A copper crisis is coming, and almost no one is ready
The copper bottleneck is a supply-rate problem, not resource exhaustion: reserves are large, but demand is compounding across every electric use at once while new supply takes about 17 years to build, so the gap opens in the late 2020s and the world pays for it in higher prices and cancelled projects, not in running out.
Every claim below is traced to its primary source. Facts link to the exact place in the document, dataset, or transcript. Calculations and inferences show how they were derived, from the numbers up. This is the whole audit trail, unedited.
FactCopper is physically irreplaceable in electrical systems; the IEA calls it a cornerstone for all electricity-related technologies.high confidenceon screen: IEA quote card▸ trail
Where in the source: Executive summary, p.7
“copper being a cornerstone for all electricity-related technologies”
Source note: Revised 2022; page reference is to the executive summary p.7.
FactClean-power build-out is copper-heavy: an offshore wind plant uses roughly seven times the copper per MW of a gas plant.high confidenceon screen: intensity ladder, kg/MW▸ trail
Full ladder gas 1,100 -> offshore wind 8,000 kg/MW. The headline ratio is ~7x, not 8x.
Where in the source: Copper intensity by technology, kg/MW
Scope: Offshore wind 8,000 kg/MW vs a gas plant ~1,100 kg/MW = 7.3x. Say 'seven times,' never '8x' (8 is the kg/MW figure).
FactAn electric vehicle uses about 2.4 times the copper of a comparable combustion car.high confidenceon screen: 53.2 kg vs 22.3 kg▸ trail
Where in the source: IEA vehicle copper content (Global EV Outlook basis)
Scope: A battery EV uses 53.2 kg of copper vs 22.3 kg for an ICE car, 2.4x. Do not use the advocacy '3 to 4x / 80 to 100 kg' figure.
FactSubstitution is a weak escape valve: alternatives displace only about 1.3% of copper use a year, and aluminum, the main one, needs ~1.6x the cross-section and corrodes more.medium confidenceon screen: 1.3% per year / 1.6x cross-section▸ trail
The 1.3% rate is the copper industry's own survey, flagged, and the only measured figure that exists.
Where in the source: Copper Substitution Survey 2022
Scope: Substitutes displaced about 1.3% of copper use in the survey year. Industry-funded; the only measured figure available.
Source note: The industry's own trade body. Funding-bias flagged; it is nonetheless the only measured substitution figure that exists.
Where in the source: ASTM B230 (61% IACS) + Tehrani 2021
Scope: Aluminum conducts at ~61% of copper, so it needs ~1.6x the cross-section for the same current, and it is more prone to creep and corrosion.
FactWorld copper consumption reached 28.2 million tonnes in 2025, having grown about 2.7% a year for the past two decades.high confidenceon screen: 28.2 Mt (2025), ~2.7%/yr▸ trail
Where in the source: Table 1, p.9 (2025 preliminary)
“World refined copper usage reached 28,195 thousand tonnes in 2025.”
Where in the source: Annex p.60, computed over 2005-2025
Scope: 20-year compound growth of refined usage. WireSift computation on the ICSG series.
Source note: Intergovernmental study group. Annex p.60 carries production and usage on one consistent basis.
CalculationMore than half of all the copper ever mined has been mined since 2000.medium confidenceon screen: over half since 2000▸ trail
cumulative_production(2000-2025) / cumulative_production(all-history) = 53.7% to 57% (crossover ~2002-2003)
- seriesUSGS historical production + ICSG cumulative usage
Two independent accounting methods both put the post-2000 share above half. The claim is stated conservatively as 'more than half'.
Where in the source: USGS Data Series 140 (historical production) cross-referenced to ICSG cumulative usage
Scope: Two methods put the share of all copper ever mined that was mined since 2000 above half, crossover around 2002-2003. WireSift derivation.
FactThe IEA projects copper demand rising to about 33 million tonnes by 2035, and as much as 36.7 million tonnes by 2040 if announced pledges are met.high confidenceon screen: ~33 Mt (2035) / 36.7 Mt (2040 APS)▸ trail
33 Mt is the Stated Policies Scenario (2035). 36.7 Mt is the Announced Pledges Scenario (2040), and must read as the pledges case, not the base case.
Where in the source: STEPS demand, p.108
Scope: Stated Policies Scenario. The 36.7 Mt figure is the separate Announced Pledges Scenario for 2040, not the base case.
FactThe average grade of copper mines has fallen about 40% since 1991, predominantly due to reserve exhaustion.high confidenceon screen: -40% since 1991▸ trail
Where in the source: p.110
“The average grade of copper mines has decreased 40% since 1991, predominantly due to reserve exhaustion.”
FactA new copper mine takes about 17 years to go from discovery to full production, so today's discoveries do not add supply until well into the 2040s.high confidenceon screen: ~17 years▸ trail
Where in the source: p.110
Scope: IEA states the average lead time from discovery to full production is about 17 years. Do not loosen to '15 to 20'.
FactThe IEA forecasts mined copper supply peaking around 24 million tonnes in the late 2020s, counting every announced project, then falling below 19 million tonnes by 2035.high confidenceon screen: <19 Mt mined by 2035▸ trail
The project pipeline is already inside this curve; new mines do not reverse the decline.
Where in the source: p.108
Scope: Mined supply. Peaks around 24 Mt in the late 2020s counting every announced project, then declines below 19 Mt by 2035.
FactFinding copper has gotten far more expensive: the cost to discover a tonne rose from about $11 in 2003 to over $100 by 2023.medium confidenceon screen: $11 (2003) to >$100 (2023)▸ trail
The series peaked near $210/t in 2012 then roughly halved; concede the cycle if challenged.
Where in the source: MinEx unit-discovery-cost series, slide 38
Scope: Cost to discover a tonne of copper. Peaked at ~$210/t in 2012, then roughly halved to ~$102 by 2023. Real Dec-2025 US$.
Source note: Industry consultant. Unit-discovery-cost figures are in real Dec-2025 US$; the series peaked at ~$210/t in 2012 then fell, so cite the cycle, not just the endpoints.
FactBuilding new copper runs over budget: Teck's Quebrada Blanca 2 was sanctioned at $5.2 billion and finished above $8.6 billion, about 65% over.high confidenceon screen: $5.2B to $8.6B, +65%▸ trail
Where in the source: Dec-2018 sanction release (base capital) vs Oct-2023 cost guidance
Scope: The linked 2018 release states QB2 base capital of US$4.739B as of Jan 1, 2019; with escalation the all-in sanction is cited around US$5.0 to 5.2B. Final guidance (Oct 2023, Teck Q3 update) was US$8.6 to 8.8B, roughly 65% above the ~$5.2B escalated sanction. One blowout of three recent Teck builds; frame as cautionary, not the norm. A dedicated Teck primary for the 2023 final figure is a link still to add.
FactPrice alone is not unlocking supply: Freeport's Bagdad expansion has cleared its own $4/lb go-ahead price for years, yet across four straight investor updates the decision slipped backward while copper rose.high confidenceon screen: $4.00/lb cleared, still not approved▸ trail
The strongest single falsification of the 'higher prices will fix it' view in the research.
Where in the source: 4Q25 8-K; 1Q26 investor presentation
Scope: Bagdad's ~2x expansion needs $4.00/lb copper; copper has traded above $4 in five of the last six years, yet across four consecutive investor presentations the approval date moved backward from end-2025 to 2H-2026 while the price rose.
EstimateOn IEA figures the refined market moves from balance toward a gap of roughly 7 million tonnes by 2035, with the shortfall opening in the late 2020s.medium confidenceon screen: ≈7 Mt short by 2035▸ trail
iea_demand(2035) - (iea_mined_supply(2035) + iea_secondary_supply(2035)) = ≈7 Mt gap by 2035
- demandIEA demand ~33 Mt (2035, STEPS)
- mined-supplyIEA mined supply <19 Mt (2035)
- secondary-supplyIEA secondary supply ~7.4 Mt (2035)
- onset-windowOnset late 2020s (Goldman, S&P, IEA)
Adding the IEA's own mined and secondary supply series and subtracting from IEA demand gives an approximate 7 Mt refined gap at 2035. This is WireSift arithmetic on IEA figures, not an IEA-published number, and must never be conflated with the IEA's separate '30% short' mined-basis statement.
Where in the source: STEPS demand, p.108
Scope: Stated Policies Scenario. The 36.7 Mt figure is the separate Announced Pledges Scenario for 2040, not the base case.
Where in the source: p.108
Scope: Mined supply. Peaks around 24 Mt in the late 2020s counting every announced project, then declines below 19 Mt by 2035.
Where in the source: p.102, STEPS secondary supply
Scope: IEA secondary (scrap-based) refined supply. Added to IEA mined supply to build the refined-basis supply line.
Where in the source: Goldman 'from 2029 onwards'; S&P peak 2030; IEA late 2020s
Scope: Three independent onset estimates cluster in 2029 to 2031.
Source note: Sell-side and commercial-data commentary. Used only to triangulate the onset window, which the IEA's own late-2020s framing brackets.
InferenceThe copper bottleneck is a supply-rate problem, not resource exhaustion: demand is compounding across every electric use at once while new supply takes ~17 years to build, so the world pays in higher prices and cancelled projects, not by running out.high confidenceon screen: closing statement▸ trail
- demandDemand compounding to ~33 Mt by 2035
- supplyMined supply peaks then falls below 19 Mt
- supply-erosionGrades down 40% since 1991
- supply-rate~17 years to build a mine
- price-lever-failsPrice cleared for years, still no approval
- gap≈7 Mt gap by 2035, onset late 2020s
Reserves are large (USGS maps billions of tonnes), so this is not exhaustion. The binding constraint is the rate at which new supply can be permitted and built against demand that has grown almost every year for a century. Because price cannot compress a 17-year build, the market clears through cost and demand destruction, not depletion.
Sources are tiered T1–T6 by independence and rigor (peer-reviewed and primary filings at the top; company and advocacy material flagged as such). A claim marked estimateis an attributed third-party figure, not a hard reported fact. Where a source is advocacy-funded or a figure is announced-rather-than-realized, it is flagged on the claim. WireSift's sourcing standard is public at /methodology.