Copper supercycle: AI data centers and energy buildout

One-line summary: AI data center expansion, EV charging, and power grid modernization are driving a 304K tonne structural copper deficit in 2026, with AI specifically projected to add 500K tonnes/year additional demand by 2030 — FCX and SCCO are the primary listed beneficiaries.

The insight

Copper is the conductive substrate for data center power distribution, EV charging infrastructure, and grid transmission. Unlike semiconductors (where demand can compress if a training run finishes), copper demand for physical infrastructure is durable: once a data center is built, its copper is in the building. The AI buildout thus creates a multi-year physical commodity demand that mining supply can't quickly match.

The chain

AI DC capex → 500kt/yr copper demand by 2030 + structural deficit + only 70% of 2035 demand covered by existing/planned mines → copper miner beneficiaries (FCX, SCCO).

Canonical: ai-capex-to-power-and-materials-cascade.

Evidence

• From 2026-05-11-autoresearch-macro-energy-critical-minerals-may-2026: Structural copper deficit of 304K tonnes in 2026. Current mine supply covers roughly 70% of projected 2035 demand — a gap that cannot be closed with permits and mine ramps inside a 5-year horizon.

• From 2026-05-11-autoresearch-macro-energy-critical-minerals-may-2026: AI data center buildout alone projected to add 500K tonnes/year of additional copper demand by 2030 — equivalent to approximately one full Escondida mine (the world's largest copper mine, ~1.2M tonnes/year) added purely from AI.

• From 2026-05-11-autoresearch-macro-semis-ai-infrastructure-may-2026: CSP CapEx of $830B in 2026 (+79% YoY) is the upstream demand signal. Every hyperscaler facility has copper-intensive power distribution and networking runs.

• From 2026-05-11-autoresearch-macro-energy-critical-minerals-may-2026: Mine project lead times are 15–20 years from discovery to production. Existing mines face declining ore grades. New supply is structurally constrained.

• From 2026-05-18-autoresearch-copper-supercycle-may-2026-update: A single major AI data center requires 40,000–50,000 tons of copper. With 50–100+ major AI DCs under construction globally in 2026, this represents 2–5M tons of incremental demand over the construction cycle — comparable to 8–20% of annual global copper production.

• From 2026-05-18-autoresearch-copper-supercycle-may-2026-update: FCX Grasberg mine incident (Q1 2026): copper sales volumes -72% YoY (82M lbs vs 290M lbs Q1 2025); unit cash cost Q1 2026 guidance was $2.60/lb vs $1.40/lb Q4 2025 baseline. SCCO 2026 guidance: 915K tons; $20.5B capex over 10 years; 5.3% CAGR to 1.6M tons by 2035.

• From 2026-05-20-autoresearch-fcx-grasberg-q2-2026-restart-copper-lme: FCX Q1 2026 actual result: unit net cash cost $1.91/lb — major beat vs. $2.60/lb guidance, driven by gold by-product credits from non-Grasberg operations. This is NOT operational recovery; Grasberg output remains collapsed. Q2 2026 unit cost guided $2.24/lb (elevated). Full-year guidance cut: 3.08B lbs (from 3.4B; 300M lbs cut); unit cost raised $1.95/lb (from $1.75/lb). Full capacity timeline: mid-2027 (not mid-2026 as prior guidance implied). See fcx entity page for detail.

• From 2026-05-20-autoresearch-fcx-grasberg-q2-2026-restart-copper-lme: Copper LME: $13,380/tonne ($6.07/lb) on May 18, 2026 — well above Goldman's $12,000 H2 "fair value" and $12,650 full-year forecast. ICSG (International Copper Study Group) now projects 150,000-tonne deficit in 2026, reversing its prior surplus call. Market pricing is consistent with ICSG deficit view, not Goldman's 490K tonne surplus model. LME net long positioning: 80th percentile.

• From 2026-05-20-autoresearch-fcx-grasberg-q2-2026-restart-copper-lme: Sulphuric acid shortage — secondary supply constraint: Hormuz closure + China export ban (May 1, 2026) → ~20% of global copper relying on SX-EW hydromet leaching faces input shortage. Chilean acid prices +44% in one month. FCX is protected (pyrometallurgical process at Grasberg); SCCO faces direct exposure (Chilean/Peruvian SX-EW operations).

• From 2026-05-18-autoresearch-copper-supercycle-may-2026-update: Secondary supply risk: China sulfuric acid export restrictions (byproduct of copper smelting) + Middle East sulfur disruptions — now confirmed and quantified in May 20 research above.

• From 2026-05-27-autoresearch-copper-supercycle-may-27-2026: ICSG abandons surplus forecast — now projects 150,000-tonne 2026 deficit. This is the first ICSG deficit call since 2009. ICSG moved from +209,000 tonnes surplus (October 2025) to -150,000 tonnes deficit for 2026. J.P. Morgan: 330,000-tonne shortfall. Goldman Sachs: 160,000-tonne surplus — now the isolated outlier. This resolves the key Goldman-vs-ICSG contradiction (640,000-tonne divergence) that previously limited conviction to medium. Conviction upgrade to medium-high now supported.

• From 2026-05-27-autoresearch-copper-supercycle-may-27-2026: Copper LME record: $6.65/lb ($13,650/tonne) on May 13, 2026 — new all-time record. Consolidated to ~$13,100–$13,400 range subsequently. Market is pricing the deficit consensus, not Goldman's surplus.

• From 2026-05-27-autoresearch-copper-supercycle-may-27-2026: FCX Grasberg Phase 2/3 restart: Initial ramp-up activities started end of March 2026 (slightly ahead of Q2 guidance); Block 1 deferred to 2027; ~85% normal production rates expected by H2 2026. Full-year 2026 guidance ~1.0B lbs copper (unchanged). Key catalyst: Q2 2026 earnings (July) — actual volume recovery + unit cost trajectory toward $1.75/lb baseline.

• From 2026-05-27-autoresearch-copper-supercycle-may-27-2026: China sulphuric acid export ban confirmed from May 2026 — not a risk flag but an implemented policy. DRC producers may curtail ~125,000 tonnes if supply chain delays extend through June. FCX (Grasberg: pyrometallurgical) fully protected; SCCO (Chilean/Peruvian SX-EW) faces direct margin compression. The acid crisis bifurcates FCX vs. SCCO risk profiles.

• From 2026-05-28-autoresearch-energy-critical-minerals-may-28: COMEX copper all-time high: $6.71/lb ($14,800/tonne equivalent) on May 13, 2026 — new all-time record. Driven by speculative buying from late 2025 + growing AI datacenter + power infrastructure demand expectations.

• From 2026-05-28-autoresearch-energy-critical-minerals-may-28: Pullback to $6.29/lb by May 27, 2026 (-6.3% from ATH). Drivers: profit-taking, China demand uncertainty, Goldman surplus framing (already in wiki). At $6.29/lb, FCX is above the $5.50-6.00/lb range that made the thesis attractive in early 2025 — fundamentally supported by structural demand story.

• From 2026-05-28-autoresearch-energy-critical-minerals-may-28: FCX Grasberg Block Cave phased restart confirmed beginning Q2 2026, on schedule. Deep Mill Level Zone and Big Gossan underground mines restarted late October 2025. Raymond James raised FCX price target on copper pricing update.

• From 2026-05-28-autoresearch-energy-critical-minerals-may-28: Causal chain refinement: Copper ATH reflects near-term speculative premium; structural demand story (AI DC + electrification) is intact at $6.29/lb. The ATH-to-pullback move does not change the thesis — it removes excess speculative premium.

• From 2026-05-30-autoresearch-energy-critical-minerals-uranium-helium-copper-nuclear: FCX Grasberg consolidated 2026 guidance ~3.1B lbs (down from 3.4B); district 2026 guided to ~1.0B lbs copper + ~0.9M oz gold (~35% below pre-mudslide pace). Full Block Cave recovery deferred to 2027–2029 (avg 1.6B lbs copper, 1.3M oz gold). FCX ~$65.69.

• lewis-hart in 2026-06-01-podcast-odd-lots-the-hidden-plumbing-of-commodity-finance: "I think part of the big story with compute, by the way, that's underfollowed, is copper. So I think everyone talks about the power — power is really important and there needs to be lots of power capacity in order to build the AI revolution. Copper is as important. And you see copper prices at record highs largely because of how much copper data centers are consuming because of all the electrical capacity that needs to be in that infrastructure." — Lewis Hart (Head of Corporate Advisory and Banking, Brown Brothers Harriman). Primary confirmation from an active commodity finance practitioner: the AI DC copper demand story is live and visible in commodity bank credit flows, not just analyst projections.

• From 2026-06-01-podcast-odd-lots-the-hidden-plumbing-of-commodity-finance: Strait of Hormuz trapped capital context: ~1,500 commercial vessels trapped; "tens of billions, maybe more than $100 billion" of capital stuck. Margin calls on hedges for trapped inventories straining commodity finance liquidity. Aframax vessel (700K barrels) financing cost: $40-45M → $70-75M overnight. System "functioning well so far" — companies came in well-capitalized from COVID + Russia-Ukraine experience. But "if the strait doesn't open sooner, you could see big strains."

• From 2026-05-30-autoresearch-energy-critical-minerals-uranium-helium-copper-nuclear: SCCO Tía María exploitation permit reinstated (Peru) — major long-standing regulatory hurdle cleared; project optionality unlocked. Q1 2026 copper mined production -4.0% YoY on lower ore grades; full-year 2026 guide 915,400 tonnes. Capex $441.9M Q1 (+39% YoY). SCCO +4.00% May 26, 2026.

• New leg — concentrate scarcity hands pricing power to miners (TC/RC collapse). From 2026-06-09-autoresearch-copper-supercycle-tc-rc-collapse-refresh: the 2026 copper concentrate TC/RC benchmark settled at $0/tonne (record low; Chinese smelter capacity outstripping concentrate — Antofagasta agreed zero charges). FCX is breaking from the benchmark it set for decades, pricing 2026 concentrate on a floor-cap system with floor and cap above the 2025 $21.25/t benchmark (7–10 smelters); SCCO's integrated smelting sidesteps third-party TC/RC entirely. This is a miner-specific value-capture channel distinct from the refined-copper price. Disruptions at Grasberg (FCX-operated) + Kamoa-Kakula tighten supply further. Copper pulled back to ~$6.3/lb off its June 2 record $6.6/lb (Fed-hike repricing + Mideast risk) — a more constructive entry. Contested near-term balance: ICSG sees a 150kt 2026 deficit; a competing view sees a tariff-driven surplus (US demand −6%).

• Practitioner corroboration of the supply-bottleneck magnitude (2026-06-10). dan-dreyfus in 2026-06-10-podcast-all-in-podcast-dan-dreyfus-america-s-critical-minerals-crisis-is: "Over the next 18 years, we're going to need 700 million tons of copper… as much copper as we mined in the last 10,000 years," and on the supply side "we're going to need five world class mega tier one mines coming online every single year… you can count on one hand and have some fingers left over the number of tier one mines that are coming on between now and the end of the decade." Chamath Palihapitiya, same source: "I thought the best performing asset was going to be copper… And that's before I talked to Dan." An active commodities investor independently sizing the same structural deficit the wiki tracks — reinforces the medium-high conviction; tradeables unchanged (FCX, SCCO). Dreyfus also frames a distinct silver bottleneck (see silver-depletion-to-solar-supply-squeeze) and a grid-capital-undersizing leg (see ai-capex-to-power-and-materials-cascade).

Names and exposures

Why copper rather than other commodities

• Copper is the one commodity where the AI buildout demand is incremental and durable (physical plant, not consumable).
• Uranium (nuclear) is a separate thesis with a different supply dynamic — see nuclear-baseload-for-ai-data-centers.
• Rare earths are a geopolitical story as much as a demand story — see us-critical-mineral-independence.
• Copper's deficit is structural, not cyclical: the 304K tonne 2026 gap isn't macro-sensitive; it's a mine-supply problem.

Contradictions / tensions

• Goldman Sachs still sees surplus (160kt, May 2026) — now the isolated outlier vs. ICSG (150kt deficit) and J.P. Morgan (330kt shortfall). Goldman's surplus view rests on China demand weakness overwhelming AI infrastructure demand. If Goldman revises to deficit, that would be a major catalyst (shift from outlier to consensus). From 2026-05-27-autoresearch-copper-supercycle-may-27-2026. [RESOLVED: Goldman-vs-ICSG contradiction no longer a ceiling on conviction — ICSG flipped to deficit; Goldman is isolated]
• ICSG flip (May 2026): ICSG moved from 209kt surplus (Oct 2025) to 150kt deficit for 2026. This is the first ICSG deficit call since 2009. The prior Goldman-vs-ICSG 640kt divergence is now Goldman-vs-ICSG+JPM — Goldman holds the contrarian bearish position. Conviction upgraded from medium to medium-high. From 2026-05-27-autoresearch-copper-supercycle-may-27-2026.
• From 2026-05-20-autoresearch-fcx-grasberg-q2-2026-restart-copper-lme: FCX Grasberg full capacity pushed to mid-2027 (not mid-2026); volume guidance cut 300M lbs. FCX Q1 cost beat ($1.91/lb) was gold-credit-driven, not operational — Q2 still guided $2.24/lb elevated. This is FCX-specific, not thesis-level.
• Copper is cyclical and extremely sensitive to China real estate and industrial demand — a China slowdown could overwhelm the AI demand signal in the short run.
• The 2030 projection (500K tonnes from AI) is analyst-modeled and may prove too aggressive if hyperscaler CapEx normalizes.
• AI DC copper demand is multi-year construction-phase (2026–2028); if construction delays persist (per electrical bottleneck data), copper demand from AI DCs may also be deferred.
• FCX and SCCO operate in jurisdictions with geopolitical risk (Indonesia, Peru, Mexico) — operational disruptions affect production more than macro demand.

What would weaken this thesis

• China economic stimulus dramatically expands Chinese copper mine output (they are the world's largest producer)
• Major recycling breakthrough dramatically increases secondary copper supply
• AI buildout stalls — CapEx consolidation leads to fewer greenfield data centers in 2027–2028

Valuation snapshot

Last refreshed 2026-07-20 (pre-open; marks are the Friday 2026-07-17 close, markets closed over the weekend). Price fills tagged twelvedata. Mkt cap / Fwd P/E are not in the Twelve Data free tier and were not re-sourced this run. The prior stamp was malformed (a 07-16 header concatenated onto an un-cleared 06-10 note) and the "Near 52w high" read on FCX was wrong — it has been corrected below.

Forward-looking outcomes (12-month)

⚠ Correction, 2026-07-17. The prior versions of these cases were internally inconsistent with this page's own evidence: they modelled copper at "$12,000+/ton" (bull), "$10,700–$11,500" (base) and "$9,000–$10,000" (bear) against evidence on this same page recording LME at $13,380/t on 2026-05-18 and a COMEX ATH of $6.71/lb (~$14,800/t) on 2026-05-13, and they cited an FCX "full-year 3.7B lbs" that had already been cut to ~3.08B. The price bands and volumes below are re-based to the page's cited evidence.

Bull case — the deficit consensus holds and Grasberg recovers into it: ICSG's flip to a 150kt 2026 deficit — its first deficit call since 2009 — plus JPM's 330kt shortfall leaves Goldman's surplus isolated (2026-05-27-autoresearch-copper-supercycle-may-27-2026). Grasberg Phase 2/3 ramp began end-March, slightly ahead of guidance, with ~85% of normal rates expected by H2 2026. Add the miner-specific TC/RC channel: the 2026 concentrate benchmark settled at $0/tonne and FCX broke from the benchmark it had set for decades, pricing on a floor-cap system with both floor and cap above the 2025 $21.25/t (2026-06-09-autoresearch-copper-supercycle-tc-rc-collapse-refresh). Copper re-tests and holds above the May ATH as AI DC demand at 40–50K tons/facility materializes. Implied price: FCX +40–60%; SCCO +25–35%. Cited: 2026-05-27-autoresearch-copper-supercycle-may-27-2026, 2026-05-18-autoresearch-copper-supercycle-may-2026-update.

Base case — copper holds the ~$6.2–6.6/lb ($13,100–$14,500/t) range it has consolidated into; Grasberg recovers on the revised timeline, not the original one: FCX earns through the incident on ~3.08B lbs full-year (guidance cut from 3.4B) with full Block Cave recovery deferred to 2027–2029, not mid-2026 (2026-05-20-autoresearch-fcx-grasberg-q2-2026-restart-copper-lme, 2026-05-30-autoresearch-energy-critical-minerals-uranium-helium-copper-nuclear); SCCO executes its 915K-tonne guide but wears the acid-cost drag. Implied price: FCX +15–25%; SCCO +10–15%. Cited: 2026-05-18-autoresearch-copper-supercycle-may-2026-update.

Bear case — Goldman's surplus call proves right and the speculative premium unwinds: copper retraces toward $5.00–5.50/lb as China industrial/real-estate weakness overwhelms the AI DC signal; Grasberg's recovery slips further against the already-deferred 2027–2029 Block Cave timeline; FCX's unit cost stays elevated (Q2 guided $2.24/lb vs a $1.75/lb baseline, and the Q1 $1.91/lb "beat" was gold-by-product-credit driven, not operational — 2026-05-20-autoresearch-fcx-grasberg-q2-2026-restart-copper-lme). SCCO additionally compresses on SX-EW acid costs. Implied price: FCX −25–35%; SCCO −20–30%. Cited: 2026-05-27-autoresearch-copper-supercycle-may-27-2026 (Goldman surplus call).

Currently undervalued vs base case? FCX: Marginal — downgraded from "Possibly," and the reason is that the old answer was conditioned on a Grasberg timeline that has since slipped. The prior read ("Possibly — if Grasberg restart completes Q2 and LME holds above $10,700") is doubly stale: copper never went near $10,700 (it is roughly $13,100–13,400) and full Block Cave recovery has been deferred to 2027–2029, with FCX's own full-year guide cut 300M lbs. At −19.0% from the high the price has partly discounted this. The structural demand case is genuinely strong and independently corroborated by practitioners — dan-dreyfus in 2026-06-10-podcast-all-in-podcast-dan-dreyfus-america-s-critical-minerals-crisis-is: "Over the next 18 years, we're going to need 700 million tons of copper… as much copper as we mined in the last 10,000 years" against "five world class mega tier one mines coming online every single year" that do not exist. But that is a decade thesis, and the 12-month FCX read is dominated by Grasberg execution, which keeps disappointing. SCCO: Marginal, and no longer "more straightforward" than FCX — that prior framing predates the acid bifurcation. SCCO's integrated smelting sidesteps the TC/RC collapse (a genuine edge), but its Chilean/Peruvian SX-EW operations take direct margin compression from China's acid export ban that FCX's pyrometallurgical Grasberg does not. The two names are now bifurcated on risk, not ranked.

Catalyst path:

• FCX Q2 2026 earnings (late July) — Grasberg ramp volume against the ~85%-of-normal H2 target, and whether unit cost moves back toward the $1.75/lb baseline from the $2.24/lb Q2 guide. The nearest gate.
• Copper price through August — the direct test of Goldman-vs-ICSG. A sustained break below ~$6.00/lb is the bear trigger; holding $6.30+ keeps Goldman isolated.
• Chinese sulphuric acid export ban duration — the SCCO-specific margin variable, and the cleanest thing separating the two names.

Related

• us-critical-mineral-independence
• nuclear-baseload-for-ai-data-centers
• tsmc-capacity-shortfall-and-pricing-power
• ai-capex-to-power-and-materials-cascade