The First Half of 2026: How Iran and AI Reshaped Asia's Markets

By the close of trading on 30 June 2026, the Asian equity story of the year had already been written twice over: once by geopolitics centred on Iran, and once by capital expenditure on artificial intelligence. Nikkei Asia's first-half scorecard, published 30 June at 04:31 UTC, put the matter bluntly: "In global financial markets, two key words dominated the first half of 2026: Iran and AI." The two drivers did not move every index in the same direction, and the gap between the year's winners and its losers is now wide enough to read as a map of which Asian economies are tethered to which of those two currents.

The H1 2026 rotation is not a clean narrative of regional advance. It is a story of dispersion. Markets that benefited from elevated energy prices, defence procurement, or AI-linked supply chains outperformed. Markets caught in the path of a renewed Iran shock, or sitting outside the AI hardware and infrastructure complex, lagged. The Nikkei review sets the frame; the underlying driver data — the data-centre build-out that now exceeds US spending on airports, marine terminals and mass transit combined, per Unusual Whales' 30 June 02:58 UTC citation — suggests that the AI leg of this trade has the longer half-life.

The Iran leg: energy, defence, and the Strait

The geopolitical leg of the trade runs through the Strait of Hormuz and the Gulf. A renewed confrontation with Iran across the first half of 2026 — the specific flashpoints of which Nikkei identifies as the dominant macro input for the period — pushed Brent and Dubai benchmarks off their 2025 dais, even as US shale failed to fully offset the disruption the way it did in earlier cycles. Asian buyers felt it directly: South Korea and Japan, the two largest Asian importers of Middle Eastern crude, saw refining margins swing wildly between late February and the end of April. India's state refiners, locked into long-dated Russian barrels, were partly insulated; Singapore's trading desks were not.

Defence names across the region rode the same wave. South Korean primes — Hanwha Aerospace, Korea Aerospace Industries, Hyundai Rotem — outperformed the Kospi by double-digit percentage points through the first five months of the year, a move consistent with European NATO demand for artillery and armour continuing to draw on Korean production lines, alongside Gulf State restocking. Indian defence PSU stocks, including Hindustan Aeronautics and Bharat Dynamics, registered a smaller but still positive premium to the Nifty, reflecting New Delhi's posture as a net beneficiary of sustained Western demand for non-Russian platforms.

The counterpoint sits with the laggards. Nikkei's scorecard highlights markets that were either directly exposed to import-cost pressure or reliant on the Gulf as a downstream customer. Thailand's SET was the clearest case: tourism receipts from the Gulf fell, and a stronger dollar-denominated import bill compressed current-account mathematics. Indonesia's IDX, while supported by commodities, gave back ground in Q2 as Jakarta moved to ration subsidised fuel to keep its fiscal arithmetic inside the deficit ceiling.

The AI leg: capex, concentration, and the new bottleneck

The second driver is harder to see in any single index move because it is concentrated in the supply chain rather than the end market. The capital expenditure number is, on its own, almost absurd. According to figures surfaced by Unusual Whales on 30 June 2026 at 02:58 UTC, current US data-centre construction spending now exceeds what the country spends on airports, marine terminals, and mass transit systems combined. That is a single line of industrial activity outranking three traditional infrastructure categories together.

For Asia, the AI leg translates into a specific basket of names. TSMC, SK Hynix, Samsung Electronics, and the Tokyo Electron / Disco / Lasertec triangle of process-tool specialists have spent six months repricing toward a regime in which advanced-node wafer demand is treated as an inelastic order book. The HBM cycle — High Bandwidth Memory — has not rolled over the way DRAM bulls feared in late 2025; instead, Hynix in particular has retained share inside Nvidia's rack-scale roadmap at a margin profile that has surprised the sell side. Korean memory, in short, is again a price-setter rather than a price-taker.

Two structural caveats belong in the same paragraph. First, the AI leg of the trade is now narrow: a handful of design wins, three memory manufacturers, two foundries, and a circle of substrate and HBM-adjacent suppliers. The Nikkei snapshot does not name the laggards inside the AI complex, but the implication is that companies exposed to AI's downstream — software, integration, application-layer SaaS — have not seen the same multiple expansion. Second, AI capex is partly a function of the same geopolitical cycle driving the Iran leg. If Gulf sovereign wealth funds cool on US hyperscaler debt issuance in H2 because their fiscal posture is being reshaped by oil volatility, the financing tailwind for the AI build-out tightens. The two keywords are not independent variables; they are correlated.

What the scorecard leaves out

The midyear snapshot is a point-in-time ranking, and a point-in-time ranking cannot tell the reader where the second half is going. Three things the scorecard does not yet capture are worth flagging.

First, the dollar. A stronger dollar through the Iran spike did what a stronger dollar usually does — punished local-currency Asian debt and compressed USD-reporting earnings for the region's exporters. Asian central banks outside Japan spent foreign-exchange reserves defending their currencies through April; whether that defence is sustainable for another two quarters is a live question that the scorecard does not address.

Second, the China question. Beijing's industrial policy continues to compress margins in legacy semiconductors, batteries, and increasingly EVs. The scorecard aggregates indices that include a heavy Mainland weighting; the dispersion inside the CSI 300 between AI-exposed names and property-exposed names is wider than the index level suggests. Regional portfolios that aggregate Asia without disaggregating China risk understating the internal Chinese rotation.

Third, the AI capex number itself is a US figure, not an Asian one. The Asian read-through is through suppliers, not through end-market capex. If US hyperscaler capex pauses — for any reason, including the Iran-driven macro shock raising the cost of capital — the Asian supplier leg of the trade is the more exposed side. The relative narrowness of the AI trade inside Asia is the structural risk that the scorecard cannot price.

Stakes for the second half

The two keywords are unlikely to disappear in July through December. The Iran file is a slow-moving security question that has now survived multiple US administrations' attempts to either contain or roll it back; the structural drivers — sanctions architecture, Israeli-Iranian shadow confrontation, Gulf monetisation of US security guarantees — outlast any single negotiating cycle. AI capex, for its part, has its own inertial momentum: rack-scale system designs commissioned in 2025 are still being delivered, and the contracted order book visible to TSMC and Hynix extends well into 2027.

What changes, if anything changes, is the margin of safety. Markets priced for an Iran ceasefire rally in the last week of June; markets priced for AI capex continuation priced for it all year. If either tail risk materialises, the dispersion that defines H1 2026 will look less like a rotation and more like the first stage of a re-rating.

For investors, the practical read is that the Asian allocation question for H2 is not whether to be long Asia — most benchmarks have already moved — but whether the overweight is in the right leg of the trade. The defence and energy basket tied to the Iran file is cyclical in a way the AI supplier basket is not, but the AI supplier basket is concentrated in a way the Iran basket is not. Holding both without acknowledging the difference in risk profile is the error the scorecard implicitly warns against.

This piece draws on Nikkei Asia's first-half 2026 markets scorecard and on data surfaced through Unusual Whales' 30 June 2026 coverage of US AI infrastructure spending. Monexus frames the period as a two-driver regime — Iran-driven energy and defence, AI-driven supplier capex — and treats the apparent independence of the two legs as the central question for the second half.