Tradesnaut Quant Research Desk · September 24, 2026 · 6 min read · AI Market Analysis
More than half of planned U.S. data centers for 2026 have been delayed or canceled due to an acute shortage of transformers. This bottleneck extends beyond the physical hardware, signaling a deeper strain on the supply chains for critical materials essential to the artificial intelligence buildout. Copper, a fundamental component in everything from server cabling to grid infrastructure, has seen its demand surge, with prices on the London Metal Exchange reaching
4,661 as of September 22, 2026. Data center construction costs jumped from $630 to $960 per square foot between 2025 and 2026. This shift in cost and project timelines indicates a structural change, driven by the expanding energy and material footprint of AI.
The mechanism
The current commodity supercycle for 'compute metals' is distinct from prior periods. Unlike cyclical demand, AI infrastructure demand is capital-committed and infrastructure-driven, leaving no inventory overhang to correct the market. A 1-gigawatt AI factory demands 50,000 metric tons of copper, and data centers could add 750,000 metric tons of new copper demand annually. S&P Global projects total copper demand will rise 50% to 42 million metric tons by 2040 from 28 million tons last year. The copper market faces a projected deficit of 10 million metric tons by 2040. The critical constraint lies not only in raw material extraction but in manufacturing capacity for components like transformers. Wood Mackenzie's Q2 2025 survey identified a 30% supply deficit for power transformers and a 10% deficit for distribution transformers. Lead times for large power transformers average 128 weeks, while generator step-up units extend to 144 weeks. The United States imports approximately 80% of its large power transformer supply. A key bottleneck is the limited global capacity and long expansion cycles for high-magnetic induction oriented silicon steel (HiB/CRGO), a vital material for transformer cores. The shortage in this specialized steel is expected to persist until at least 2028, driven by aging grids, new energy connections, and AI. Building new transformer plants takes two years, but scaling the specialized workforce for coil winding, a craft requiring years of training, is a far slower process.
Who is exposed
Companies heavily invested in data center expansion face direct exposure to these material and component shortages. Hyperscale operators, including Amazon, Microsoft, Google, Meta, and Oracle, project over $600 billion in infrastructure spending in 2026, with roughly $450 billion allocated to AI infrastructure. This significant capital expenditure is increasingly consumed by rising raw material costs. Data center construction costs per square foot increased sharply between 2025 and 2026, with AI-optimized facilities costing 50-100% more. Electrical systems and cooling represent 40-45% and 15-25% of the data center budget respectively, areas highly dependent on copper and specialized materials. Mining companies, particularly those focused on copper and silver, see a direct benefit from the rising demand. Freeport-McMoRan (FCX), a major copper producer, reported Q2 2026 earnings per share of $0.74, exceeding analysts' expectations of $0.60 by 23.33%. The company’s stock traded at 72.36 at the last close. Silver, trading at 58.05, also finds increasing industrial applications in electronics, power infrastructure, and renewable energy systems, with the Silver Institute projecting a sixth consecutive year of supply deficits in 2026. Companies manufacturing electrical grid components, power equipment, and specialized steels are similarly exposed to the demand surge and the challenges of scaling production. Furthermore, a helium crisis, triggered by geopolitical events in the Gulf, is impacting semiconductor supply chains, as ultra-pure helium is critical for chip fabrication, with prices doubling.
Quantitative Outlook
The market data reflects the ongoing impact of these trends. Freeport-McMoRan (FCX) closed at 72.36, indicating market recognition of copper's strategic importance, despite a slight daily change of -0.22 (-0.30%). The United States Copper Index Fund (CPER) closed at 40.65, showing a modest daily gain of 0.04 (0.10%). The iShares Silver Trust (SLV) traded at 58.05, with a change of -0.11 (-0.19%). These movements occur within a broader context of a supply-constrained environment. S&P Global projects copper demand will rise to 42 million metric tons by 2040, against a projected supply deficit of 10 million metric tons. The persistent deficit in the silver market, now in its sixth year, further highlights material scarcity. Data center capital expenditures are forecast to accelerate, with hyperscalers spending over $600 billion on infrastructure in 2026. These costs are exacerbated by transformer price increases of 80% over five years by May 2026. A shift in this picture would require significant, long-term investments in new mining projects, the acceleration of transformer manufacturing capacity, and the development of a skilled labor force in specialized areas like coil winding. Government initiatives, such as the invocation of the Defense Production Act in April 2026 to prioritize domestic transformer manufacturing, suggest recognition of the problem but do not offer immediate solutions given multi-year lead times for equipment and worker training. Investors should monitor these long-term supply-side responses and any technological breakthroughs that could significantly reduce material intensity in AI infrastructure.
Tags: Copper, Silver, Critical Materials, AI Data Centers, Transformers
