Copper Is Near Record Highs. Is AI Data-Center Demand Really Driving It?

by Aug 31, 2026Market Radar

Independent market context reviewReviewed by Andreas Torgersen · BSc Finance, BI Norwegian Business School

MarketInsiderLab separates structural market themes from short-term price drivers using company filings, earnings releases, primary-source disclosures and documented market evidence. References to companies are for research and context, not stock recommendations.

Human reviewedPrimary-source focusedNo stock recommendationsMethodology disclosed
KEY TAKEAWAY

Copper has become a real physical input to the AI infrastructure buildout, but the 2026 copper rally cannot be explained by AI alone. Trade policy, inventory geography and mine supply are also moving the market. The useful question for investors is therefore not “which miners mention AI?” but “where is copper exposure measurable in production, earnings and actual customer demand?”

Copper is increasingly described as an “AI metal.” That is directionally useful but incomplete. Data centers need dense electrical infrastructure, while the grids serving them require more substations, transformers and transmission. Copper sits throughout that chain. But 2026 price action also reflects tariff expectations, regional inventory dislocations and constrained mine supply. This analysis separates the structural AI story from the near-term market mechanics.

LME copper peak, Aug. 2026$14,343/t
2026 forecast global surplus639k t
BHP FY26 copper output~2 Mt
Teck Q2 copper production135.9k t

Why Copper Is Suddenly Part of the AI Trade

The AI boom is not only a semiconductor story. Every large data-center campus needs high-voltage connections, transformers, switchgear, backup systems, cooling equipment and dense internal power distribution. Copper appears throughout that stack. It also sits upstream in the transmission expansion needed to bring new generation to data-center clusters.

This is why the copper theme fits naturally beside MarketInsiderLab’s analysis of AI data-center power-grid stocks. The earlier analysis maps the companies building the physical power system. Copper is one of the raw materials flowing through that same buildout.

The causal chain matters: AI can strengthen long-term copper demand without being the main reason copper rises in a given month. Reuters reported in late August 2026 that copper was approaching record highs even as analysts still expected a global surplus, because prospective U.S. tariffs were pulling refined metal into the United States and draining inventories elsewhere.

The 2026 Copper Rally Is More Complicated Than “AI Demand”

Structural demand

AI + grids + electrification

Data centers add load and require physical electrical infrastructure. That raises copper’s strategic importance over a multi-year horizon.

Near-term pricing

Tariff-driven inventory shifts

Expected U.S. tariffs have encouraged shipments into U.S. inventories, tightening availability elsewhere without necessarily implying an immediate global shortage.

Supply constraint

New mines are slow

Large copper projects can take years to permit, finance and build. Higher prices do not translate into instant new supply.

Narrative risk

Price is not volume

A miner can report stronger earnings because copper prices rise even while its own production falls. That distinction matters when comparing exposure.

THE CLEAREST AI-TO-COPPER EVIDENCE

Rio Tinto → AWS turns the narrative into a documented supply chain

In January 2026, Rio Tinto announced a two-year collaboration under which Amazon Web Services will use the first copper produced with Rio Tinto’s Nuton bioleaching technology in components of its U.S. data centers. Rio Tinto says data centers use copper in electrical cables and busbars, transformer and motor windings, printed circuit boards and processor heat sinks.

Johnson Camp, Arizona
Nuton produces refined copper cathode at the mine gate.
AWS U.S. data centers
Copper moves into components used in physical AI infrastructure.

This does not prove that AI is the dominant driver of global copper prices. It does something more useful: it provides direct evidence that hyperscale data-center demand is entering the copper supply chain. That is stronger evidence than inferring AI exposure simply because a company owns copper mines.

5 Listed Companies With Measurable Copper Exposure

BHPBHPScale + growth pipeline
SCCOSouthern CopperHigh price sensitivity
TECKTeckRising production
IVN / IVPAFIvanhoe MinesConcentrated exposure
RIORio TintoDirect AWS evidence
Company Current evidence Why the theme matters What can break the thesis
BHP ~2 Mt FY26 copper output; copper generated more than half of underlying EBITDA Large existing production plus a multi-country growth pipeline Diversified portfolio reduces pure copper sensitivity
Southern Copper Q2 sales +40.6% while copper sales volumes fell 1.5% High sensitivity to copper pricing Recent earnings strength was price-led more than volume-led
Teck Q2 copper production 135,900 tonnes, +25% YoY Captures both higher pricing and operating growth Project execution and strategic transformation risk
Ivanhoe Mines Kamoa-Kakula produced 64,328 tonnes in Q2 2026 More concentrated copper growth exposure Asset, jurisdiction and operating concentration
Rio Tinto AWS will use Nuton copper in U.S. data-center components Direct evidence connecting copper supply to hyperscale data centers Rio remains diversified; Nuton volumes are small relative to global copper markets

1. BHP: Copper Has Become the Earnings Engine

BHP is still a diversified mining giant, but its financial mix has shifted. In FY2026 the company said copper contributed more than half of underlying EBITDA for the first time. BHP produced roughly 2 million tonnes of copper for a second consecutive year and says its project pipeline across Chile, Australia and Argentina could lift copper production by around 40% by FY2035.

2. Southern Copper: Price Exposure Is Not the Same as Production Growth

Southern Copper is a useful lesson in how commodity exposure works. In Q2 2026, net sales increased 40.6% and net income rose sharply, but copper sales volumes fell 1.5% and mine production declined 3.5% year over year. The company clearly benefited from the copper rally, but the quarter was driven more by price than by higher physical output.

3. Teck: Higher Price and Higher Volume

Teck showed a different pattern. Q2 copper production rose 25% year over year to 135,900 tonnes, while the company also benefited from record copper prices. That combination gives the operating evidence more weight than a purely narrative-driven “AI copper” thesis.

4. Ivanhoe Mines: Concentrated Exposure, Concentrated Risk

Ivanhoe Mines offers a more concentrated copper profile through Kamoa-Kakula in the Democratic Republic of Congo. The operation produced 64,328 tonnes of copper in Q2 2026. Concentration can increase sensitivity to copper prices, but it also makes mine-specific execution, grades, recovery rates, power and jurisdictional risk more important.

5. Rio Tinto: The Most Direct AI Evidence, But Not a Pure-Play Copper Stock

Rio Tinto deserves a separate category. It is a diversified miner rather than a pure copper play, but the AWS agreement gives it unusually direct evidence of AI-infrastructure demand. AWS will use Nuton copper produced at Johnson Camp in components of U.S. data centers, while AWS also provides cloud and analytics support to help optimize the Nuton process.

The agreement should not be overstated: the project is targeting roughly 30,000 tonnes of refined copper across a four-year deployment period, tiny relative to the global copper market. Its importance is evidentiary rather than volumetric. It shows that the AI-data-center narrative has moved from forecasts into an identifiable commercial supply chain.

What the AI-Copper Narrative Gets Right

  • AI data centers require electrical infrastructure, not just GPUs.
  • Copper is embedded in cables, busbars, transformers, motors, circuit boards and cooling systems.
  • Grid expansion adds copper demand outside the data-center fence as well as inside it.
  • Long mine-development timelines can make sustained demand growth difficult to match quickly.
  • The Rio Tinto–AWS agreement provides direct commercial evidence of data-center copper demand.

What the Narrative Can Overstate

  • AI is not the only driver. Construction, grids, EVs, industrial demand and trade flows all matter.
  • A tight regional market is not the same as a global shortage. Inventory geography has been heavily distorted by tariff expectations.
  • High copper prices do not guarantee higher production. Grades, outages, water, permits and mine plans determine volumes.
  • Direct AI evidence does not mean material earnings exposure. The Rio/AWS deal is strategically interesting, but current Nuton volumes are small relative to the global market.
MARKETINSIDERLAB CONCLUSION

The AI-copper thesis is real as a structural demand story, but too simplistic as an explanation for the 2026 copper rally. Near-term pricing is being shaped by tariffs, inventory geography and supply constraints alongside long-run demand from AI infrastructure and electrification.

The strongest listed-company evidence comes from different places: BHP for scale and long-term growth, Southern Copper for price sensitivity, Teck for rising production, Ivanhoe for concentrated copper exposure, and Rio Tinto for the clearest documented AI-data-center supply-chain link.

Structural driverAI + grid electrification
Near-term driverTariffs + inventories
Best evidenceProduction + customer contracts
Main narrative riskAttributing every copper move to AI

Sources

Editorial disclosure: MarketInsiderLab independently analyzes public market narratives, company disclosures and alternative data. References to companies and themes are for research and commentary only.
Investment disclosure: This article is for informational and research purposes only. It is not investment advice, a recommendation to buy or sell securities, or a price target.