When the lights flicker

Brent crude briefly touched nearly $120 a barrel on 9 March before falling back to around $88 by 11 March, as traders swung from pricing a prolonged disruption to betting that the war might end sooner than feared and policymakers explored an emergency reserve release. Yet the operating picture around the Strait of Hormuz remains far less reassuring than the oil price suggests. UKMTO and JMIC advisories continue to report severe GNSS and GPS spoofing, AIS anomalies, electronic interference and sharply reduced traffic through the waterway. The gap between markets and operations is widening: oil has retreated, but ships are still being struck, traffic remains heavily degraded, and one of the most dangerous escalation risks, the laying of mines in or near the Strait, still hangs over every transit.

Most commentary has focused on the obvious consequences: petrol prices, inflation and the political pain that follows both. The more important question sits further down the chain. This is not simply a temporary energy shock. It exposes how deeply the digital economy still depends on gas flows, specialist materials, export terminals, grid connections, insurance markets, shipping corridors and the infrastructure that holds them together. Once those systems come under stress at the same time, the line between an energy crisis and a technology crisis starts to disappear.

What comes into view very quickly is a question of control: over the gas that keeps fabs running, the specialist materials inside the chips, the terminals and processing assets that feed them, and the maritime routes that connect production to market. None of those dependencies is new. What the war has done is strip away the illusion that they can still be treated as background conditions rather than strategic constraints. The digital economy likes to imagine itself as software, scale and abstraction. In reality it still runs on molecules, minerals, substations, ports and ships.

The fabrication countries

Taiwan and South Korea remain central to advanced semiconductor production, but both economies sit on top of energy systems that are far more exposed than many investors like to admit. South Korea still buys around 70 per cent of its oil and 20 per cent of its LNG from the Middle East. Taiwan generates more than 40 per cent of its electricity from LNG and, in 2025, imported roughly one-third of that LNG from Qatar, with Australia supplying roughly another third. US LNG accounted for about 10 per cent in 2025, though Taipei has said it wants that share to rise to roughly 30 to 33 per cent in 2026.

That leaves Taiwan more exposed on gas than many investors assume. Crude sourcing has shifted heavily towards the United States, but gas remains the sharper vulnerability because it underpins a large share of the power system and still arrives by sea. Taipei has said petroleum and LNG inventories remain above legal minimums, while local reporting has put gas reserves only a little above the 11-day legal threshold. That is not a crisis in itself. It is enough to show how quickly a maritime disruption can become a power security issue.

Seoul can still chase alternative cargoes, and Taipei is doing the same, but emergency substitution only buys time. Taiwan has said it would buy more LNG from the United States and could coordinate with Japan and South Korea if disruption persisted. That helps at the margin. It does not change the underlying reality that once disruption runs from weeks into months, contingency planning turns into a harder contest for replacement supply, industrial prioritisation and resilience under pressure.

Markets did what they usually do in the opening phase of a shock: they priced fear first and asked harder questions later. Some of that panic unwound as oil retreated. The reversal says more about the speed with which sentiment swings than it does about any real improvement in the operating picture. Brent fell sharply even as conditions around Hormuz remained degraded and further vessels reported damage.

For the strongest parts of the semiconductor industry, whether memory or leading-edge logic, direct energy costs alone are unlikely to be decisive. SK Hynix's fourth-quarter 2025 revenue of 32.8 trillion won, and 58 per cent operating margin make the point clearly enough in memory. At the logic end of the stack, the bigger risk lies less in the electricity bill than in concentrated materials, imported energy and the infrastructure needed to keep fabrication running.

The sharper risk sits below the electricity bill, where specialist inputs and supply concentration matter more than most headline oil models allow for. South Korea has already flagged risks to key semiconductor materials, including helium and bromine sourced from the Middle East. The same problem is now emerging in sulphur and sulphuric acid. Gulf disruption has pushed sulphur prices sharply higher, tightening a supply chain that matters not just for fertilisers and metals but for semiconductor wafer cleaning as well. That is what a real systems shock looks like: not one commodity moving in isolation, but multiple industrial inputs tightening at once.

The Gulf’s AI bet

Saudi Arabia and the UAE have spent years and tens of billions positioning themselves as global AI and data centre hubs. The logic was sound: abundant power, geographic centrality between Asian and European markets, and sovereign wealth funds willing to take long-duration bets. Public disclosures show over $50 billion of AI and data centre infrastrucutre investment across the two countries. Saudi Arabia alone has $26.8 billion of named projects. Microsoft’s UAE commitment runs to $15.2 billion through 2029. TAQA Group is putting over AED 37 billion into power infrastructure for Abu Dhabi’s data centres.

These are assets designed to operate for twenty or thirty years. The return profiles assume stable operations over that horizon. AWS facilities in the UAE and Bahrain have now taken physical damage from drone strikes. Insurance premiums had already hit six-year highs before the war. QatarEnergy’s North Field expansion, which was supposed to take LNG capacity from 77 to 110 million tonnes per annum by 2027, was already running behind schedule. If this disruption goes past a month, the LNG surplus the market had priced in for this year disappears, and EU gas storage is at 30 per cent heading into summer refill season. Anyone writing cheques for Gulf data centre projects this week is doing very different maths from the maths they were doing three weeks ago. The cost of physical hardening, redundant power, and war risk insurance is now a line item that did not exist in the original pro formas. For some projects, that changes the viability calculation entirely.

Gas turbines and compute

For years, the cloud was sold as though it floated above industrial reality. The AI build-out now taking shape looks much more like the construction of large, power-hungry industrial facilities whose economics depend on secure energy, transmission access, generation capacity and the speed with which new power can be brought online. PJM has already been fast-tracking combined data-centre and power-generation projects because demand is surging and the grid cannot always move quickly enough to meet it.

That ought to end the fiction that compute floats free of energy reality. Operators reach for co-located power and behind-the-meter generation because speed matters and grid queues are too slow, yet that solution binds more of the compute stack directly to energy-market volatility and to the infrastructure needed to keep that energy flowing. The oil reversal changes none of that. It merely shows how quickly markets recover once traders convince themselves that enough barrels will still move. The more important question is whether the physical system underneath that confidence is genuinely secure.

Hyperscalers at the top of the market can absorb more volatility than everyone below them. Mid-tier data-centre operators, enterprise builders and regional platforms have much less room. A world of unstable energy inputs, tighter financing conditions and greater infrastructure risk therefore favours scale, cash flow and vertical control. Even when the war premium comes off the oil price, the financing logic underneath the build-out can still shift in favour of the players with the strongest balance sheets.

Taiwan is the larger warning

Hormuz should also be read as a warning about Taiwan, because a blockade of the Taiwan Strait would not simply interrupt the export flow of semiconductors. It would hit the most concentrated part of the global advanced compute stack, with Taiwan accounting for roughly 90 per cent of the world’s most advanced leading-edge logic below 5nm while also hosting the CoWoS packaging capacity used in AI GPUs that has become one of the supply chain’s key bottlenecks. A blockade would therefore place immediate pressure not only on advanced chip output, but on the imported energy, LNG receiving system, shipping access and power security that keep Taiwan’s fabrication base running.

The Taiwan parallel matters because a crisis in the Gulf constrains energy and specialist inputs first, then spreads up the chain into semiconductors and data centres, whereas a crisis in the Taiwan Strait would hit the advanced compute stack and the energy system underneath it at the same time. Taiwan generated more than 40 per cent of its electricity from LNG in 2025, imported about one-third of its LNG from Qatar and another third from Australia, and is only now trying to lift US LNG to roughly one-third of supply in 2026. Gas reserves sitting only a little above the legal minimum make the point sharper.

The economic consequences of a real blockade would not stop at semiconductors. Shipping insurance, rerouting costs, industrial production, electronics supply chains, cloud capacity planning and global confidence would all take a hit at once. That is why Taiwan matters here. Iran shows how quickly one maritime chokepoint can reprice energy, insurance and logistics. Taiwan would apply the same logic to the world’s most concentrated advanced compute ecosystem. CSIS has argued that Taiwan’s dependence on trade and imported energy makes blockade or quarantine scenarios especially severe, with consequences extending well beyond Taiwan itself.

There is also a precedent problem here. Unilateral US action in Venezuela and now Iran does not determine Chinese strategy on Taiwan, which is driven primarily by Beijing’s own objectives and timing. But it does weaken Washington’s ability to argue from principle when it opposes coercive attempts to shape the government of another polity. Reuters reported in January that analysts saw the Venezuela strike as something that could embolden China’s territorial claims and give Beijing more rhetorical ammunition on Taiwan, even if it did not make an immediate attack more likely. The more the United States normalises force, sanctions and economic strangulation in pursuit of regime change, the easier it becomes for Beijing to present pressure on Taiwan not as an aberration, but as another version of great-power coercion in a world that has already accepted the logic elsewhere.

China's calculation

China enters this period with more room than most Asian economies because it bought roughly 1.38 million barrels a day of Iranian oil last year, or about 13 per cent of its seaborne imports, while also building strategic inventories estimated at around 900 million barrels, or roughly three months of imports. Beijing’s new five-year energy plan also calls for more stockpiling and continued work on the Power of Siberia 2 pipeline. Immunity would be too strong a word. Buffer is not.

That difference gives China more room than Japan, South Korea or Taiwan to absorb a short disruption, but the more important point is that Beijing’s leverage no longer stops at rare earths. China’s power sits not just in mining but in processing, conversion, export licensing, manufacturing concentration and the ability to leave supply legally available while making it operationally slower, scarcer or more expensive. It has expanded rare-earth export controls, tightened restrictions around related equipment and widened its controls across strategic minerals.

The leverage, therefore, extends far beyond one narrow slice of the supply chain. China continues to dominate strategic processing capacity, and its wider industrial strength remains visible across batteries, graphite and other critical-mineral value chains. In strategic terms, that means Washington can still wield sanctions, finance, insurance, naval reach and the security architecture of seaborne trade, but Beijing can answer through industrial concentration, refining, export permits, stockpiles and overland energy links.

That is why the Iran war matters beyond the Gulf. Every time Washington uses maritime power, sanctions or financial pressure to shape commodity flows, Beijing gets another reason to deepen control over processing, logistics and overland alternatives. Every time Beijing tightens export licensing or controls, Washington gets another reason to derisk away from China. The two systems are no longer separating neatly into energy on one side and technology on the other. They are converging into a single contest over who controls the critical links between raw material, energy, compute and distribution.

What breaks if this lasts

Higher energy costs hit first, but fertiliser, freight and industrial inputs follow close behind once disruption spreads through shipping and infrastructure. The chain reaction is already moving beyond fuel. Sulphur prices have jumped, fertiliser markets are tightening, diesel has become a growing constraint on transport and manufacturing, and the same shipping disruption is now feeding into chemicals and metals used much further up the industrial stack. When energy, transport, petrochemicals and fertiliser all start moving together, the inflation channel is broader than the oil screen on a terminal suggests.

The technology sector does not escape any of that, because it still depends on the same stressed physical economy as everything else. When pressure builds across energy, transport, industrial inputs and financing conditions at the same time, operating costs rise, demand weakens, capital becomes harder to secure and political tolerance for expensive technological disruption narrows. Markets repriced rate expectations almost as soon as oil spiked, even if some of that move later softened. Policymakers are now considering an unprecedented strategic reserve release to calm prices, while import-dependent economies have already started using fuel-stabilisation mechanisms and conservation measures to contain the fallout. That tells you how seriously governments take the risk of a wider chain reaction.

The larger danger for the technology sector therefore lies not simply in a higher cost base but in the collision between a more expensive, less predictable physical world and very large capital commitments that were justified in a more benign one. The bigger the build-out, the more exposed those assumptions become. If energy remains volatile, infrastructure security costs rise, and the supply chain for specialist materials becomes more political at the same time, then expected returns on AI infrastructure, fab expansion and electrification projects all start to look less secure than they did when capital was cheaper and shipping felt invisible.

The balance of probability

Although Brent’s retreat suggests traders believe flows can still be maintained, or restored quickly enough to avoid a lasting supply shock, conditions around Hormuz remain far more fragile than market pricing implies.

What this crisis has exposed is the narrowness of the digital economy’s physical base. Semiconductor production depends on imported energy and concentrated specialist inputs. Gulf AI campuses depend on infrastructure that can be struck. Data-centre expansion is pushing operators ever closer to hard energy reality. Taiwan remains the clearest example of how advanced compute and energy vulnerability can sit inside the same maritime theatre. China, meanwhile, is deepening both stockpile resilience and industrial leverage across the materials chain, not only through rare earths but through refining, licensing and the administrative power to decide what leaves the country, in what form and for whom.

None of these risks sits comfortably inside a single team or a single spreadsheet, which is precisely why so many organisations still struggle to think about them properly. Energy, materials, compute, logistics, maritime security and sovereign risk no longer belong in separate silos. They are converging into the same bottleneck. The board-level question is no longer whether energy security matters to technology strategy. It is whether companies are still pretending these risks can be managed independently when the real world is presenting them as one interconnected system.

Control over gas routes, specialist materials, processing capacity, fabrication, energy infrastructure and the shipping lanes that connect them will shape the technology landscape for years to come. The Iran war has made that visible in one chokepoint. Taiwan would make it visible in another. The industry built its future on the assumption that gas would flow, ships would sail, materials would arrive and physical systems would remain politically boring. That assumption is now being tested in public.