Ming-Chien Chyu, PhD, PE, Founding President, Healthcare Engineering Alliance Society (HEALS), and professor at Texas Tech University.
getty
Geopolitical tensions in the Middle East have sent shockwaves through the global high-tech industry. This disruption is a reminder of a deeper structural reality: Modern technology is built on an intricately interdependent global supply chain.
The 2026 conflict has exposed how tightly coupled these systems are, particularly in areas like energy, materials and logistics where the Middle East plays an outsized role. What might once have seemed like a distant geopolitical issue is now surfacing as a direct operational and financial risk for companies building the world’s most advanced AI systems, semiconductor fabs and data center networks.
By 2028, data centers are projected to consume roughly 12% of U.S. electricity. Energy already accounts for 30% to 60% of data center operating expenses, and the surge in crude oil and natural gas prices across Europe and Asia has sharply increased the cost of running AI systems. Inference—now responsible for 80% to 90% of an AI model’s lifetime energy use—has become significantly more expensive, compressing margins for companies that rely on high‑volume API calls.
This pressure comes as autonomous, agentic AI systems are becoming mainstream. These systems run continuously, reasoning and planning in loops, and are far more sensitive to energy‑price volatility than earlier generations of AI.
As a result, pricing itself is starting to evolve from static, compute-based models toward more dynamic frameworks that account for energy volatility, signaling a potential decoupling of AI pricing from purely software-driven economics.
The hardware ecosystem supporting AI is also under strain. While high‑end GPUs often ship by air, most supporting infrastructure—cooling systems, racks, power‑distribution units and bulk memory—moves by sea. Dubai and Jebel Ali, two major transshipment hubs, have been heavily disrupted. Air‑freight congestion and maritime diversions around Africa have added weeks of delay to infrastructure build‑outs.
Memory markets were already tight before the crisis. DRAM, NAND and especially high‑bandwidth memory were in short supply due to surging AI demand. There is now a premium on these components, raising costs for companies training next‑generation frontier models.
One of the most severe impacts today has been on the specialized gases and chemicals required for semiconductor fabrication. Qatar produces about 30% of the world’s helium, a critical input for lithography machines. Helium is essential for cryogenic cooling of superconducting magnets and for maintaining the thermal stability of silicon wafers, but conflict has reduced global helium output by roughly one‑third.
Unlike oil, helium cannot be easily stockpiled at scale. Taiwan and South Korea—home to the world’s most advanced fabs—are heavily dependent on Qatari helium. If a fab runs out, lithography machines must be shut down, and restarting them can take weeks of recalibration.
Shipping disruptions compound the problem. With vessels rerouting around Africa, transit times have increased by 10 to 20 days. Liquid helium slowly boils off during transport, meaning a significant portion of each shipment evaporates before arrival. In response, companies like TSMC are accelerating investments in helium‑recovery systems potentially capable of recycling up to 90% of existing stocks.
The Gulf also exports 45% of the world’s sulfur, used to produce sulfuric acid—the most widely used chemical in chip fabrication. A shortage affects not only cutting‑edge 2-nanometer and 3-nanometer chips but also mature‑node chips used in automobiles and medical devices. Meanwhile, bromine supplies from Israel and Jordan—critical for etching and flame retardants—have been disrupted, raising costs for semiconductor manufacturers.
While oil often dominates headlines, natural gas (LNG) is the hidden backbone of the AI revolution. Semiconductor fabs are among the most energy‑intensive facilities in the world, relying heavily on natural gas for both power and chemical processes. The 2026 Gulf conflict has removed 20% of global LNG supply.
Although the U.S. is the world’s largest natural‑gas producer, domestic delivery is constrained by aging pipelines and infrastructure bottlenecks. The issue is not a shortage of gas in the ground but the inability to transport it efficiently to data center hubs.
This constraint is beginning to reshape how tech companies think about infrastructure—not just as a scaling problem, but as a control problem—pushing them toward greater ownership or influence over power generation and delivery.
Natural gas has become the AI industry’s swing fuel, providing roughly 40% of the electricity used by U.S. data centers. As LNG spot prices doubled in early March, the cost of running AI models rose sharply. Training a next‑generation frontier model now requires hundreds of millions of dollars in electricity alone. Some labs have paused training runs to avoid burning through capital at unsustainable rates.
To bypass slow public-grid expansion, many tech giants began building on-site natural-gas power plants in 2025. This behind-the-meter strategy was intended to provide greater certainty for AI workloads by reducing dependence on constrained public infrastructure.
Instead, the current disruption reveals an important limit: Private generation does not eliminate exposure to upstream fuel markets. When gas becomes scarce or politically prioritized for residential heating and essential industries, energy independence proves more conditional than absolute.
East Asian tech economies are uniquely vulnerable. Taiwan and South Korea rely on LNG for 25% and 19% of their primary energy consumption, respectively, and 83% of LNG passing through the Strait of Hormuz in 2024 was destined for Asia.
In response, the crisis is accelerating a broader rethinking of energy strategy, with greater emphasis on diversification into sources like small modular reactors and geothermal—not just as alternatives, but as mechanisms for reducing systemic exposure to single-region supply shocks.
Taiwan's advanced fabs require constant baseload power, and the island has only about 11 days of LNG reserves. Any disruption to shipments from Qatar or the UAE could destabilize the grid and halt production of the world’s most advanced chips. China has more diversified energy sources but lacks firm power in its eastern tech hubs, which depend heavily on imported LNG.
The high-tech industry is unlikely to remain as cost-efficient, globally fluid or energy-agnostic as it has been over the past decade. Instead, this moment is accelerating a structural shift toward a “fortress tech” model, where proximity to stable allies and reliable energy sources matters more than raw compute.
If current pressures persist, the industry may be forced to make difficult trade-offs between growth, cost and supply continuity—including the possibility of reduced wafer output in the near term.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。