Copper and the Electrification Supercycle: Why 2026 Breaks the Bear Case: Deluair Consultancy

The copper market enters 2026 with a supply book structurally short of the demand it has signed up to serve. Codelco is guiding sub 1.4 million tonnes of mined output, the lowest since 1998, while Freeport's Grasberg is past peak grade, BHP's Escondida is grinding through head grade decay, and Glencore's Collahuasi expansion remains stalled on a water permit pending five years. First Quantum's Cobre Panama has been idled since November 2023, and Las Bambas blockade economics still trim 60,000 to 90,000 tonnes a year. AI training halls land at 3 to 4 kg of copper per kW against a traditional 2 kg, every battery electric vehicle embeds 83 kg against 22 kg, and Chinese smelter treatment and refining charges have collapsed to zero or negative, the cleanest signal that concentrate is the binding constraint. This brief sets out the price band, attribution, and capital allocation.

The mined supply problem is structural, not cyclical #

Global mined copper supply ran at roughly 22.6 million tonnes in 2024 according to ICSG provisional data, against refined usage of 27.3 million tonnes that the scrap pool and Chinese custom smelting closed only at zero or negative treatment margins. Codelco's 2024 production of 1.328 million tonnes was the lowest since 1999, and 2025 guidance of 1.34 to 1.40 million tonnes was missed at the lower bound. The 2026 internal plan, articulated in the company's January investor update, is anchored at 1.39 million tonnes with a downside skew toward 1.35 if El Teniente slips again.

El Teniente is the swing factor. The block cave transition, originally targeted for sustained 460,000 tonnes by 2025, has been pushed to 2027 at the earliest and head grade has fallen to 0.69 percent from 0.84 a decade ago. Andina is the partial offset, with Andes Norte extending mine life to 2058, but its grade is 0.70 percent and first new tonnes do not arrive before 2028. Chuquicamata Underground ramps slowly and Radomiro Tomic sulphide remains in feasibility with a capex envelope expanded from 4.0 to 5.5 billion dollars. Codelco now produces roughly 250,000 tonnes a year less than a decade ago, against a market that needs 6 to 8 million additional tonnes by 2035.

The story is not idiosyncratic. BHP's Escondida is guiding 1.18 to 1.30 million tonnes for fiscal 2026 against 1.32 million delivered in fiscal 2024, with head grade of 0.74 percent in December 2025 against 0.92 in 2018. Freeport's Grasberg block cave is past peak ore tonnage, and the company has flagged a 35 percent decline in Indonesian copper sales for 2026 versus 2024 as overburden management becomes the throughput constraint.

Stranded barrels and permitting drag #

Beyond operating mines, the 2026 balance is shaped by tonnes that are not flowing. First Quantum's Cobre Panama, before the November 2023 Supreme Court invalidation of its concession, produced 331,000 tonnes in 2023 and held a 350,000 tonne nameplate. The current administration has tied restart to a national environmental audit, copper export taxation, and a sovereign equity stake. Restart timelines now sit in the second half of 2027, removing roughly 1.2 million cumulative tonnes from the concentrate book between 2024 and 2027.

MMG's Las Bambas continues to operate, but blockade history through 2022 to 2024 has compressed the operating envelope. Output of 322,000 tonnes in 2024 sits well below the 380,000 to 420,000 nameplate band. Anglo American's Quellaveco runs at design, but the company's wider asset rationalization, including the coal divestment and pending sale of De Beers and nickel, has pulled growth capital away from Peruvian expansion.

The North American greenfield pipeline is the slowest moving piece. Rio Tinto and BHP's Resolution project in Arizona, with an indicated resource of 1.787 billion tonnes at 1.52 percent copper, has been in permitting since 2014 and the Final EIS has been re issued and re withdrawn three times under successive administrations. A realistic first production date of 2031 to 2033 is now standard in sell side models, with capital intensity north of 18,000 dollars per annual tonne. Brownfield expansions at existing Chilean and Peruvian operations cost 8,000 to 12,000 dollars per annual tonne, which is why incremental supply, when it appears, appears there first.

Operator and asset	2024 actual	2026 guidance midpoint	2030 internal plan	Head grade trajectory
Codelco group, Chile	1,328	1,390	1,500	0.61 percent declining to 0.58 percent
BHP Escondida, Chile	1,320	1,240	1,150	0.74 percent declining to 0.62 percent
Freeport Grasberg, Indonesia	780	560	620	0.96 percent then transitional dip
Glencore Collahuasi, Chile (50 percent JV with Anglo)	560	545	610 conditional on permit	0.83 percent declining to 0.75 percent
First Quantum Cobre Panama	97 stockpile shipments	0 to 90 conditional	330	0.41 percent flat
MMG Las Bambas, Peru	322	380	420	0.71 percent declining to 0.65 percent
Ivanhoe Kamoa Kakula, DRC	437	600	800 plus	5.3 percent declining to 4.1 percent
Anglo American Quellaveco, Peru	320	320	300	0.62 percent flat

Major copper mine output: 2024 actual versus 2026 guidance versus 2030 internal plan (kt of contained copper)

The DRC is the only meaningful new tonne #

Ivanhoe Mines Kamoa Kakula in the DRC is the single largest source of new mined copper this decade. Phase 1 and Phase 2 combined produced 437,000 tonnes in 2024, Phase 3 commissioning lifted 2025 output toward 520,000 tonnes, and the project's 500,000 tonne per year direct to blister smelter, energized in late 2025, captures the value previously paid to Chinese custom smelters. Head grade at Kakula remains in the 5 percent range, an order of magnitude above any other major operating copper mine, which is why project economics survive the 2018 mining code revisions and the logistics frictions of landlocked Lualaba Province.

The DRC contribution cannot close the global gap alone. CMOC's Tenke Fungurume and Kisanfu, the China backed Sicomines complex, and a wider second tier of Chinese owned operations together added roughly 1.3 million tonnes of DRC mined supply in 2024 against 700,000 tonnes in 2018. The structural ceiling is logistics. The Lobito corridor rail rehabilitation, financed by the US DFC and the EU Global Gateway, is the most important infrastructure variable for non Chinese offtake. Until Lobito throughput materially exceeds its current 1.5 million tonne annual capacity, Kamoa Kakula and peers remain capacity constrained on rail wagons and Walvis Bay diversions, not on metallurgy.

Demand is no longer a forecast: AI grids and EVs are landed #

Copper intensity in AI infrastructure has been understated. A traditional hyperscale data center at 8 to 12 kW per rack embeds roughly 2 kg of copper per kW of IT load across busways, distribution, transformers, and chillers. AI training halls densified to 80 to 120 kW per rack, with liquid cooling manifolds, 33 kV medium voltage distribution, and onsite substations carrying 300 MW or more, are landing at 3 to 4 kg per kW. Stargate, Hyperion, Stargate 2, Saudi Humain, and the Reliance Jamnagar AI builds collectively represent 20 to 30 GW of incremental IT load decisioned through 2030. At midpoint intensity that is 2.5 to 3.6 million tonnes of additional copper demand from AI alone over the period.

EV copper intensity is harder than the consensus number. A battery electric passenger vehicle embeds approximately 83 kg of copper across motor windings, battery pack busbars, inverters, and harness, against 22 kg in an ICE comparator. PHEVs sit at 60 kg. Heavy duty BEV trucks reach 200 to 300 kg. The IEA Global EV Outlook puts 2030 BEV plus PHEV sales between 41 and 50 million units in the announced pledges scenario, which translates to 2.8 to 3.4 million tonnes of incremental copper demand against 2024. DC fast charging at 350 kW per port and medium voltage cabling between substations and charging plazas add another 0.4 to 0.6 million tonnes by 2030.

Substitution risk is real on the HV transmission line itself, where aluminum conductor steel reinforced cable already takes the bulk of overhead lines. It is not real in motor windings, in battery pack internals, in inverters, or in the medium voltage distribution that connects AI compute to the grid. The boundary in 2026 looks the same as it did in 2016: aluminum wins in the air, copper wins inside the building.

End use	Copper intensity	Functional unit	Substitution exposure
Internal combustion passenger vehicle	22 kg	per vehicle	low
Battery electric passenger vehicle	83 kg	per vehicle	none in motor or pack
Heavy duty BEV truck	200 to 300 kg	per vehicle	none
DC fast charging plaza, 8 ports at 350 kW	1,200 to 1,800 kg	per plaza	low
Traditional hyperscale data center	2.0 kg per kW	per kW IT load	moderate at MV bus
AI training hall, liquid cooled, 100 kW per rack	3.5 kg per kW	per kW IT load	low
Onshore wind turbine	3,500 kg per MW	per MW nameplate	moderate at generator
Offshore wind turbine, fixed bottom	8,000 kg per MW	per MW nameplate	low, dense cabling
Utility scale solar PV with inverter	2,500 kg per MW	per MW DC	moderate
High voltage transmission line, AC	low, aluminum dominant	per km	high, aluminum substitutes

Copper intensity by end use, 2026 (kg of copper per functional unit)

TC/RC at zero, COMEX shorts, and the LME to SHFE arb #

The cleanest market signal that the bottleneck is concentrate, not refined metal, is the collapse of treatment and refining charges. The 2024 benchmark, set between Antofagasta and Jiangxi Copper, came in at 80 dollars per tonne and 8.0 cents per pound. The 2025 benchmark fell to 21.25 dollars and 2.125 cents. By Q1 2026, spot TC/RC for clean concentrate had touched zero and traded negative for several weeks, meaning Chinese smelters were paying miners for the privilege of processing their concentrate. CSPT formally proposed coordinated production cuts in March 2025 and again in February 2026, and the Tongling, Jinchuan, and Daye smelters have announced 2026 maintenance schedules that withdraw 8 to 12 percent of Chinese smelting capacity.

Inventory positioning is consistent. Combined LME, SHFE, and COMEX warehouse stocks ended March 2026 at roughly 280,000 tonnes against a five year average of 410,000 tonnes. The SHFE to LME arbitrage has run persistently positive on import parity since November 2025, pulling units east, while COMEX's premium over LME widened to 1,100 dollars per tonne in February 2026 on tariff hedging by US fabricators. Speculative net long positioning on COMEX, having flipped briefly to net short in October 2025 on China demand pessimism, was rebuilt to more than 70,000 contracts by early April 2026, the largest since the 2022 squeeze.

Price scenarios, 2026 to 2030 #

The 2026 to 2030 band sits in three scenarios. The base case, anchored on the announced supply pipeline materializing 70 percent on time and demand growing with the IEA's Stated Policies trajectory, places LME three month copper at 9,500 to 11,000 dollars per tonne across 2026 to 2028, lifting to 10,500 to 12,000 by 2030 as Cobre Panama, Resolution, and the second wave of Chilean brownfields fail to fully close the gap. The bear case, requiring Cobre Panama at full nameplate by 2027, Collahuasi permitted in 2026, and Chinese property completions stalling EV adoption, takes the band to 8,000 to 9,500 but does not reach the prior cycle's lows because the cost curve has shifted up. The bull case, with AI capex accelerating beyond the announced pipeline and one further major Chilean curtailment, takes it to 12,000 to 14,500 on a sustained basis.

The probability weighted expected price for 2027 sits near 10,800 dollars per tonne, with the 90th percentile right tail above 13,000. The 90th percentile incentive price for new greenfield supply, fully loaded at a 10 percent real return, sits at 11,500 to 12,000. That is the price that has to print, and persist, for Resolution and the next tier of Argentine and Pakistani projects to reach FID. Until it prints, the bear case for copper is mathematically a bet against electrification itself.

Capital allocation implications #

For end users, the dominant exposure is no longer price. It is allocation. Hyperscalers locking in 300 MW substations through 2028 are signing copper into 18 month order books at fabricators whose concentrate exposure runs through Codelco, Freeport, and BHP. Automakers running 2027 BEV launches face a similar problem. The strategic move is direct streaming or offtake from second tier producers in the DRC, at Oyu Tolgoi underground, and in Argentina's Vicuna district, executed through prepayment structures that lock concentrate without taking equity exposure to permitting risk.

For producers, the capital call is binary. Brownfield expansion at 8,000 to 12,000 dollars per annual tonne pays back inside the price band the market is already discounting. Greenfield at 18,000 dollars and a fifteen year permitting tail does not, until the bull case prints. The miners that compound shareholder value through 2030 will fund the brownfield, run existing operations harder against grade decay, and resist political pressure to greenlight greenfields at incentive prices not yet confirmed by the market.

Sources #

Cite this brief

@misc{hossen2026coppersupplysupercycle2026,
  author = {Hossen, Md Deluair},
  title  = {Copper and the Electrification Supercycle: Why 2026 Breaks the Bear Case},
  year   = {2026},
  url    = {https://deluair.com/consultancy/insights/copper-supply-supercycle-2026},
  note   = {Deluair Consultancy briefs}
}

Hossen, M. D. (2026). Copper and the Electrification Supercycle: Why 2026 Breaks the Bear Case. Deluair Consultancy briefs. https://deluair.com/consultancy/insights/copper-supply-supercycle-2026

Hossen, Md Deluair. "Copper and the Electrification Supercycle: Why 2026 Breaks the Bear Case." Deluair Consultancy briefs, 2026-04-26. https://deluair.com/consultancy/insights/copper-supply-supercycle-2026.

On the watchlist

Upcoming dates that bear on this brief.

See the full firm watchlist for the rest of the calendar.

Throughout 2026 Trade

DRC Code Minier review and royalty regime

Whether the export royalty rate moves up and the implication for CMOC, Glencore, and the Indonesian HPAL substitution flow.

Related insights

Adjacent reading.

Industrial policy and supply chains

Copper and the Electrification Supercycle: Why 2026 Breaks the Bear Case