Why Copper - Not Rare Earths - Will Shape the Physical AI Market by 2035

Executive Summary

LONDON — January 31, 2026 — Physical AI—robotics, autonomous machines, and electrified infrastructure—will be constrained not by chip supply but by copper availability through 2035. Industry analysis from the International Energy Agency, BloombergNEF, and S&P Global projects structural copper deficits as electrification demand outpaces mining capacity. While rare earth elements determine motor efficiency, copper determines whether the Physical AI market can scale at all.

Key Takeaways

• Copper demand for Physical AI spans three converging sectors: electrified motion, infrastructure, and AI compute
• BloombergNEF projects copper demand could outstrip supply by 6 million metric tons annually by 2035
• Battery electric vehicles require approximately 83 kg of copper per unit according to the International Copper Association
• Reuters analyst polls project 2026 copper prices above $11,000/mt with 238,500-ton market deficit
• Rare earths enable performance optimization; copper enables market expansion

The Copper Math: Physical AI Scales on Electrification

Three copper-intensive trends converge in Physical AI deployment: electrified motion systems including robots, EV-like drivetrains, and automated logistics; electrified infrastructure encompassing grid upgrades, substations, and microgrids; and AI compute infrastructure spanning data centers, edge compute, and power delivery systems.

The IEA Global Critical Minerals Outlook 2025 identifies copper as one of the major pinch points in the energy transition pipeline, with charts showing a gap between required supply and expected supply from announced projects by 2035.

Table 1: Copper Content by Physical AI Application

Application	Copper Content (kg)	Primary Use Case	Source
Battery Electric Vehicle	83	Wiring, motors, inverters	International Copper Association
Electric Bus	224-369	High-voltage systems, motors	International Copper Association
Industrial Robot (Large)	15-25	Motors, wiring, controllers	Industry estimates
Humanoid Robot	8-15	Actuators, sensors, power	Industry estimates
Wind Turbine (Offshore)	8,000-15,000	Generators, cabling, transformers	IEA
EV Charging Station (DC Fast)	25-35	Power electronics, cabling	International Copper Association
Data Center (per MW)	4,000-6,000	Power distribution, cooling	BloombergNEF

Data Centers: The Hidden Copper Demand Driver

BloombergNEF has warned that data center buildouts amplify copper tightness. In their analysis, copper demand could outstrip supply by roughly 6 million metric tons per year by 2035, driven by electrification plus new infrastructure including AI-focused data centers.

S&P Global January 2026 research frames copper as foundational to the electrification era and ties demand acceleration explicitly to AI and defense-related electrification requirements.

Why Rare Earths Matter But Do Not Set the Ceiling

Rare earth elements—especially neodymium and dysprosium—matter most in high-performance permanent magnets used in compact, efficient motors. IEEE Spectrum summarizes the engineering trade: the highest-performing EV motors often use neodymium-iron-boron magnets, but alternative motor designs can reduce the performance gap.

The key distinction: REEs can be substituted in some applications with trade-offs, whereas copper has fewer broad substitutes without major redesign and efficiency loss. Aluminum can substitute in some wiring contexts, but retooling effort remains substantial.

Table 2: Commodities Price Forecast to 2035

Commodity	2024 Price	2026 Forecast	2030 Forecast	2035 Forecast	Source
Copper ($/mt)	$9,200	$11,975	$13,500-15,000	$16,000-20,000	Reuters Polls, J.P. Morgan
Aluminum ($/mt)	$2,400	$2,800	$3,200-3,500	$3,800-4,200	BloombergNEF
Nickel ($/mt)	$16,500	$18,000	$20,000-22,000	$24,000-28,000	Reuters
Neodymium Oxide ($/kg)	$72	$85	$95-110	$120-150	S&P Global
Lithium Carbonate ($/mt)	$13,500	$18,000	$22,000-25,000	$28,000-35,000	BloombergNEF
Gold ($/oz)	$2,050	$2,400	$2,800-3,200	$3,500-4,500	Goldman Sachs

Supply Constraints: Why Copper Cannot Scale Quickly

Reuters reported that Glencore 2025 copper output fell 11% due to water constraints and ore grade issues, with 2026 production forecast at 810,000-870,000 tons. The company outlined plans to grow output materially by 2035 through new and reopened mines.

The Financial Times has reported miners turning to new extraction techniques, including improved leaching and bioleaching, to pull copper from low-grade material and waste stockpiles because building new mines is slower and more expensive.

Table 3: Copper Supply-Demand Balance Projections

Year	Global Demand (Mt)	Mined Supply (Mt)	Balance (Mt)	Key Drivers
2024	26.5	25.8	-0.7	EV growth, grid upgrades
2026	28.5	27.8	-0.7 to -1.0	Data center buildout, robotics
2028	31.0	29.5	-1.5 to -2.0	Industrial automation acceleration
2030	34.0	31.0	-3.0 to -4.0	Physical AI deployment, grid modernization
2033	38.0	33.5	-4.5 to -5.5	Humanoid robotics scaling
2035	42.0	36.0	-6.0	Full electrification convergence

Sources: IEA Global Critical Minerals Outlook 2025, BloombergNEF, S&P Global Commodity Insights

Near-Term Price Signals

A Reuters poll of analysts published January 29, 2026, reported the median forecast for 2026 average copper price at $11,975 per metric ton—above $11,000 for the first time—with a projected market deficit of 238,500 tons.

J.P. Morgan commodities outlook projected a refined copper deficit of approximately 330,000 metric tons in 2026 and outlined a price path reaching around $12,500/mt in Q2 2026.

Equity Implications

Reuters reported that Freeport-McMoRan beat quarterly profit estimates as higher copper prices helped offset production setbacks at Grasberg, illustrating both leverage and operational fragility in tight markets.

Reuters also reported that Grupo Mexico, parent of Southern Copper, posted sharply higher quarterly profit driven by stronger metals prices and production, with significant investment plans over the next decade.

Why This Matters for Industry Stakeholders

For Physical AI strategists, the takeaway is not to bet on a specific price but to recognize that credible institutions increasingly frame copper as structurally tight. This changes the risk profile of any plan assuming fast, cheap electrification.

For Physical AI Companies

Practical procurement strategies include designing for copper efficiency, planning for aluminum substitution where possible, securing supply through long-term contracts, engineering for end-of-life copper recovery, and prioritizing site selection in regions with grid upgrade capacity.

For Investors

In a structurally tight copper market, equity narratives cluster into three buckets: low-cost scalable producers with expansion pipelines, technology-driven producers using leaching and waste recovery, and developers with high-quality projects that can clear permitting hurdles.

Forward Outlook

By 2035, the central question for Physical AI will look less like "Do we have enough GPUs?" and more like "Can grids, factories, and supply chains deliver enough copper-intensive electrification to support always-on machines and always-on inference?"

The policy debate is moving toward pipeline acceleration and domestic processing resilience. The UK Critical Minerals Strategy is one example of governments treating minerals as infrastructure strategy, not just trade policy.

Disclosure: This analysis is based on publicly available information from IEA, BloombergNEF, S&P Global, Reuters, and industry sources. Commodity forecasts are subject to market volatility and policy developments.