely, frequent policy whipsaws will deter capital-intensive projects, especially in semiconductors and other high-capex sectors.
For developing economies, H1’s pattern is a mixed blessing. On the one hand, South–South trade and demand for AI hardware in emerging markets present visible opportunities for export growth; on the other, competition for high-tech value-chain roles is fierce and dominated by established manufacturing hubs in East Asia. To capture more of the AI dividend, developing countries need targeted industrial policies: skills development for advanced manufacturing, investment incentives for data-centre clusters and regulatory frameworks that protect privacy and enable cross-border data flows under trusted arrangements. Without those elements, they risk remaining low-value nodes even while global trade in AI hardware expands.
What about forecasting numbers beyond the WTO’s cautious 2026 outlook? A reasonable scenario planning approach suggests three plausible paths. In a downside scenario, wider tariff escalation and weakening global GDP merchandise trade volumes could track below one per cent growth in 2026 and risk mild contraction in parts of 2027. In a moderate baseline, tariffs remain at current levels, inventories normalise, but AI investment continues at a slower pace, the WTO’s 0.5 per cent for 2026, followed by a modest rebound in 2027, is plausible.
In an upside scenario, sustained global cooperation on trade facilitation, continued digital investment and stable macroeconomic conditions could lift trade volumes to low-to-mid single-digit growth by 2027, led by digital and capital goods. However, because first-half growth was concentrated in a few product lines, downside risks are asymmetric: modest policy or demand shocks could erase gains faster than steady growth can restore them.
Policy priorities follow directly. Major economies should avoid mistaking front-loaded import surges for durable demand and align fiscal and industrial support with productive investment rather than temporary consumption. In trade policy, multilateral coordination to limit escalation and manage strategic goods is essential. For developing economies, targeted industrial policy, public investment and partnerships to integrate into AI-hardware value chains are key to converting today’s spike into a stable export base.
Finally, business leaders must adopt operational realism. Inventory normalisation will pressure margins and cash flows in some sectors; firms should prioritise working capital management and hedging strategies. Long-term capital allocation should favour projects with defensible competitive advantages, proximity to demand centres, specialised manufacturing capability, or lower-cost renewable energy for power-hungry data centres. And because trade growth in 2025 is uneven, corporate strategies should be regionally nuanced: Asia-centric manufacturing and supply lines for AI hardware; localised distribution and services for consumer goods; and contingency plans for energy-linked supply disruptions.
The first half of 2025 provided a valuable diagnostic: global trade can still surprise on the upside, particularly when structural investment cycles (AI) and tactical behaviour (front-loading) align. But that surprise is fragile. The underlying engines of broad-based trade growth remain weak, tariff risks are real and delayed, and sectoral divergence is stark. Policymakers should avoid complacency; firms should plan for a two-speed world where digital capital goods drive value even as many traditional sectors tread water.
The smart response is simple in principle – invest in competitiveness, preserve open and predictable trade rules and build resilience – but hard in practice. If H1 taught us anything, it is that volatility can create opportunity.
The writer is a trade facilitation expert, working with the federal government of Pakistan.
