The Iran War Is Threatening the Entire AI Supply Chain — From Hormuz Helium to Chip Fabs to Data Centers, Here’s What Could Break

Georgi Kuzmanov Senior Equity Analyst & Founder

For more than three years, artificial intelligence has been the engine powering global markets — driving trillions in investment, pushing indices from New York to Tokyo to record highs, and reshaping every sector from health care to finance. But the technology that was supposed to be immune to old-economy constraints has a dirty secret: it runs on fossil fuels, rare chemicals, and manufacturing chains that cross more than 70 borders before a finished chip reaches an end user. The war with Iran has laid this vulnerability bare.

As the Strait of Hormuz enters its second month of effective closure, the consequences are cascading through the entire AI infrastructure stack. East Asian chip fabrication plants face energy shortfalls. Qatar’s halted helium production is disrupting wafer cooling. Nearly half of all planned U.S. data centers are at risk of delay or cancellation. And the private credit market that finances much of this buildout is in the early stages of a liquidity crisis. What follows is a layer-by-layer analysis of how the Iran conflict is threatening the AI boom — and what it means for investors.

Layer 1: The Chemical and Energy Inputs

Helium — The Invisible Bottleneck

Roughly one-third of the world’s helium supply comes from Qatar, where it is extracted as a byproduct of liquefied natural gas processing at the massive Ras Laffan industrial complex — the largest LNG facility on Earth. Helium is not a luxury input for chipmakers; it is essential. The gas is used to cool silicon wafers during fabrication, and there is no commercially viable substitute at scale. South Korea and Taiwan, which together account for the overwhelming majority of advanced semiconductor manufacturing, source the bulk of their helium from Qatar.

Ras Laffan has been offline since March 2 following Iranian drone strikes. Qatari officials have stated that the damage was more severe than initially reported, with lasting consequences for the facility’s infrastructure. Phil Kornbluth, founder of Kornbluth Helium Consulting, estimates that even after the conflict ends, it will take four to five weeks to restart production at the facility, followed by an additional two to three months to fully restore the global helium supply chain to its pre-crisis state.

Sulfur and Bromine — The Forgotten Dependencies

Helium gets the headlines, but two other chemicals critical to chip manufacturing are equally exposed. Approximately half of all seaborne sulfur shipments — sulfur is used for wafer cleaning during fabrication — transits the Strait of Hormuz. Even before the war, sulfur supplies were constrained due to surging demand from the EV battery and technology sectors. The closure has turned a tight market into a critical one.

Bromine tells an even more concentrated story. The Dead Sea is the world’s largest source of bromine, a chemical essential for etching circuit patterns onto silicon wafers. South Korea imports nearly its entire bromine supply from Israel. With Israel engaged as a direct belligerent in the conflict and shipping routes through the region disrupted, bromine deliveries are irregular at best and halted at worst.

Energy — The Existential Threat to East Asian Fabs

The semiconductor supply chain begins and ends with energy. Taiwan’s TSMC fabricates approximately 90% of the world’s advanced-node semiconductors and nearly all of the high-end AI chips designed by Nvidia, the world’s most valuable company. South Korea’s Samsung Electronics and SK Hynix together produce roughly 80% of global high-bandwidth memory and nearly 70% of all DRAM — the chips that power AI systems, cloud data centers, smartphones, and automobiles.

Both countries are almost entirely dependent on imported fossil fuels for electricity generation. Taiwan relies on the Middle East for more than a third of its LNG needs. South Korea imports virtually all of its energy. One-fifth of the world’s oil and LNG moves through the Strait of Hormuz. The arithmetic is straightforward: if the strait stays closed, the fabs that produce the world’s AI chips face energy rationing within months.

Layer 2: Chip Manufacturing Under Stress

The inputs crisis flows directly into fabrication. TSMC, Samsung, and SK Hynix have not yet announced production curtailments, but the supply constraints are building behind the scenes. Cathay Pacific Cargo, which handles approximately 30% of global wafer transport, has reported limited access to its regional hub in Dubai due to the conflict, creating logistics bottlenecks even for finished chips awaiting delivery.

If the Hormuz closure extends into mid-April and beyond — as BCA Research’s chief strategist Marco Papic has warned could trigger the world’s first post-COVID supply chain disruption — chip prices will surge sharply as manufacturers rationalize output and compete for increasingly scarce inputs. The memory chip market, already tight due to the AI-driven demand cycle, would be the first to feel the impact. SanDisk’s 168% year-to-date return reflects the market’s early pricing of this scenario.

Nvidia, whose H100 and B200 chips are the workhorses of the AI revolution, is particularly exposed through its manufacturing dependency on TSMC. If TSMC is forced to ration production — prioritizing military and government contracts over commercial AI orders, as some analysts have speculated — the ripple effects through the entire AI ecosystem would be immediate and severe. Every hyperscaler, every AI startup, and every enterprise customer is downstream of TSMC’s fabrication capacity.

Layer 3: Data Center Construction Is Stalling

The consequences extend beyond chips to the physical infrastructure that houses them. According to a Bloomberg analysis of Sightline Climate data, nearly half of the data centers planned to come online in the United States in 2026 face delays or outright cancellation. Of the approximately 12 gigawatts of new capacity announced for this year, only one-third is actually under construction. The rest remains in early planning stages and is increasingly at risk.

The primary bottleneck is not land or permits — it is electrical components. Batteries, transformers, and switchgear represent less than 10% of a data center’s construction cost but are absolutely essential for completion. “If one part of the supply chain is delayed, the entire project cannot be completed,” said Andrew Likens of Crusoe Energy Systems. “Right now, the situation is a very complicated puzzle.”

U.S. demand for these components has outstripped domestic supply, forcing companies to source from Canada, Mexico, South Korea, and China — extending lead times and adding logistical complexity at a moment when global supply chains are already strained by the Hormuz disruption.

The Energy Cost Squeeze

In the United States, where hyperscalers are expected to spend $650 billion on AI infrastructure this year, approximately 75% of the planned on-site power generation relies on natural gas. But American LNG exporters are redirecting shipments to Europe and Asia, where the Hormuz-driven shortfall commands premium prices. This dynamic is pushing domestic energy costs higher, directly impacting data center economics. Electricity represents roughly half of a data center’s operating expenses.

Layer 4: The Middle East Data Center Pivot

The war has also disrupted the emerging narrative of the Middle East as a global data center hub. In recent years, Gulf states — eager to diversify their economies and align with Washington by distancing from Beijing — had attracted major investment commitments from Nvidia, Oracle, Microsoft, and OpenAI for regional data center projects. Iran’s retaliatory strikes early in the conflict deliberately targeted data centers in neighboring countries, causing disruptions to banking, payment processing, and cloud services within the first week.

“If geopolitical risk in the Persian Gulf continues to rise, companies may accelerate projects in Northern Europe, India, or Southeast Asia, where energy supplies, regulatory frameworks, and security conditions are more predictable,” said Patrick Murphy, head of the geopolitical unit at Hilco Global.

Tess de Blanc-Knowles, senior director at the Atlantic Council, offered a more nuanced view: companies are unlikely to abandon the region immediately, but they may slow the deployment of new capital or pause planned partnerships. The distinction matters — even a temporary deceleration in Gulf data center investment would redirect hundreds of billions of dollars and reshape the global map of AI infrastructure for a decade.

Feng Qi Yu, a professor of energy systems engineering at Cornell University who visited the UAE recently, noted that even before the war, the region faced challenges — particularly water consumption for cooling. Now the primary advantage the Middle East offered — cheap and reliable access to seemingly unlimited energy — is no longer a certainty. “In the short term, many companies will probably reassess,” he said.

Layer 5: The Financial Architecture Is Cracking

Perhaps the most underappreciated risk is the convergence of the AI supply chain disruption with the private credit crisis. A significant portion of data center construction is financed through the private credit market — the same $1.8 trillion sector currently experiencing a liquidity crisis. Blue Owl Capital, whose flagship BDC received 40.7% withdrawal requests this week, has significant exposure to technology-adjacent lending including AI infrastructure. The private credit freeze is reducing the capital available for AI-related capex at precisely the moment costs are rising.

Paul Kedrosky, an investor and fellow at MIT’s Institute for the Digital Economy, told Time magazine that the war had made him “significantly more worried about systemic economic risks related to AI” because “the consequences are unknown in terms of how it affects this highly interconnected energy and information network.”

The chain of transmission runs from rising energy costs to higher data center operating expenses to less favorable unit economics to reduced private credit availability to slower construction to constrained AI capacity to lower revenue growth for hyperscalers to declining equity valuations to further capital withdrawal. Each link reinforces the next.

Elon Musk’s Space Data Center Gambit

In a characteristic display of first-principles thinking, Elon Musk has proposed building data centers in orbit — using SpaceX’s Starship rocket to deploy modular computing infrastructure beyond the reach of terrestrial energy disruptions. The concept is technically fascinating but faces insurmountable near-term obstacles, according to experts.

Microsoft’s 2015 Project Natick tested a similar concept underwater, deploying a sealed data center off the Scottish coast that used seawater for cooling. Despite the technological promise, the project was shelved due to lack of commercial demand and economic viability. “These challenges will likely be even more serious in space than underwater,” said Roy Chua of the research firm AvidThink.

The fundamental problem is obsolescence. AI chip generations turn over rapidly — once a satellite data center is launched, its hardware cannot be upgraded or repaired. Cost estimates for the space data center program range from $1 trillion to several trillion dollars, and the logistics would require approximately 3,000 SpaceX launches per year — roughly eight per day, compared to the 167 launches the company completed in all of 2025. Starship, the vehicle designed for this mission, has not yet achieved a stable orbit in testing.

The Investment Implications

For investors, the Iran war’s threat to the AI supply chain creates a complex risk landscape that the market has only partially priced.

There are reasons to temper the alarm. AI companies continue to grow revenue rapidly — Anthropic reportedly doubled its revenue run rate since late 2025. Bloomberg noted that some traders are using AI tools more intensively, not less, to analyze the war’s market impact. As Professor Yu of Cornell observed: “Look at how powerful these AI models are and how quickly they are being adopted. Demand is clearly growing, not stopping because of this conflict.”

The most likely outcome is not the death of the AI boom but its deceleration and geographical redistribution. Companies will become more selective about where they deploy capital. Data center investment will shift toward regions with more stable energy supplies — Northern Europe, India, Southeast Asia. Chip manufacturers will accelerate diversification away from single-source chemical dependencies. And valuations will adjust to reflect a world where AI infrastructure costs more and takes longer to build than the market assumed six months ago.

“My view is that companies will probably be more selective,” said Professor Yu, “but not necessarily less committed.” That may be the most optimistic credible assessment available. The less optimistic one comes from BCA Research’s Papic: mid-April is the deadline, and the clock is running.

Georgi Kuzmanov

Senior Equity Analyst & Founder at AlphaEdge. Columbia University MSFE (2011–2013). Covering equities, macro, and geopolitics for serious investors.