Chinese miners are growing fastest, on the lowest cost base in the industry, out of orebodies Western majors sold them. This is not a story about state policy or subsidy. It is a story about equipment, ore grade, and capital cost; and about Western miners, when given the freedom, choosing Chinese mining equipment, too.
That is the price signal textbook commodity economics says should bring production online: rising prices, new mine development, more supply.
Almost every major producer responded.
Over the two years to the end of 2025, CMOC grew copper output 77% and Zijin 8%; Rio Tinto grew 42%, BHP 18%, Barrick 15%. Freeport fell 20%, most of it a September 2025 mud rush at Grasberg. Almost the whole industry is expanding into the price.
What differs is where the growth comes from and what it costs to lift. CMOC's 322 kilotonnes of new copper came out of Tenke Fungurume and Kisanfu, two orebodies it bought from Freeport, which now produces nothing from either. And it mines them at a fraction of what the same metal costs Barrick.
The West is running out of good rock (by Western standards)
Western majors are expanding where they already own tier-one rock — Escondida, Oyu Tolgoi — and essentially nowhere else. The marginal orebody, the one that sets the next decade of supply, does not work on their operating model.
Grade is falling globally. The production-weighted average copper head grade is down roughly 40% since 1991, and at the large mines that set the cost curve S&P Global puts it near 0.5%. Morenci in Arizona, Escondida in Chile, Grasberg in Indonesia — the Western flagship mines — are working progressively into the tail of their orebodies. Each tonne of copper requires more rock moved than it did a decade ago.
Discoveries have collapsed. Global new copper discoveries are down 80% from the 1990–2010 average — from ~50 Mt per year to ~8 Mt per year. Over 200 copper mines are expected to run out of ore before 2035. Without new discoveries, future production is capped by what is already known.
Building a new mine in the OECD takes 10–15 years. Greenfield copper or gold projects in the Americas, Australia, or Canada face permitting, litigation, environmental review, and Indigenous consultation stages that push first pour a decade or more from discovery. A mine commissioned in 2026 was explored in 2012. Rising prices today produce no new metal until 2040.
At 0.45% grade with 10-year development cycles and Western operating costs, many orebodies that still contain copper simply do not pencil out. The rock is there. The economics are not.
If you can't mine it, China can.
Chinese miners are running the same cycle, in many of the same countries, under the same physical geology — and the Chinese pair expanded copper output 28% over two years against 6% for the five Western majors combined. The difference is operational capability. Four things Chinese miners can do that Western miners have shown they cannot:
Operate at lower grade. A mine at 0.30% copper grade that is marginal for Caterpillar-equipped Western operators is profitable for XCMG-equipped Chinese operators. Julong, Zijin's flagship Tibetan copper mine, books ore down to a copper-equivalent cut-off of 0.17% against a reserve grade of 0.28% — it mills what a Western operator would send to the waste dump. This is a capability constraint, not a choice.
Operate outside the OECD. Zijin operates profitably in Tibet, Kazakhstan, Serbia, DRC, Kyrgyzstan, Ghana. CMOC runs two of the world's largest copper-cobalt operations in DRC. These are not impossible jurisdictions — Kazakhstan in particular has an orderly mining regulator and a stable currency. Western majors simply do not have the on-the-ground operating experience, supply-chain relationships, or financing structures to run mines there. The expertise left the industry in the 2010s consolidation wave, and nobody has rebuilt it.
Acquire and restart orebodies Western miners wrote off.
In each case the orebody did not change; the owner did, and the new owner is expanding production from rock the old one had written down.
Finance mine development at 2–3%. Western miners pay 5–7% on debt against 2–3% for Chinese resource financing, a spread that tracks the gap between US and Chinese policy rates. Over a 25-year mine life, 400 basis points of discount rate is worth 30–50% of project net present value.
The cost curve, in $/kg copper
The all-in sustaining cost (AISC) in mining is the cash cost to extract the resources — known as C1 in industry nomenclature — plus the maintenance capex required to keep the mine running.
Freeport's 2026 cash cost guidance is $3.86/kg. Barrick's is $4.85–$5.40/kg — $4,850–$5,400/t. At $13,198/t, that is a cash margin of roughly $8,070/t.
Zijin's C1 is $0.70/lb — $1,543/t — which leaves a cash margin of $11,655/t on the same metal at the same price.
Why is the cost so much lower?
The 60–70% delta comes from equipment, capital, power and plant design.
Equipment cost. Chinese OEMs — XCMG, SANY, LiuGong, Tonly — sell mining trucks at 40–50% below Caterpillar and Komatsu prices on comparable specifications. A Cat 797F: roughly $5M. An XCMG XDE400 at the same 400-tonne payload: roughly $2.5M (neither OEM publishes list prices, so both are trade estimates). Faster delivery, customisation included, local parts supply.
Capital cost. Debt at 2–3% against the Western 5–7%.
Self-supplied power. Chinese operators increasingly bundle mine development with power infrastructure — the Heshima Hydropower Project at CMOC's DRC operations, dedicated solar plus battery storage at Chinese domestic mines, industrial-scale nuclear in the planning pipeline. Western miners pay grid rates ($0.08–0.15/kWh) or diesel generator rates ($0.20–0.35/kWh) for remote sites. Over a mine life, self-supplied clean power reduces operating costs by 15–25%.
Throughput design. Chinese mines are designed for high throughput at low grade from day one. Julong Phase 2 processes 350,000 tonnes of ore per day. Western mines have legacy plant sized for the higher-grade ore the orebody used to contain, so their fixed costs spread across fewer tonnes of rock and land higher on every tonne of metal.
What is coming next — electric first, then autonomous
None of that cost gap yet includes electric or autonomous fleets, which are barely deployed at copper mines.
Zijin's $0.70/lb C1 — $1,543/t — is achieved with conventional diesel-electric fleets at Julong, Kamoa-Kakula, Čukaru Peki, and Bor. The full electric + autonomous + battery-swap + renewable-powered stack is currently deployed at coal, not copper. When it migrates to copper mines — which Chinese equipment makers are actively targeting — the Chinese cost curve drops another $400–$1,100 per tonne.
At 0.50% grade, both diesel and electric work; electric plus autonomy saves ~$557 per tonne of copper. At a 0.17% cut-off, diesel is economically impossible — and the electric + autonomous fleet saves ~$1,640 per tonne of copper.
Chinese miners are both investing in the technology and mining the ore where it pays out most: the ore Western operators have already written off.
The western fleets have 15–20-year replacement cycles. Diesel trucks purchased in 2024 will still be operating in 2044. There is no catch-up path that does not require writing off existing capital.
For Freeport, Barrick and Newmont, their most credentialed Western peer has already concluded that Chinese mining equipment is the better value. They can follow Rio Tinto and concede the procurement relationship to Chinese suppliers, or they can stay with Caterpillar and pay the cost-curve penalty indefinitely.
Caterpillar's response — retrofit versus ground-up
Caterpillar has 690 autonomous trucks in operation at end of 2024, a target of 2,000+ by 2030, a partnership with CRH for battery-electric trucks and multiple pilots. All of it on the wrong architecture.
Caterpillar retrofit: MineStar Command is an autonomy software and sensor layer added to existing truck designs. The chassis — 793, 797F, 794 AC — was designed for a driver. Cabin, HVAC, seat, climate control, pressurisation, dust filtration, sound insulation. The diesel engine remains. Electric trucks are a separate, earlier-stage development program. Two parallel roadmaps with unclear convergence.
XCMG ground-up: The ZNK95 was designed cabless from first principles. No driver compartment eliminates cabin structure, HVAC, pressurisation, dust filtration, seat, operator controls. Weight savings redirected to larger battery and more payload. Electric only — no diesel option muddles the engineering. Integrated from day one with the Huawei 5G-A network, cloud dispatch, and battery-swap stations.
A Cat 793 with MineStar autonomy: approximately $3.5M all-in. An XCMG ZNK95 with equivalent autonomy and electric drive: approximately $1.7M. Caterpillar can match XCMG on autonomy software — and likely will within 3–5 years. It cannot match them on system cost without a ground-up chassis redesign, which means writing off decades of chassis design IP.
The divergence
In 2026 the supercycle prints are historic: copper set an all-time high of $14,528/t on the three-month contract in January and was still settling above $13,000 in April, gold averaged $4,876/oz through the first quarter, and cobalt metal more than doubled over the course of 2025. Zijin's 2025 net income was $7.4 billion — a record, up 60%. CMOC's was $3 billion — up 50%. Their 2026 production is guided higher again.
Freeport's output fell and Barrick's gold with it; Newmont's gold barely moved. Unit costs rose at all three, and earnings growth lagged the commodity move.
The Western mining industry today looks like the solar industry of 2012, the battery industry of 2018, or the electric-vehicle industry of 2022. Each time the combination was the same: a cheap and powerful industrial base, electrification of the operating cost structure, autonomy eliminating the labour floor, and a lower cost of capital.
The marginal tonne of copper has moved east. Western operators are not incompetent; the industry moved on without them, onto a different chassis, a different power source and a different capital structure: a Chinese one.
The supercycle is real, but not for them.
Get the next post by email. No spam.
