The race to build artificial intelligence infrastructure is running headlong into a hard physical reality: the world's most critical chokepoint for industrial inputs sits in one of the planet's most volatile regions.
Nearly $650 billion in AI spending is committed to U.S. infrastructure this year alone, according to Moody's analysts. That investment rests on an assumption that has suddenly become precarious: that supply chains delivering essential materials will continue to function.
The Iran-Qatar conflict has exposed the fragility of those chains. Last month, Iranian attacks struck Qatar's Ras Laffan complex, which supplies roughly 30% of the world's helium. Helium is not a luxury input for chip makers. It is indispensable. The facility's operator says it can no longer fulfill existing contracts.
The damage means production cannot immediately restart even if the Strait of Hormuz reopens. A year ago, helium was plentiful enough that prices were falling. Now supplies have flipped to shortage, and spot prices have climbed sharply.
"The AI economy runs on tokens, tokens run on GPUs and GPUs depend on Qatari helium, Israeli bromine, and liquefied natural gas tankers with a single, 21-mile-wide exit from the Persian Gulf," Moody's analyst David Pan explained to Axios. That narrow passage handles roughly a fifth of global LNG shipments, a critical fuel source for the data centers powering AI systems.
The crisis reveals a broader pattern now defining the 2020s economy: supply shocks exposing dependencies that seemed manageable during decades of stability. The AI boom is simply the latest casualty of this new fragility.
Existing buffers provide temporary relief. Stored helium, long-term supply contracts, and recycled gases can buy time for semiconductor makers. But these are stopgaps, not solutions. The underlying problem persists: the physical foundation of the AI economy threads directly through a region engulfed in conflict with no clear resolution.
Complicating matters further, President Trump extended a ceasefire with Iran this week, yet Tehran's lead negotiator Mohammad Bagher Ghalibaf said the U.S. naval blockade makes "reopening the Strait of Hormuz impossible," signaling that the practical obstacles to restoring normal flow remain unresolved.
There is another layer to the pressure. The AI buildout is not merely facing supply constraints, it is also generating extraordinary demand. BlackRock portfolio managers Simon Wan and Tom Becker recently noted that component prices have surged 17-fold over the past year, a stark departure from decades of steady cost declines. Global spending on data centers and defense is driving this exponential surge.
"Whereas the 2020 jump in microelectronics prices were primarily attributable to supply chain disruptions, we now see a booming global industrial cycle generating excessive demand," they wrote. The distinction matters: supply problems can be solved with time and investment, but demand outpacing available supply creates a harder constraint.
For hyperscalers betting hundreds of billions on AI infrastructure, the equation is simple but unforgiving. They need steady flows of helium, bromine, natural gas, and semiconductors. Geography and geopolitics have made that assumption dangerous.
Author James Rodriguez: "The AI economy has discovered what every supply chain manager learned the hard way in 2020: betting everything on a single narrow passage through a warzone is a strategy with an expiration date."
Comments