Compute crunch and copper squeeze
Reports say advanced AI usage is outpacing global computing capacity, forcing product and pricing tradeoffs as companies contend with limited compute. At the same time, a market analysis warns AI infrastructure demand could add about 110,000 tons of copper in 2026—an 80% jump from two years earlier—creating competing pressure from defense and tariffs. (ibtimes.com) (markets.financialcontent.com)
Artificial intelligence companies are hitting a harder limit than software: not enough chips, servers, and power to keep up with demand. (ibtimes.com) The bottleneck is not just chip design. It is also the physical process of packaging advanced chips, building data centers, and connecting them to the grid, areas where suppliers have struggled to expand as fast as customers are ordering capacity. (microsoft.com) (reuters.com) Microsoft told investors in April 2025 that demand for its artificial intelligence services was growing faster than the data-center capacity it was bringing online, and Amazon said in April 2026 that demand for Amazon Web Services capacity and custom chips still exceeded supply. (microsoft.com) (cnbc.com) That shortage is showing up in product decisions. Companies have delayed rollouts, limited access, and pushed customers toward higher-priced plans because the most advanced models cost far more to run than standard cloud software. (ibtimes.com) (microsoft.com) Each new server hall also needs large amounts of copper, the metal that carries electricity through cables, transformers, cooling systems, and backup equipment. S&P Global said artificial intelligence infrastructure alone could add about 110,000 metric tons of copper demand in 2026, up 80% from two years earlier. (markets.financialcontent.com) (spglobal.com) S&P Global said data-center electricity demand is expected to rise from about 100 gigawatts in 2022 to 550 gigawatts by 2040, while defense spending and grid expansion are pulling on the same copper supply. (spglobal.com) The International Energy Agency said global electricity demand from data centers is set to more than double by 2030 to about 945 terawatt-hours, with artificial intelligence-optimized facilities the biggest driver of that increase. In the United States, the agency said data centers are on course to account for almost half of electricity-demand growth through 2030. (iea.org) Copper miners and market analysts argue that new supply will take years because large mines need permits, financing, and construction before they can produce metal at scale. S&P Global projects a potential 10 million metric ton copper shortfall by 2040 without major supply expansion. (spglobal.com) Technology companies say the spending is backed by real revenue. Amazon said on April 9 that its artificial intelligence business inside Amazon Web Services had reached a more than $15 billion annual revenue run rate, even as capacity constraints persisted. (reuters.com) (cnbc.com) The immediate question is no longer whether companies want more artificial intelligence. It is whether chip packaging plants, power grids, and copper supply can be built fast enough to keep the next wave online. (ibtimes.com) (iea.org)
