LME copper cash settlement sustains at or above each of $12,000 / $15,000 / $20,000 / $25,000 per tonne (≥1 quarter average)

---
title: LME copper cash settlement sustains at or above each of $12,000 / $15,000 / $20,000 / $25,000 per tonne (≥1 quarter average)
status: draft
dimensions: ["metals","commodities","utilities","housing"]
horizon: medium
trigger: LME copper cash settlement price (3-month rolling average) first sustains at or above each of {$12,000, $15,000, $20,000, $25,000} per tonne for at least 1 calendar quarter.
timeline: {"p10":2026,"p50":2028,"p90":2031}
confidence: medium
sub_gates: [{"slug":"copper-sustained-12k","p50":2026,"why":"LIKELY ALREADY RESOLVING. LME cash $13,545/t on 22 May 2026, Q2 2026 3-month rolling avg ~$13,000/t. Q1 2026 average was ~$12,900/t per KGHM bulletin. Sustained $12,000+ quarter is functionally locked: even Goldman bear case ($11,000-11,200/t H2 2026) only happens if 15% US tariff is announced AND demand stays soft. Most-likely first sustained-quarter at $12K+: Q1 2026 (i.e., already done) or Q2 2026 ending June 30. Modal P50 = 2026."},{"slug":"copper-sustained-15k","p50":2028,"why":"Canonical tier. UBS targets $15,000 by Mar 2027 (Feb 2026 note); Citi bull case (30% prob) $15,000 avg year-end 2026; Goldman 2035 base case $15,000. Convergence of analyst bull cases points to 2027-2029 window. Requires (a) US tariff resolution clearing the LME-COMEX arb, (b) China demand recovery, (c) deficit widening as ICSG projects 150-330kt 2026 deficit. P50 2028 anchors midpoint of these bull cases."},{"slug":"copper-sustained-20k","p50":2030,"why":"Requires accelerated electrification + supply shock. ICSG / S&P deficits widen significantly post-2030 as mine output peaks. Tail-bull paths: combined Codelco production collapse (already trending), Peruvian political shift after April 2026 election, AI data center capex sustaining 165% growth to 2030 per Goldman. Citi 'just-in-case' inventory build to 3 months consumption alone would require ~$27,885/t. P50 anchors at decade-end."},{"slug":"copper-sustained-25k","p50":2032,"why":"Tail scenario combining AI buildout exceeding plans + EV inflection past 50% market share + mine supply collapse + sustained Chinese restocking. McKinsey 37Mt-by-2031 demand path + S&P 10Mt cumulative deficit through 2035 imply this tier becomes plausible only post-2031. Hyper-bull case: $27,885/t implied by Citi 3-month-inventory restock thesis. P50 = 2032 reflects this is a regime-change scenario, not a continuation."}]
history: [{"date":"2026-05-25T00:00:00.000Z","p10":2026,"p50":2028,"p90":2031,"why":"Initial estimate. Anchored to LME cash $13,545/t (22 May 2026 close, Westmetall). $12K tier is functionally already firing for Q2 2026 average; $15K tier converges to 2027-2029 across UBS/Citi/Goldman bull cases (P50 = 2028); $20K and $25K tiers require regime-change events (mine supply collapse, accelerated Chinese restock, AI-capex super-cycle) and anchor late-2020s to early-2030s. P90 2031 reflects Goldman's explicit '$15,000 by 2035' as long-run target — the canonical tier is reasonably likely to fire well before then, but a global recession or aluminum-substitution shock could push it out to early 2030s."}]
cross_gate: [{"other":"residential-solar-storage-0.04","relation":"enabled_by","strength":"strong","note":"Solar deployment requires massive copper for inverters, wiring, grid interconnects, BESS busbars. PV systems use 2-3x the copper of fossil alternatives per MW deployed; residential solar accelerates this conversion. IEA estimates copper for low-carbon energy systems grows from 7.9 Mt/yr in 2025 to 17.3 Mt/yr by 2050. Cheap residential solar storage IS the demand pull for the high-copper-intensity grid build-out."},{"other":"humanoid-retail-20k","relation":"correlates","strength":"medium","note":"Humanoid robots use ~5-10 kg copper each across motors (WRSM rotor windings doubles Cu vs PM motor), batteries, wiring, sensors. 10M+ household units = 50-100kt incremental annual demand (small but meaningful at the margin)."},{"other":"humanoid-10m-households","relation":"enabled_by","strength":"medium","note":"Volume humanoid deployment pulls copper through motor windings, charging infrastructure, and the household-level grid hardening required to support always-on robotic platforms. Same direction as the broader electrification thesis."},{"other":"ai-agent-30pct-knowledge-work","relation":"correlates","strength":"strong","note":"AI data center power demand is THE largest single new copper consumer in every analyst projection. BHP: AI data center copper grows 6x to 3 Mt/yr by 2050 (9% of global demand). Bloomberg: 1.1-2.4 Mt incremental NA demand by 2030. Morgan Stanley: 740kt data center copper by 2026. Copper intensity at 40-70 t/MW for AI factories vs 20-30 t/MW for traditional servers. Agent capability + deployment scales compute → power → copper."},{"other":"robotaxi-unit-economics-5-cities","relation":"enabled_by","strength":"medium","note":"EVs use 60-83 kg copper vs ICE ~23 kg (ICSG); robotaxi adoption accelerates EV fleet conversion + adds charging infrastructure copper demand. Benchmark forecasts EV copper demand >2.5 Mt by 2030 (vs ~1 Mt in 2024). 800V architectures + aluminum substitution thrift Cu intensity per vehicle, but volume growth dominates."},{"other":"metals-bom-30pct","relation":"competes","strength":"medium","note":"Aluminum substitution is real but limited (~10-20% of Cu use in batteries, busbars, wiring). Sodium-ion batteries eliminate Cu foil entirely. Recycling can supply ~35% of demand by 2050 (still leaves 65% for mining). If metals-BoM gate fires via recycling/substitution path, copper benefits less; if it fires via supply diversification (new mines, asteroid mining), copper deficit narrows directly."},{"other":"smr-first-oecd-deployment","relation":"correlates","strength":"weak","note":"SMR-powered data centers still need copper for site transmission + grid interconnect + cooling. SMR-on-site only shortens the transmission leg; site-level copper (40-70 t/MW) is the same. IEA: SMRs come online ~2030 to meet data center demand notably in China/Japan/US."},{"other":"commodity-natgas-ai-power","relation":"correlates","strength":"strong","note":"Same underlying driver: AI data center power demand. IEA: gas + renewables together meet >50% of incremental data center demand to 2035. Higher copper price and higher natgas price both reflect the AI-electrification-bottleneck thesis. If one fires, the other typically fires."},{"other":"commodity-uranium-smr-bull","relation":"correlates","strength":"medium","note":"SMR-powered hyperscale data centers need uranium AND copper (40-70 t/MW for the site). The bullish thesis stack is the same — accelerating AI compute demand exceeding incumbent grid capacity."}]
key_dependencies: [{"factor":"AI data center capex sustained growth","kind":"gate","direction":"accelerates","linked_gate":"ai-agent-30pct-knowledge-work","impact":"If AI agent gate fires, data center copper demand hits 1.4-2.3 Mt incremental by 2030, pulling $15K P50 forward ~2 years; stalling AI capex removes the single largest structural demand pillar and shifts P50 past 2030."},{"factor":"US Section 232 copper tariff magnitude","kind":"regulation","direction":"both","linked_gate":null,"impact":"Goldman base case 15% tariff compresses LME price to $11,200/t H2 2026, delaying $15K tier by 1-2 years; tariff cancellation unwinds ~150kt Comex inventory back to LME and tightens global supply, shifting P50 from 2028 to 2026-2027."},{"factor":"Codelco operational recovery or collapse","kind":"market","direction":"both","linked_gate":null,"impact":"Codelco audit revealed 26,875t overstatement and El Teniente runs reduced for >=5 years; failure to reach 1.7 Mt 2030 target removes ~400kt/yr and shifts $20K P50 forward to 2028; successful governance reform and output recovery shifts $15K P50 back toward 2030."},{"factor":"Chinese copper demand recovery or contraction","kind":"market","direction":"both","linked_gate":null,"impact":"China is 58% of global refined demand; a 5% contraction adds 1.5-2 Mt negative demand pull and pushes all tiers out 2-3 years (Citi bear case $10,000/t); restocking to 3-month inventory alone implies $27,885/t and would shift $20K P50 from 2030 to 2027."},{"factor":"Peruvian election political stability","kind":"event","direction":"both","linked_gate":null,"impact":"April 2026 election determines fate of $64bn project pipeline; pro-mining outcome unlocks Conga and adjacent projects within 5-7 years, shifting $20K P50 out past 2032; leftist nationalization or continued instability removes ~12% of global supply growth and shifts $15K P50 from 2028 to 2026-2027."},{"factor":"Residential solar storage deployment scale","kind":"gate","direction":"accelerates","linked_gate":"residential-solar-storage-0.04","impact":"PV systems use 2-3x copper per MW vs fossil alternatives; IEA projects copper for low-carbon energy growing from 7.9 Mt/yr to 17.3 Mt/yr by 2050; if solar gate fires on timeline, it adds the largest absolute demand bucket and shifts $15K P50 from 2028 to 2026-2027."},{"factor":"Aluminum and sodium-ion substitution pace","kind":"market","direction":"delays","linked_gate":"metals-bom-30pct","impact":"EV copper intensity could drop to 20-30 kg/vehicle (vs 60-83 kg) via aluminum wiring and sodium-ion batteries eliminating Cu foil; if metals-BoM gate fires via substitution path, demand growth slows by 30-50% and shifts $15K P50 out 2-3 years toward 2031."}]
external_calibration: {"metaculus":null,"manifold":null,"expert_consensus":"Goldman Sachs end-2026 $10,710 H1 avg / $11,200 H2 / $15,000 by 2035 (long-run); Citi base $13,000 / bull $15,000 EoY 2026 (30% probability) / Trump-resolution-scenario $14,423-27,885 for inventory restocking; UBS $15,000 by Mar 2027; Cochilco (Chile govt) 2026 avg $5.55/lb = $12,235/t (raised from $4.95/lb = $10,913/t); BHP 50Mt-demand-by-2050 base case; ICSG 150kt 2026 deficit (J.P. Morgan: 330kt; ING: 600kt); IEA 30% supply deficit by 2035; Red Cloud $6/lb avg by 2030 = $13,228/t; Bank of America $14,500 by 2028; Trading Economics Q4 2026 forecast $6.83/lb = $15,051/t.","market_size":"Global refined copper market ~28 Mt/yr × ~$13,500/t = $378 bn/yr. LME copper futures notional volume ~$80 trillion/yr. ETF AUM (COPX, JJC, CPER combined) ~$5 bn. Mining-equity sector (BHP/FCX/SCCO/FQM combined market cap) ~$400 bn at May 2026."}
last_updated: "2026-05-25T00:00:00.000Z"
sources_count: 22
---

## TL;DR

LME copper cash settlement closed at **$13,545/tonne on 22 May 2026** (3-month $13,600/t; Westmetall) [1] — already trading in the lower end of the bull-case zone the four-tier ladder probes. The Citi-Goldman-UBS bull-case mid-point converges on **$15,000/tonne as the next regime-defining level**, with timelines ranging from "average year-end 2026 in the bull scenario" (Citi, 30% probability) [4] to "by March 2027" (UBS) [13] to "by 2035 as long-run target" (Goldman) [14][15]. The asymmetric setup: structural deficit, 50-year-low LME inventories, ageing mine grades, geopolitical concentration (Chile + Peru = 40% of supply), and the AI-data-centre demand shock all push the same direction. The 4 tiers and modal P50s:

| Tier | Sustained-quarter ≥ | P50 year | Driver |
|---|---|---|---|
| $12,000/t | Q2 2026 (likely already) | 2026 | Current 3M roll already $13,600; bear case $11,200 |
| $15,000/t | 2027-2029 | **2028** (canonical) | UBS/Citi/Goldman bull-case midpoint |
| $20,000/t | 2029-2031 | 2030 | Mine output peaks ~2030 + AI-capex sustained |
| $25,000/t | 2031-2033 | 2032 | Tail: BoM-scale restock + Chinese SPR build |

The trigger is **sustained ≥1 quarter at LME cash settlement** (3-month rolling average) — not a single intraday touch. This is a stricter resolution standard that filters out short squeezes (e.g., the Trafigura LME warehouse withdrawal of 51,000+ tonnes on 22 May 2026 that drove cancellations to 1974 lows [11][12] is a single-week dislocation, not a sustained-quarter event). Of the four tiers, the $12,000 tier is effectively a base-case yes given current prices, the $15,000 tier is the central debate, and the $20,000 / $25,000 tiers price the AI-supercycle tail.

## Current state (24 May 2026)

**LME cash settlement**: $13,545/tonne 22 May; $13,600/t 3-month; LME stock 391,900t (down 1,200t day-on-day) per Westmetall [1]. KGHM internal report (13 May 2026): cash $14,097/t, monthly avg $13,453.69/t, annual avg $12,908.32/t [2]. The May 2026 daily range has been $12,895–$14,097/t — a wide ~9% trading band reflecting (a) the Iran-war / Strait-of-Hormuz dynamic dragging dollar bid + GDP fears, (b) tariff anticipation pulling LME stock into Comex warehouses, (c) Chinese restocking sporadically firing. The single most striking data point: **LME visible inventory is at a 50-year low** following Trafigura's 51,000-tonne withdrawal on 22 May (largest single-session withdrawal since 2013); cumulative 2-month drawdown >150,000 tonnes, total stock the lowest since 1974 [11][12].

**Technical context (per TradingKey 16 May 2026 analysis) [10]**: copper is consolidating below the Fibonacci 0.382 extension at **$14,233.50** (key short-term resistance). A clean break above opens the path to the 0.5 extension at **$15,000**. MA60 and MA144 bullish alignment intact; medium-term trend bullish. Major intraday high in 2026 was **~$14,100/t in mid-May**, just below the all-time-high of **~$13,310/t set in January 2026** (Comex; LME cash hit a fresh ATH later in Q1).

**Comex-LME spread**: Comex HG continues trading at a structural premium to LME cash equivalent — implied tariff embedded at 15-20% per Goldman analysis [5][6]. The end-June 2026 US Section 232 tariff decision is the next macro catalyst.

**Cochilco** (Chilean govt copper commission, 20 May 2026) raised its 2026 avg-price forecast from **$4.95/lb to $5.55/lb** ($10,913 → $12,235/t) and cut Chilean production to **5.3 Mt for 2026** (down 2-5% from prior estimate), citing lower ore grades, scheduled maintenance, operational constraints [16]. **Codelco** internal audit (May 2026) revealed 26,875-tonne 2025 production overstatement; new Kast-government board takes over 25 May 2026 [17][18]. Q1 2026 Chilean mine output fell 6% YoY [17].

## Demand drivers

**AI data centres** are the dominant new structural demand. Quantitative bracket:
- **BHP**: AI data center copper grows 6× from ~500kt (2024) to ~3 Mt/yr by 2050 (9% of total demand) [3]
- **Bloomberg NEF / BI**: Global incremental copper demand 2-2.5 Mt by 2030 from data centers; NA copper demand growth 3% CAGR through 2035 [7]
- **Morgan Stanley**: Global data center copper consumption reaches 740,000 tonnes in 2026; more than doubles by 2028 [10]
- **Goldman / IEA**: AI drives 165% increase in data center power demand by 2030; 945 TWh data center electricity by 2030, 1,200 TWh by 2035 base case [9]
- **CommodityNode disaggregation**: 1.4-2.3 Mt incremental AI-driven copper demand by 2030 across generation (375-750kt), grid T&D (600-900kt), facility itself (450-700kt) [20]
- **IEA Critical Minerals 2025**: AI data centers alone could consume 250-550kt/yr by 2030 (up to 2% of global demand) [19]

Copper intensity per MW for AI factories is **40-70 t/MW** (vs 20-30 t/MW for traditional cloud) — driven by liquid-cooled racks at 50-100 kW/rack, higher power-distribution copper, redundant power paths, and the on-site grid + cooling infrastructure [21]. The "50-tonne rule" (50 t/MW central estimate for AI builds) is becoming consensus.

**EV transition** adds ~2.5 Mt by 2030. Per-vehicle Cu intensity is declining from 99 kg (2015) → 62 kg (2030 forecast) via thinner foils (41→26 kg), wiring localization (30→17 kg), aluminum substitution in busbars/harnesses [21][22][23][24]. But unit-volume growth dominates: total EV copper demand projected to 2.5+ Mt by 2030 vs ~1 Mt in 2024 [22]. Note: International Copper Study Group 2018 estimate had a passenger BEV at **83 kg Cu** vs 23 kg for ICE — a 3.6× lift; the "EV = 4× copper of ICE" rule-of-thumb is reasonable for the 2024-2027 vintage fleet but understates as thrifting accelerates.

**Grid + renewables build-out** is the largest absolute demand bucket. McKinsey: transmission build-out pushes global copper demand to **~37 Mt/yr by 2031** [12]. IEA: copper for low-carbon energy systems grows 7.9 Mt/yr (2025) → **17.3 Mt/yr by 2050** [3]. A 1 GW wind turbine farm requires 2,866 tonnes of copper (India Economic Survey 2026); 1,194 truckloads of 400t to process the ore [25]. Solar PV requires 5-6× the copper of equivalent fossil generation per MW.

**Residential / commercial electrification** (heat pumps, induction cooktops, EV charging) adds gradual demand growth — incrementally ~150-200kt/yr of marginal Cu demand globally, dominated by retrofit + new-build copper wiring upgrades for elevated household electricity intensity.

## Supply drivers

**Mine pipeline is structurally thin.** Per BHP and S&P Global modeling:
- Global refined demand projected 27 Mt (2024) → 33 Mt (2035) → 42 Mt (2040) → 50 Mt (2050) [3]
- Mine output peaks around **2030** then declines as existing operations age out [3]
- $500-800 bn mining capex required through 2040 (IEA); BloombergNEF says $2.1 trillion across all energy-transition metals by 2050 [19]
- New copper mine **15-20 year lead time** from discovery to first production; project pipeline insufficient to close post-2030 deficit [3]

**Chile (~24% of global mined supply)** is structurally declining. Q1 2026 output -6% YoY; Codelco production -8% YoY [17][26]; Cochilco cut 2026 forecast to 5.3 Mt (-2 to -5%) [16]; new Kast-administration governance reform of Codelco underway with debt at $25.9 bn (vs. $972M new debt in 2025) [18]. Codelco 2026 target 1.344 Mt — only 10kt above the troubled 2025 print, and that's before the new board's renegotiation [27]. El Teniente will operate at reduced level for ≥5 years post-2025 fatal accident [26][27].

**Peru (~12% of global supply)** is on political ice. Nine presidents since 2016; interim president José María Balcázar installed Feb 18, 2026 [28]. April 2026 elections critical: $64 bn project pipeline frozen [28]. Newmont keeping Conga project in care-and-maintenance for ≥10 years [28]. Only Tía María (Southern Copper) actively advancing.

**Indonesia** (Grasberg force majeure) and **DRC** (Tenke / Kamoa political+infrastructure constraints) round out the supply-disruption picture; mine supply growth estimates fell to **1.4% for 2025** per ICSG, slowing further to 0.7% in 2026 [9][29].

**Recycling** caps at ~35% of supply by 2050 even under generous assumptions [22]. Citi "Call on Scrap" framework: every $1,000/t price rise → 150-200kt/yr additional scrap supply [4]. The 2024 Apr-May price spike triggered notable destocking; the question for 2026-2027 is whether another bullish leg can release fresh scrap or whether the bin is empty.

**Grade decline**: Cathles 2024 paper and India Economic Survey 2026 [25] both note that operating mine grades are now 0.5-0.6%; new projects 0.4-0.5%. Lower grades mean 2-4× more ore must be processed per tonne Cu — energy + waste-stripping costs scale punishingly. The marginal-cost-of-new-supply curve has steepened materially in the last decade.

## Per-tier reasoning

**$12,000/tonne sustained quarter — P50 2026**: Q1 2026 average price was $12,908/t per KGHM [2]; Q2 2026 to date averaging ~$13,000/t. Goldman's most bearish post-tariff forecast is $11,200/t H2 2026 [5] — and that's predicated on the US imposing a 15% tariff that diverts but doesn't destroy demand. Even Goldman's "fair value" of $11,100/t for 2026 [5] sits within the bull-quarter zone. **The $12K tier looks resolved-yes for either Q1 or Q2 2026** unless an aggressive bear case (Chinese collapse + recession + tariff shock combined) materializes by mid-2026. Modal P50 = 2026.

**$15,000/tonne sustained quarter — P50 2028 (CANONICAL)**: Three independent bull cases converge on $15,000:
1. **UBS (27 Feb 2026 note)**: raised target +$500/t across horizon, $15,000/t by end-March 2027 [13]
2. **Citi (May 2026 note)**: $15,000/t average year-end 2026 in 30%-probability bull scenario, contingent on Strait of Hormuz reopening + Chinese restocking + EV/AI demand acceleration [4]
3. **Goldman**: $15,000/t as 2035 long-run target ($11,500 in 2025 dollars) [14][15]

The price needed to incentivize 1.3-month → 2-month global Cu inventory restock per Citi is **$14,423/t** [4] — close to but below the $15K tier. For 3-month inventory restock, **$27,885/t**. Bank of America Dec 2025 target was already $14,500/t [20]. Trading Economics Q1 2027 forecast: $7.04/lb = **$15,520/t** [30]. The lower-bound case for $15K: Q4 2027 / Q1 2028 (UBS scenario). Upper-bound: Q4 2029. **P50 = 2028** anchors the analyst midpoint.

**$20,000/tonne sustained quarter — P50 2030**: Requires demand-supply gap widening past the 2030 mine-peak threshold + sustained AI capex + at least one major supply shock. Plausibility stack:
- ICSG / S&P deficit widens significantly post-2030
- 2.5 Mt EV demand by 2030 hits its full force (currently ~1 Mt)
- 1.4-2.3 Mt AI-driven incremental demand by 2030 [20]
- Codelco fails to reach 1.7 Mt 2030 target (analyst consensus already skeptical)
- Peruvian Conga / Yanacocha re-blocked by political shifts

Citi's "inventory restock to 3 months consumption" path would alone require **$27,885/t** [4]. P50 = 2030 reflects: tier is structurally plausible by decade-end, but a sustained quarter requires multiple drivers firing.

**$25,000/tonne sustained quarter — P50 2032**: Tail scenario. Requires (a) global cumulative deficit per Wood Mackenzie / S&P (~10 Mt 2025-2035) [29] forcing margin-cost-of-supply to step up sharply, (b) BoM-scale Chinese strategic reserve build pulling 1-2 Mt/yr off the market for 3-5 years, (c) AI-capex sustaining 30%+ data-center-power growth through 2030 (Goldman bull case), (d) at least one major supply collapse (Codelco operational implosion, multi-mine Chilean strike, Peruvian leftist nationalization). Modal P50 = 2032 reflects the regime-change character. Note that Goldman 2035 target of $15K (in 2025 dollars) doesn't imply $25K is impossible — it implies $25K nominal becomes plausible if (i) general USD CPI inflation runs hot + (ii) the supply-deficit scenario materializes. The two together push the nominal print materially above analyst long-run central forecasts.

## Counter-arguments / what would have to be wrong

1. **Chinese demand collapse** — Q4 2025 Chinese refined Cu demand was -8% YoY per Goldman; if China's property-sector deleveraging + EV-stimulus exhaustion compound through 2026-2027, the 58% of global demand from China could shrink fast. Bear-case 5% China demand contraction in 2026-2027 = 1.5-2 Mt of negative demand pull. Citi bear case (20%) targets $10,000/t [4]. This is the single biggest reason copper doesn't go to $20K.

2. **AI buildout slowdown** — if hyperscaler capex compresses on poor model ROI / energy permitting bottlenecks / regulatory backlash, the 1.4-2.3 Mt of AI-driven Cu demand may materialize at half speed. IEA "headwinds case" cuts 2035 data center demand 20% [9]. The CommodityNode bear case in the AI-copper thesis assumes Microsoft/Google/Meta/AWS pause $100bn+ of their committed $350bn capex [20].

3. **Aluminum substitution accelerates beyond consensus** — Tesla, GM, Porsche all shipping aluminum-wiring EVs; CATL/BYD/LG aluminum busbars; Octovalve cooling. CleanTechnica argues EV Cu intensity should fall to **20-30 kg/vehicle** (not 60-80) by 2030 [24]. Sodium-ion batteries (CATL, BYD volume-shipping in 2026) eliminate Cu foil entirely. Solar PV silver-thrifting precedent: high prices induced 19% demand destruction in 2026 per Metals Focus. If Cu prices stay above $13K/t for 2+ years, substitution at scale becomes economically rational; this is the structural copper-bear-case 5-year argument.

4. **Superconductor / room-temp breakthroughs** — true superconductors (whether Korean LK-99 derivatives revived, or room-temp under-pressure variants) would obsolete copper transmission in stages over 10-15 years. Low probability before 2030, but the tail is fat enough that consensus bull cases hedge by capping at $15K rather than $25K.

5. **US tariff resolution removes the speculative bid** — Goldman base case: 15% US tariff announced mid-2026 implemented 2027 [5][6]. If tariff is materially smaller / delayed / killed, the Comex/LME spread compresses, the US-warehoused inventory unwinds back into global supply, and 2026-2027 prices soften. Trafigura's 51K-tonne withdrawal [11][12] is the trade unwinding into the tariff event; post-decision, that 150kt could reverse.

6. **Cochilco surplus forecast** — Cochilco's own May 2026 report projects a modest 12,000-tonne refined surplus for 2026 (vs. 124kt deficit in 2025) [16]. ICSG October 2025 update flipped from surplus expectation to 150kt deficit, but the gap between ICSG and Cochilco shows the supply-demand balance is genuinely contested. If Cochilco proves right, near-term price ceiling is much closer to $12-13K than $15K+.

## Cross-gate dependencies

Three couplings dominate:

**Strong enable from `ai-agent-30pct-knowledge-work`** — The single largest new copper demand bucket is AI data center build-out. BHP's 6× growth to 3 Mt/yr by 2050; CommodityNode's 1.4-2.3 Mt incremental NA demand by 2030; Goldman's 165% data center power increase by 2030. AI capability driving compute scaling drives power drives copper. If the AI gate fires (>30% knowledge work agent-augmented by 2030), the copper $15K tier is very likely. If AI capex stalls, the copper bull case loses its biggest structural pillar.

**Strong enable from `residential-solar-storage-0.04`** — Cheap residential solar + storage requires grid integration, inverter copper, BESS busbars, EV charger backhaul. PV systems use 2-3× the Cu intensity of fossil per delivered MWh; IEA forecasts copper-for-low-carbon-energy grows from 7.9 Mt → 17.3 Mt by 2050 [3]. The residential solar gate firing means demand pull for the high-Cu-intensity grid build (already the largest absolute Cu demand). Critically: both gates share the same supply-pinch risk — if Cu goes structurally tight at $15K+, grid build slows, which forces solar deployment to plateau, which is a negative feedback on the residential-solar gate.

**Strong correlate with `commodity-natgas-ai-power`** — Same underlying AI-data-center-electrification driver; gas + renewables together meet >50% of incremental data center demand to 2035 per IEA. The two commodity gates fire together or stall together.

**Medium correlates**: `humanoid-10m-households` (5-10 kg Cu per robot × 10M+ units = 50-100 kt incremental), `humanoid-retail-20k` (motor windings), `robotaxi-unit-economics-5-cities` (EV fleet conversion), `commodity-uranium-smr-bull` (SMR-powered data centers still need site-level Cu).

**Substitution risk from `metals-bom-30pct`** — if the metals-BoM gate fires via supply diversification (new mines, asteroid mining, deep-sea mining), copper deficit narrows; if it fires via recycling + substitution (Al for Cu, Na-ion eliminating Cu foil), copper price rally is capped.

**Weak correlates**: `commodity-gold-upside` (monetary vs industrial thesis can de-couple), `smr-first-oecd-deployment` (SMR-on-site still copper-intensive on the grid side).

## Evidence and sources

1. [Westmetall — LME Copper Cash Settlement / 3-Month / Stock daily series](https://www.westmetall.com/en/markdaten.php?action=table&field=LME_Cu_cash) — 22 May 2026 cash $13,545/t, 3M $13,600/t, stock 391,900t; 14 May daily high $13,986/t. Accessed 2026-05-25.
2. [KGHM Polska Miedź — Daily Base Metals Bulletin 13 May 2026](https://kghm.com/sites/default/files/2026-05/2026-05-13%20EN.pdf) — Cash $14,097/t; 3M $14,140/t; monthly avg $13,453.69/t; annual avg $12,908.32/t; stock 398,550t. Accessed 2026-05-25.
3. [BHP Insights — How copper will shape our future](https://www.bhp.com/news/bhp-insights/2024/09/how-copper-will-shape-our-future) — 50 Mt copper demand by 2050 (+70%); 2.6% CAGR to 2035; existing mines declining; LRMC-based pricing methodology; data center copper demand grows 6× to 3 Mt/yr by 2050. Accessed 2026-05-25.
4. [Futunn / Citi — Commodity outlook bull case copper $15,000](https://news.futunn.com/en/post/67965263/citi-s-latest-commodity-outlook-in-a-bull-market-scenario) — Citi base $13,000/t 2026; bull $15,000/t year-end 2026 (30% prob); bear $10,000/t (20% prob); 2026 deficit 56kt; "just-in-case" 2-month restock = $14,423/t, 3-month = $27,885/t; AI data centers 150kt/yr incremental 2026-2027. Accessed 2026-05-25.
5. [ADVFN / Goldman Sachs — Copper outlook April 2026](https://uk.advfn.com/market-news/article/14020/goldman-trims-copper-outlook-amid-weaker-demand-but-keeps-long-term-bullish-stance) — 2026 avg $12,650/t (down from $12,850); 2026 surplus 490kt; long-term $15,000/t by 2035; fair value $11,100/t. Accessed 2026-05-25.
6. [Goldman Sachs Research — Copper to decline from record highs in 2026](https://www.goldmansachs.com/insights/articles/copper-prices-forecast-to-decline-from-record-highs-in-2026) — 2026 range $10,000-11,000/t; H1 avg $10,710/t; 2026 surplus 160kt; demand>supply from 2029; 2035 target $15,000/t ($11,500 in 2025 USD); 15% US tariff base case mid-2026. Accessed 2026-05-25.
7. [Bloomberg / BI — Copper demand set for data-center boost](https://www.bloomberg.com/professional/insights/commodities/copper-demand-is-set-for-data-center-boost/) — Global Cu demand +2 Mt by 2030 from data centers; 27-33 t/MW data center intensity; NA demand 3% CAGR to 2035; APAC + EMEA additional 800kt-1.3 Mt by 2030. Accessed 2026-05-25.
9. [IEA — Energy and AI special report (April 2025)](https://iea.blob.core.windows.net/assets/de9dea13-b07d-42c5-a398-d1b3ae17d866/EnergyandAI.pdf) — Data center electricity demand >2× to 945 TWh by 2030, 1,200 TWh by 2035 base case; range 700-1,700 TWh in 2035; first SMRs come online ~2030. Accessed 2026-05-25.
10. [TradingKey — Copper price forecast May 2026](https://www.tradingkey.com/analysis/commodities/more/261902344-copper-price-forecast-historical-highs-15000-tradingkey) — Fibonacci 0.382 resistance $14,233.5; 0.5 ext $15,000; MA60/MA144 bullish; Morgan Stanley 740kt 2026 data center Cu demand; Scotiabank 350kt 2027 deficit. Accessed 2026-05-25.
11. [Discovery Alert — Trafigura LME withdrawal 50-year low](https://discoveryalert.com.au/copper-lme-withdrawals-trafigura-comex-arbitrage-tariffs/) — 51,000+ tonne single-session 22 May 2026 withdrawal; 150,000+ tonne 2-month drawdown; LME stocks lowest since 1974; spot $13,660/t at withdrawal; +13% monthly price appreciation. Accessed 2026-05-25.
12. [Crypto Briefing / Bloomberg — Trafigura pulls 51,000 tons](https://cryptobriefing.com/trafigura-copper-lme-warehouse-withdrawal/) — Confirmation of 22 May withdrawal; Comex stocks +550% since Feb 2025 tariff probe; late-June 2026 US tariff ruling expected. Accessed 2026-05-25.
13. [Tradingpedia — UBS raises copper $15,000 by March 2027](https://www.tradingpedia.com/2026/02/27/copper-outlook-bullish-as-ubs-cites-limited-supply/) — UBS +$500/t across horizon; $15,000/t by Mar 2027; 2025 deficit ~200kt; 2026 deficit 520kt (raised from 407kt). Accessed 2026-05-25.
14. [Fool.com.au — Goldman Sachs copper $12,650 2026 / $15,000 by 2035](https://www.fool.com.au/2026/02/10/copper-price-forecast-for-2026-goldman-sachs/) — H1 2026 avg ~$13,000/t; end-2026 $11,200/t; long-term $15,000/t; 600 kt 2025 surplus (largest since 2009). Accessed 2026-05-25.
15. [Mining Reporters — Codelco 1.344 Mt 2026 target](https://www.miningreporters.com/noticia/news/2026/01/codelco-copper-output-2026-1344mt-up-10000-tonnes) — Codelco 2026 target 1.344 Mt (+10kt vs 2025); 2030 target 1.7 Mt; debt $24 bn; $5 bn annual capex; El Teniente fatal accident July 2025; 2025 output 1.33 Mt (declining-grades context). Accessed 2026-05-25.
16. [Mining Reporters — Cochilco cuts production raises price forecast](https://www.miningreporters.com/noticia/news/2026/05/chile-cuts-copper-production-tight-global-market) — 2026 Chilean production 5.3 Mt (-2 to -5%); 2027 5.5 Mt; 2026 price avg raised to $5.55/lb from $4.95/lb; global demand 28.2 Mt 2026 / 28.8 Mt 2027; 12kt surplus 2026. Accessed 2026-05-25.
17. [BNamericas — Codelco overestimated production by 27,000t](https://www.bnamericas.com/en/features/chile-codelco-overestimated-copper-production-by-almost-27000t-and-faces-output-declines-and-control-failures) — Internal audit 26,875t overstatement (2% of 2025 production); Q1 2026 Chilean output -6% YoY; Codelco Q1 -8% YoY; new board takes over 25 May. Accessed 2026-05-25.
18. [BNamericas — Codelco audits debts data tampering](https://www.bnamericas.com/en/analysis/chile-codelco-faces-audits-debts-and-revelations-of-copper-data-tampering) — Debt $25.957 bn / $972M new 2025; Kast governance reform incoming; privatization / holding model under consideration. Accessed 2026-05-25.
19. [The Deep Dive — IEA 30% copper supply deficit by 2035](https://thedeepdive.ca/iea-warns-copper-supply-to-face-30-deficit-by-2035/) — IEA: demand 27 Mt (2024) → 33 Mt (2035) → 37 Mt (2050); 30% deficit base / 40% aggressive climate; $500-800 bn mining capex needed through 2040; BloombergNEF $2.1 trillion all energy-transition metals by 2050; data centers 250-550kt/yr by 2030. Accessed 2026-05-25.
20. [CommodityNode — Copper AI moment](https://commoditynode.com/copper-ai-data-center-demand-surge/) — Late-Mar 2026 LME $11,840/t; AI Cu intensity 40-60 t/MW; 1.4-2.3 Mt incremental AI demand by 2030; cumulative deficit 9.7-10 Mt 2025-2035; Goldman $15,000/t by 2028; BoA $14,500/t. Accessed 2026-05-25.
21. [Skillings — Copper demand AI data centers 2026](https://skillings.net/copper-demand-ai-data-centers-2026-outlook-and-price-drivers/) — 50-tonne rule for AI builds (50 t/MW central); traditional 20-30 t/MW vs AI 40-70 t/MW; 2026 DC Cu demand ~300kt base / 400kt+ bull. Accessed 2026-05-25.
22. [AInvest — Copper structural deficit electrification](https://www.ainvest.com/news/copper-structural-deficit-electrification-reshaping-long-term-supply-demand-balance-2605/) — EV intensity 99 kg (2015) → 62 kg (2030); EV total demand 1 Mt (2024) → 2.5 Mt (2030); recycling caps at 35% by 2050; aluminum substitution ~2% of total use. Accessed 2026-05-25.
23. [Mining.com — Visualizing the decline of copper usage in EVs](https://www.mining.com/web/visualizing-the-decline-of-copper-usage-in-evs/) — Benchmark Mineral Intelligence; copper foil 41 kg → 26 kg; wiring 30 kg → 17 kg; aluminum substitution in busbars/wiring/charging cables. Accessed 2026-05-25.
24. [CleanTechnica — Electrification won't crash on copper](https://cleantechnica.com/2025/06/12/electrification-wont-crash-on-copper-debunking-latest-claims/) — Bearish-Cu case: EV intensity could drop to 20-30 kg; 800V architectures save 6-10 kg; aluminum wiring saves 10-15 kg; sodium-ion eliminates Cu foil entirely. Accessed 2026-05-25.
25. [Economic Times — India Economic Survey 2026 copper shortage](https://economictimes.indiatimes.com/industry/indl-goods/svs/metals-mining/economic-survey-2026-world-faces-copper-shortage-as-ai-data-centres-drive-power-demand/articleshow/127770289.cms) — 1 GW wind turbine = 2,866 t Cu = 1,194 truckloads ore; mine grades 0.5-0.6%; new projects 0.4-0.5%; volatility from Indonesia/DRC/Chile mine outages. Accessed 2026-05-25.
26. [Mining Reporters — Codelco weak January 2026 output](https://www.miningreporters.com/noticia/news/2026/03/codelco-output-drop-december-2025-production-surge) — January 91kt (4th lowest of decade); Dec 2025 172.3kt (highest of decade) under scrutiny; El Teniente reduced for ≥5 years per Codelco senior exec. Accessed 2026-05-25.
27. [Northern Miner — Codelco 20,000-tonne output gap](https://www.northernminer.com/news/codelco-faces-questions-over-20000-tonne-output-gap/1003891076/) — Audit suggests senior executive authorized inclusion of volumes failing finished-product criteria; March output 110.9kt (-10%); 1.7 Mt 2030 target now under question. Accessed 2026-05-25.
28. [BNamericas — Political stability key to Peru $64 bn mining pipeline](https://www.bnamericas.com/en/analysis/political-stability-key-to-unlocking-perus-us64bn-mining-project-pipeline) — Nine presidents since 2016; Balcázar interim from Feb 18 2026; April 2026 election; Conga in care-and-maintenance ≥10 years; only Tía María progressing; voter intention López Aliaga 14.6% / Fujimori 10.3% / 38.5% undecided. Accessed 2026-05-25.
29. [MacroAnalytix — AI copper supercycle 2026](https://macroanalytix.com/p/ai-copper-supercycle-2026-data-center-energy-bottleneck-commodity-deficit) — US DC demand 183 TWh (2024) → 426 TWh (2030); S&P 75.8 GW (2026) → 134.4 GW (2030) DC grid power; Goldman 165% DC power increase 2030; ICSG 2026 deficit 150kt; JPM 330kt; ING 600kt; S&P 10 Mt cumulative deficit through 2040. Accessed 2026-05-25.
30. [Trading Economics — Copper forecast 2026/2027](https://tradingeconomics.com/forecast/commodity...copper) — Q2/26 $6.34/lb; Q3/26 $6.63/lb; Q4/26 $6.83/lb ($15,051/t); Q1/27 $7.04/lb ($15,520/t). Accessed 2026-05-25.

## Cross-gate synthesis (decision-level)

The copper supercycle gate sits at the structural intersection of the project's four largest physical-AI capability gates: **ai-agent-30pct-knowledge-work** (data center power), **residential-solar-storage-0.04** (grid build copper), **robotaxi-unit-economics-5-cities** (EV fleet copper), and **humanoid-10m-households** (motor windings + household-grid hardening). If any three of these four fire on their stated timelines, the $15K tier resolves by 2028. If all four fire, $20K becomes plausible by 2030. The tail-bull $25K scenario requires a fifth driver: either Chinese strategic reserve build (geopolitical) or a Codelco-scale supply implosion (operational) layered on top of demand acceleration. Copper is the cleanest physical-AI proxy with the lowest substitution risk and the most analyst-consensus structural deficit.