The semiconductor industry is no stranger to supply shocks, and the current environment still appears to be a manageable headwind.
The Iran war sent oil prices surging 58% in a single month, with the most immediate effects concentrated in high energy-importing economies. The conflict is also introducing new pressure points for the AI buildout, where several critical inputs trace back to the Persian Gulf. The AI supply chain entered this conflict with little slack. High bandwidth memory was already sold out for the year and advanced chip packaging had backlogs extending 1-2 years1. While chip companies have various levers they can pull to manage a supply shock, it does underscore how the AI investment boom is increasingly supply-constrained, not demand-constrained.
A ballooning helium shortage
Most people associate helium with party balloons. But the semiconductor industry is now one of helium’s largest consumers, relying on it to cool wafers during the etching process and to maintain the controlled environments required for advanced lithography. Unlike most industrial inputs, helium cannot simply be produced to meet demand. It forms through radioactive decay deep underground over billions of years, exists in small concentrations within natural gas deposits, and can only be captured as a byproduct when that gas is processed. Once released, it escapes into the atmosphere.
Qatar produces roughly a third of the world’s helium, and Iranian strikes have caused damage to critical infrastructure slashing output by 14%, with repairs that could take up to five years2. There’s no quick way to offset that production, and no current substitute for its role in chipmaking—or healthcare, aerospace and defense for that matter3.
There are supply buffers, and the cost impact should be modest. Major Asian chipmakers hold roughly three to six months of inventory, and even at double the price, helium accounts for less than 1% of production costs. But most liquid helium has an effective shelf life of 35 to 48 days before it escapes its containers, and if the Strait of Hormuz remains closed into the summer, dwindling inventories could pose a binding constraint on AI chip production.
Cost pressures for fabs
Energy costs may pose a more inflationary concern. In South Korea, where Samsung and SK Hynix dominate production of high bandwidth memory, industrial power prices have risen 39-55% year-to-date4. Taiwan is also strained, with 97.7% of its energy imported and only about seven days of LNG reserves. TSMC alone consumes 7-10% of Taiwan’s total electricity.
Chipmakers are quickly adapting—TSMC is prioritizing production at its most advanced nodes and Samsung has deployed helium recycling systems to extend existing inventory, while both governments have activated emergency energy protocols. These responses help mitigate a short-term supply shock, but an extended conflict may see these cost pressures passed along to customers, pressuring margins for U.S. tech companies.
A manageable risk, for now
The semiconductor industry is no stranger to supply shocks, and the current environment still appears to be a manageable headwind. AI producers like TSMC are expected to receive priority in helium allocation, and chipmakers have moved quickly to secure alternative sourcing and extend existing inventory. That said, a prolonged disruption would test the resilience of global supply chains and pace of the AI buildout. It may not derail structural demand for AI hardware, but it could temper it. For investors, the duration of this conflict will increasingly matter for the AI supply chain, posing meaningful second-order effects.