Copper, long recognized as a barometer of global industrial economic health and often dubbed "Dr. Copper," is now experiencing a record-breaking rally reminiscent of a hot technology stock, driven by massive investor bets. This surge is primarily linked to the unprecedented global wave of AI infrastructure construction and the soaring electricity demands from AI training and inference, which are expected to fuel a dramatic increase in copper demand. The fate of this industrial metal is now inextricably bound to the AI frenzy. A copper super-cycle is highly likely entering its initiation phase, but this is not the traditional cycle driven by a real estate boom. Instead, it is a new type of super-cycle shaped by the combined forces of "massive expansion in AI computing infrastructure demand, accelerated power grid construction, global energy transition, hard asset inflation, and constrained mine supply."
Within the AI computing supply chain, AI GPUs/ASICs form the computational core, HBM is central for data throughput, and power is the prerequisite for data center operation. Copper serves as the foundational conductor building the unprecedented power and computing systems and connecting them. Without sufficient copper, there would be inadequate high-speed copper cables, copper foil, and other copper components inside cloud server racks, nor the necessary power transmission, transformation, distribution, and liquid cooling systems for AI data center infrastructure. Lacking these, the AI computing empire would remain confined to chip orders and capital expenditure plans, unable to materialize into complete data centers. This highly conductive metal, crucial for data center construction and power grids, has recently seen its price movements closely mirror the stock fluctuations of AI computing leaders like NVIDIA and ASML. LME copper prices hit a record closing high last week, then retreated alongside volatility in AI-related stocks; on Friday, prices rallied strongly again as investors returned to the sector. Since the second half of 2025, traditionally low-volatility LME copper has surged by 50%, repeatedly setting new all-time highs. This sharp volatility reflects that copper's exposure to all the transmission lines, transformers, and critical electrical equipment needed to power AI data centers has become a key pillar for market optimism on copper's prospects. Currently, this factor is outweighing concerns about the impact of high oil prices from the Iran war on the more traditional industrial base closely tied to copper demand. A soaring stock market is unusually driving copper prices higher—the correlation between copper and U.S. tech stocks has reached its highest level since 2012. Note: Chart shows 40-day rolling correlation.
The head of base metals trading at DH Fund Management Co., one of China's largest hedge funds, stated, "This rally is primarily driven by structural thematic trading around AI computing." She holds a neutral to bullish outlook on copper, noting that further gains might require a new wave of investor capital flowing into tech-related sectors, coupled with tightening mine supply. Transitioning from an industrial barometer to the "power vascular system" for computing, copper is effectively shifting from a traditional "global industrial barometer" to an "AI infrastructure hard asset." The near-synchronous movement of copper prices with core AI computing stocks like NVIDIA, ASML, and Broadcom indicates that global capital has incorporated copper into extended/derivative trading of the AI computing chain: as long as AI data centers continue to expand, high-speed copper cables, copper foil, power grids, transmission lines, transformers, distribution equipment, liquid cooling systems, and backup power must expand in tandem.
AI is not the sole force behind the significant rise in LME copper prices. Investors have also been flocking to hard assets like copper as crucial inflation hedges, while long-term underinvestment in new mines risks creating severe supply shortages. Matt Miskin, Co-Chief Investment Strategist at Manulife John Hancock Investments, said, "Commodities have gone from a neglected asset class to one that is increasingly attractive to multi-asset investors." "For copper, it's really a three-legged stool: AI demand, inflation hedging, and a hot, running macroeconomic environment."
Analysts' views on the impact of the global AI super-wave on copper demand vary widely, with forecasts ranging from about 125,000 tons annually over the past three years to an estimated data center demand of 1.1 million tons in 2025. Prominent commodity trader Mercuria expects demand growth of about 350,000 tons this year. While this represents only about 2.5% of annual demand expectations, Nicholas Snowdon, Mercuria's Head of Metals Research, stated at an industry conference in Hong Kong earlier this month that he anticipates growth in the AI computing infrastructure sector to resemble trends seen during the golden development periods of China's EV and renewable energy industries; the latter two have become major sources of copper demand in less than a decade. Predictions for AI-related copper demand show significant divergence—analysts' forecasts for AI-driven copper demand vary greatly. Note: Marex data reflects the midpoint of its 100,000 to 150,000-ton demand forecast. S&P Global data is based on AI-related copper demand projected to grow from 1.1 million tons in 2025 to 2.5 million tons by 2040.
However, some analysts persist in warning that the copper market is prematurely pricing in demand increases. Recent research from commodity broker Marex Group and Oxford University points out that despite commitments from Microsoft, Alphabet's Google, and Amazon totaling about $580 billion for U.S. data center projects this year, the domestic U.S. data center industry faces severe bottlenecks in securing labor, power, equipment, permits, and community acceptance. Guy Wolf, Global Head of Market Analytics at Marex, said, "For copper, what really matters most is not the data center construction frenzy itself, but the power generation and transmission networks needed to support these data centers." "The AI narrative is indeed bullish for copper, but the actual realization of metal demand may be further out than many investors assume."
Even so, Wolf noted that given the massive scale of recent inflows into tech stocks, traders looking to bet on a reversal in this rally also face short-selling risks. As copper prices have risen, Wall Street money managers have added a net long position of approximately $14 billion in London and New York futures markets so far this quarter. Over the same period, the total market capitalization of Nasdaq 100 companies has increased by $7.8 trillion, according to institutional calculations. Wolf stated, "We haven't seen this level of institutional interest since the China economic super-cycle," referring to the period earlier this century when historically rapid Chinese growth fueled a commodity boom. "Even a small portfolio allocation shift into hard assets could overwhelm this market."
As shown in the chart above, tech stock inflows could overwhelm the copper futures market—in Q2, Nasdaq 100 market cap increased by over $200 billion daily. Bearish investors face another risk of getting burned: the re-emergence of a profitable trade opportunity linked to the risk of potential U.S. tariffs on copper imports. In recent weeks, New York Comex copper prices have jumped above London Metal Exchange prices, allowing traders to capture arbitrage opportunities by shipping metal to the U.S. On Friday, LME copper futures were around $13,630 per ton, while Comex prices were about $400 higher. Cancellation warrants for copper in LME-tracked warehouses saw their largest increase since 2013, indicating traders may be seeking to move stocks to Comex warehouses to capture higher prices. This dynamic could ultimately trigger a new round of bidding wars among buyers for available metal, even in a weaker demand environment.
Earlier this year, a rebound in Chinese demand supported copper prices, with strong spending in power grids and renewables offsetting losses in the property sector, which was the engine driving copper demand for much of the 21st century. However, prices near record highs are starting to dampen manufacturer demand. In the past, investors have often been caught off guard when rapid declines in Chinese spot demand overturned bullish bets on future growth. But Matthew Fine, who began investing in copper miners after joining Third Avenue Management as a portfolio manager in 2017, remains convinced the long-term trajectory is significantly higher. He stated that even under relatively conservative expectations, including new demand from AI data center-related industries, copper demand might only grow by 2.7% annually by 2040. The key point, however, is that mining companies are increasingly unable to keep pace with demand expansion because new projects are becoming scarcer and more costly to develop.
It's Not Just About GPUs and HBM! Copper Becomes the "Physical Vascular System" for AI Infrastructure, as Hard Asset Super-Cycle Gets a Computing Catalyst. Copper is indispensable in AI infrastructure precisely because it is the foundational metal for power transmission and distribution systems. A larger portion of AI's pull on copper lies not inside the data center buildings themselves, but in the restructuring of the power systems behind them. For copper, what truly matters is not the AI data centers per se, but the power generation and transmission networks required to operate them. Higher power density in AI server racks means data centers require higher-capacity transformation, lower-loss distribution, liquid cooling pump systems, backup energy storage, and stronger grid connections; this makes copper the "physical vascular system that converts power into computing power." AI data centers cannot run on AI GPUs/ASICs, HBM/NAND, data center CPUs, and broader AI server clusters alone. They require a stable power delivery system from generation sources, transmission networks, substations, UPS, busbars, cables, transformers, and switchgear all the way to the server racks and chips.
Forecast data indicates that global grid expansion is driving a new wave of copper demand, with grid infrastructure copper demand expected to increase from 12.52 million tons in 2025 to 14.87 million tons in 2030, with AI data centers and electrification further amplifying this demand. Furthermore, the copper bull market is not driven by AI alone, but by a triple resonance of AI demand, inflation hedging, and insufficient mine supply. Investors are using hard assets like copper as inflation hedges, while long-term underinvestment in new mines is creating a severe supply gap. Related research from S&P Global also warns that global copper demand could rise from about 28 million tons in 2025 to about 42 million tons by 2040, driven by AI infrastructure, energy transition, and grid upgrades, while mine supply growth struggles to keep pace.
Copper prices have recently moved in high synchrony with core AI stocks like NVIDIA and ASML, with the 40-day rolling correlation rising to its highest since 2012. This indicates capital has incorporated copper into "AI computing chain spillover or AI derivative trading": the massive expansion of AI data centers necessitates large-scale expansion of power grids, transmission, transformers, distribution equipment, cooling systems, and backup power, bringing copper closer to being the underlying bottleneck asset for AI hardware infrastructure.
"Commodity Bull Flag Bearer" Declares AI Computing Frenzy Ignites Super-Cycle. Jeff Currie, the veteran Wall Street strategist who accurately predicted the "commodity super bull market" during the global COVID-19 pandemic, recently stated in a media interview that the world is in the early stages of a new super-cycle of commodity demand. This cycle, driven by the collision between the frenzy of AI computing infrastructure construction and long-term underinvestment in the capacity of raw materials like energy and metals, could last another decade or longer. Currie's latest view aligns with that of Bank of America strategist Michael Hartnett, suggesting that the U.S.-China strategic competition, the difficulty of a complete cessation of Middle East geopolitical conflicts, and the global AI race are bringing heightened focus to core supply chain issues related to traditional energy. As resource-exporting nations like Indonesia plan new agencies to tighten commodity exports, and industrial metal prices for copper, aluminum, and nickel continue to rise this year—primarily due to surging demand for key metals and minerals essential for building AI data centers under the AI wave—commodities are effectively becoming an "AI core investment trend" beyond the AI computing infrastructure chain covering AI GPUs/ASICs, data center CPUs, HBM/NAND/HDD storage, 2.5D/3D advanced packaging, liquid cooling systems, optical interconnect supply chains, and data center power chains. It's not about buying models, chips, and high-performance AI cloud servers, but investing heavily in the underlying energy, metals, chemicals, and resource security premiums that support the capacity expansion of AI computing infrastructure.
Currie noted in the interview that U.S. oil inventories are particularly tight, while a sulfuric acid shortage has helped push copper to record highs, as this petroleum refining-derived chemical is crucial for copper production. The primary raw material for sulfuric acid production is sulfur. In modern industry, about 90% of global sulfur is recovered on a large scale from petroleum refining and natural gas processing. As per the core judgment of "commodity bull flag bearer" Currie—the AI data center construction frenzy is colliding with long-term underinvestment in energy and materials capacity, potentially placing the world in a commodity super-cycle that could last 10 years or even longer.
A recent research report from Wall Street giant Barclays indicates that as the AI construction boom transmits to commodity prices, developing nations like Chile, Peru, Brazil, Indonesia, and China stand to benefit. Among the emerging markets mentioned in Barclays' report, Chile and Peru are major copper exporters. Copper is crucial for AI-related electrification trends due to its extensive use in grids, transmission infrastructure, and core power management wiring in data centers. Additionally, Chile controls lithium, a raw material critical for the global energy storage technology roadmap—when we mention "battery energy storage," the vast majority of new projects opt for lithium batteries, firmly locking lithium into the position of "biggest beneficiary" in the global storage technology path. Indonesia is the largest producer of nickel, a key input for batteries and storage systems needed to support the growing power demands driven by AI. Meanwhile, China controls rare earth magnets, which are vital for semiconductor manufacturing equipment, data center infrastructure, robotics, and many other fields.