The Iran conflict is threatening chip supplies, submarine cables, and energy costs simultaneously. For an industry building agentic commerce on the assumption of abundant AI infrastructure, the timing could not be worse.

On March 5, South Korea’s KOSPI index crashed. The immediate cause was surging oil prices following Iran’s threats to block the Strait of Hormuz. But as Australian analyst Shanaka Anslem Perera argued on X, the market crash was not a Korea story. It was the first live stress test of the AI infrastructure buildout’s most critical single point of failure.

Global DRAM inventory sits at two to three weeks. NAND at three to four weeks. There is no buffer. The $440 billion global memory supercycle, projected for 2026, depends on fabrication plants that cannot operate without imported oil and LNG flowing through contested waters. And right now, those waters are a war zone.

Every agentic commerce protocol announced this year, every AI fraud model deployed, every real-time payment system running on cloud infrastructure, depends on chips, energy, and data connectivity that the Iran conflict is placing at risk. The payments industry has spent the past six months debating trust layers, agent verification standards, and checkout protocols. Almost nobody is discussing what happens if the physical infrastructure underneath all of it becomes constrained.

The Strait of Hormuz handles approximately one fifth of the world’s oil and more than a fifth of its liquefied natural gas. When tensions escalated following US and Israeli strikes on Iran beginning February 28, traffic through the strait dropped by roughly 70 percent almost immediately. Major carriers suspended transits. Crude prices surged around eight percent after the first strikes. Spot prices for LNG leapt by up to 50 percent following the closure of Qatar’s Ras Laffan megaterminal.

For the semiconductor industry, energy is not a line item. It is the foundational input. Chip fabrication plants are among the most energy-intensive manufacturing facilities on Earth, operating massive cleanrooms and advanced lithography systems around the clock. When energy prices rise, running those factories becomes materially more expensive. When supply becomes uncertain, production planning becomes impossible.

South Korea is at the centre of this vulnerability. The country produces approximately two thirds of the world’s memory chips through Samsung Electronics and SK Hynix. It imports 97 percent of its energy through the Strait of Hormuz. As Perera noted, every NVIDIA Blackwell chip, every Google TPU, and every hyperscaler expansion relies on high-bandwidth memory manufactured overwhelmingly in one country that is entirely dependent on energy flowing through one contested waterway.IDC says that if the conflict lasts only a few weeks, the impact on the electronics industry will be modest but visible. Inflation across key areas is already present. But if the conflict drags on, the analyst firm warns it could worsen component shortages and reignite supply chain instabilities not seen since 2021.

The energy disruption is the most immediate risk. The data infrastructure disruption may be the most consequential.

Rest of World reported that for the first time, both the Strait of Hormuz and the Red Sea are effectively closed to commercial traffic. Seventeen submarine cables pass through the Red Sea alone, carrying the majority of data traffic between Europe, Asia, and Africa.

Amazon, Microsoft, and Google spent years building data centres across the Gulf, betting that the region would become the world’s next great hub for artificial intelligence. The January 2026 Pax Silica initiative brought the UAE and Qatar into a US-led effort to keep advanced semiconductors away from China. Abu Dhabi’s G42 cut ties with Huawei. Humain aligned with US chip suppliers. Billions of dollars in technology infrastructure were deployed on the assumption of regional stability.

That assumption is being tested. The undersea cables connecting Gulf data centres to Africa, South Asia, and Southeast Asia pass through two narrow passages, both of which are now disrupted. In February 2024, three Red Sea cables were cut by the dragging anchor of a cargo ship struck by a Houthi missile, disrupting 25 percent of traffic between Asia, Europe, and the Middle East. One cable took five months to repair because repair vessels could not safely access the area. If multiple major cables were severed now, with crews locked out of both chokepoints, the disruption could last far longer.

As Rest of World noted, US policy was designed to protect chips from Beijing, not infrastructure from Tehran. The security frameworks underpinning the US-UAE AI partnership focused on supply chain control and geopolitical alignment, not on physical defence during high-intensity conflict. The gap between digital ambition and physical vulnerability has never been more exposed.

Beyond energy and data, the Iran conflict threatens specific materials that the semiconductor industry cannot do without.

Qatar is a major global supplier of helium, a gas essential for heat management during chip fabrication. Helium has no viable alternatives in this application. It is produced in only a handful of countries, and Qatar is among the largest players. South Korea’s trade ministry warned this week that semiconductor production could be disrupted if key materials cannot be sourced from the Middle East.

Petrochemical feedstock prices are rising as crude surges, lifting input costs across the plastics sector, including components used in circuit board laminates, insulators, and semiconductor packaging. Electronics Weekly reports that market prices for these materials had been softening before the conflict. That trend has reversed.

The rare earth dimension adds a further layer of complexity. Iran sits on significant rare earth element deposits, and negotiations between Tehran and Washington reportedly discussed minerals just before the strikes began. China, as one of Iran’s few allies, expected to be the lead partner in developing those deposits. With Presidents Trump and Xi set to meet in the coming weeks, the intersection of rare earth access, semiconductor supply, and geopolitical positioning creates a tangle of competing interests that extends well beyond the immediate military conflict.

A less visible but potentially significant risk is the diversion of semiconductor manufacturing capacity toward military needs.

The speed at which the US and its allies are exhausting ordnance has raised concerns about component supplies. Electronics Weekly reports that missile interceptors used to counter Iranian drone attacks require radiation-hardened, high-reliability memory and dedicated FPGAs and processors for guidance systems. Capacity for these defence-grade components is already sold out at specialist fabrication plants.

US Secretary of State Marco Rubio highlighted the scale of the problem, noting that Iran produces over 100 missiles per month compared to six or seven interceptors that can currently be built. The administration could invoke the Defence Production Act to prioritise domestic production at facilities including Intel Foundry and Micron, and leverage Chips and Science Act subsidies over Samsung, SK Hynix, and TSMC for additional capacity.

If defence production is prioritised, commercial semiconductor availability gets squeezed further, on top of existing allocation battles driven by the AI data centre buildout. New Electronics warned that for an industry still recovering from the chip shortages of 2021 to 2023, the timing could hardly be worse. Chipmakers are already facing bottlenecks from AI demand that has tightened supplies to manufacturers of smartphones, laptops, and automobiles. A defence diversion would add a third source of demand pressure on already constrained capacity.

This is where the geopolitical story connects directly to our industry.

Every agentic commerce protocol being built right now assumes abundant, affordable AI compute. Mastercard’s Verifiable Intent uses cryptographic proof layers that require processing power at scale. Google’s Universal Commerce Protocol assumes AI agents can reason over product catalogues, compare options, and execute transactions in real time. OpenAI’s Agentic Commerce Protocol assumes 700 million weekly ChatGPT users can seamlessly transition from discovery to checkout through AI-powered interfaces.

The AI fraud detection systems that Visa, Mastercard, and every major bank depend on require continuous model training and inference at enormous scale. Real-time payment systems like FedNow, which has crossed 1,500 participating institutions, run on cloud infrastructure that consumes energy and chips in quantities that are growing, not shrinking.

If memory production slips because Korean fabs cannot secure affordable energy, AI model training costs rise. If data centre expansion in the Gulf stalls because submarine cables are at risk, latency increases and redundancy decreases. If defence production diverts semiconductor capacity, commercial chip availability tightens further. Each of these outcomes pushes out the timeline for agentic commerce at scale.

The cascade effects are worth tracing. Higher chip costs mean higher cloud compute costs. Higher cloud costs mean higher operating expenses for every payment processor, fraud detection system, and AI-powered checkout platform that runs on cloud infrastructure. Those costs ultimately pass through to merchants as higher processing fees or to consumers as reduced investment in product innovation. The payments industry spent years compressing margins through technology efficiency. A sustained chip shortage or energy price spike reverses that trend.

Cross-border payments are particularly exposed. The SWIFT network, correspondent banking infrastructure, and real-time cross-border systems like Visa Direct and Mastercard’s Move platform rely on data centres and connectivity that span the exact geographies now at risk. The G20’s 2027 target for faster, cheaper cross-border payments was already under pressure. A prolonged Hormuz disruption would make those targets unreachable.

The stablecoin infrastructure being built on top of all of this is equally dependent. Fiserv’s FIUSD, PayPal’s PYUSD, and Circle’s USDC all require blockchain nodes, validation infrastructure, and exchange connectivity that run on the same cloud platforms and semiconductor supply chains now under stress. The GENIUS Act may have created the regulatory framework for stablecoins, but the physical infrastructure to support them at scale faces the same chokepoint risks as every other digital financial service.

Bloomberg Opinion framed it directly: a world where we can generate AI videos in seconds from our phones might seem remote from the physical realities of warfare in the Persian Gulf. In fact, they are closely intertwined. The building blocks of the technology industry are deeply dependent on petroleum flowing through the Strait of Hormuz.

The payments industry has spent the first quarter of 2026 debating which agentic commerce protocol will win, which trust layer will become the standard, and which AI models will power the next generation of fraud detection. Those debates assume the infrastructure to support them will be available when needed. That assumption is no longer safe.

Polish EMS group Tstronic captured the mood in a statement cited by Electronics Weekly: “Global political risk encourages abandoning strict just-in-time in favour of a just-in-case strategy.”

For payments companies that depend on cloud infrastructure, the implication is worth considering carefully. Just-in-time has been the default operating model for digital services: scale compute on demand, rely on global supply chains for hardware, and trust that capacity will be available when needed. The Iran conflict is a reminder that supply chains have physical dependencies, that those dependencies concentrate in vulnerable geographies, and that conflicts can disrupt multiple critical inputs simultaneously.

The strategic response for payments and fintech companies is not to retreat from AI or slow the buildout. The economics and competitive dynamics make that impossible. But it does argue for a more deliberate approach to infrastructure resilience. Multi-region cloud deployments with genuine geographic diversity. Supply chain visibility that extends beyond tier-one providers to the raw materials and energy sources underneath. Contingency planning that accounts for simultaneous disruptions rather than isolated incidents.

Some of the largest players are already moving in this direction. Microsoft, Google, and Amazon have all accelerated data centre investments in regions less exposed to Gulf chokepoints, including Scandinavia, the US Midwest, and Japan. But the hyperscalers building the infrastructure are not the only ones who need to plan for this. Every payments company running AI workloads on that infrastructure inherits the same geographic risk, whether they know it or not.

The uncomfortable lesson is not new. The pandemic taught it. The 2021 chip shortage reinforced it. But the payments and AI industries have been behaving as though the lesson was temporary rather than structural. The buildout has continued at full speed, with data centres expanding, model training intensifying, and agentic commerce protocols proliferating, all on the assumption that the physical infrastructure will hold.

It probably will, for now. IDC’s assessment that a short conflict produces modest impact is likely correct. But the vulnerability has been exposed. And the next disruption, whether it comes from Hormuz, the Taiwan Strait, or somewhere else entirely, will find an AI economy that has grown even more dependent on concentrated physical chokepoints.

For an industry that prides itself on building the future, the present is delivering a pointed reminder: the digital economy still runs on atoms, energy, and geography. No protocol can route around that.

The payments industry spent Q1 debating which agentic commerce protocol will win. The Iran conflict is asking a more fundamental question: what happens when the physical infrastructure underneath all of it becomes contested?