China's primary aluminum output reached 37.08 million tonnes in 2020, up 4.9% year on year, reversing a 0.9% decline in 2019 and setting a fresh national record that re-anchored the global supply curve around Chinese smelters [S6]. That single national figure is larger than the combined annual output of the next several producing countries, and it is the reference point every specifier and buyer should hold when reading capacity tables.
Country-level capacity is unevenly distributed and concentrated in a handful of large integrated smelters. Brazil's Albras (Alumínio Brasileiro S.A.) sat at 460,000 tonnes/year nameplate in 2017, unchanged from 2016 and matching the long-run median of 406,000 t/yr across 33 years of CEIC observations back to December 1985 [S5]. For buyers mapping long-term supply risk, the contrast between China's millions-of-tonnes flexibility and single-asset producers like Albras defines the reliability profile of each origin.
How Country Capacity Is Reported and What the Numbers Actually Cover
Primary aluminum capacity is published in tonnes-per-year of nameplate, not actual output, and utilization rates shift with power prices, curtailment orders and grid availability. Brazil's Albras data series on CEIC runs from December 1985 onward with 33 observations, a 406,000 t/yr median, and a 460,000 t/yr figure that held flat across 2016 and 2017 [S5]. That kind of stability is rare — most national capacity tables show step-changes tied to greenfield potline additions or multi-year curtailments, and the 2020 Chinese record is one such step-change: output rose 4.9% in a single year, with spot prices climbing from roughly 12,000 yuan/t in H1 2020 to about 16,000 yuan/t by year-end [S6].
For a specifier, the practical distinction is between nameplate and effective capacity. Nameplate is the design throughput of the pots under standard amperage and current efficiency; effective capacity is what a smelter can actually deliver given power-availability agreements, environmental caps, and maintenance schedules. China hit its 37.08 Mt 2020 output by leaning on higher prices to restart curtailed pots — a reminder that the same nameplate envelope can swing 5-10% on price alone [S6]. The same dynamic applies in reverse: when power tariffs rise or carbon rules bite, nameplate stays constant but effective output drops.
Country-Level Snapshot: China, Brazil, and the Mid-Tier Producers
China is the structural anchor of global primary aluminum supply, accounting for the majority of the world's output and capable of swinging millions of tonnes on policy and price signals. Its 2020 record of 37.08 Mt was paired with a price recovery from approximately 12,000 yuan/t in H1 2020 to roughly 16,000 yuan/t later in the year, an effective +33% price move that flipped smelter behaviour from curtailment to restart [S6]. For downstream processors buying billet, slab or P1020, the policy lesson is that Chinese domestic price bands gate export availability, and the 2026 aluminum price band reference is built on top of that primary-market signal.
Brazil sits in the second tier with concentrated ownership. Albras's 460,000 t/yr nameplate in 2017 is a single-asset number — losing a potline at Albras is a national-level event in Brazil, whereas a comparable disruption in China is absorbed across dozens of smelters [S5]. The other major producing countries (India, Russia, Canada, Australia, the UAE) cluster in similar single-digit-million-tonne ranges, and the global ex-China total historically runs in the 25-30 Mt/yr band, meaning China's 37.08 Mt alone is materially larger than the entire rest of the world combined [S6].
Selection Criteria for Specifiers: Origin, Power Mix, and Carbon Load
Origin dictates more than logistics. Smelters in hydro-heavy grids (Canada, Norway, parts of Brazil, Iceland) typically score better on Scope 2 carbon intensity than coal-heavy grids, and that differential now flows through to buyer specifications under EU CBAM and similar mechanisms. Effective capacity — what a smelter actually delivers in a given quarter — is governed by grid availability, water levels for hydro, and curtailment orders tied to environmental or pricing policy. The 2020 Chinese pattern — output rising 4.9% on a +33% price move from 12,000 to 16,000 yuan/t — is the canonical example of price-elastic primary supply [S6].
Specifiers evaluating sourcing should also consider the downstream cluster around the smelter. A country with large primary capacity but limited casthouse capacity exports most of its metal as P1020, forcing the buyer to remelt; a country with integrated billet, slab and value-added rolling or extrusion allows the buyer to skip the remelt step. For context on how downstream specifiers pair primary metal with forming equipment, see the aluminum die casting machine cluster guide and the broader aluminum alloy reference for grade selection downstream of the smelter decision.
Comparison: China vs Brazil vs Mid-Tier Producers on Four Decision Criteria
On a four-criteria basis, the country choices line up as follows. Output scale: China 37.08 Mt (2020) dwarfs all others; Brazil's Albras alone sits at 460,000 t/yr (2017) [S5][S6]. Capacity stability: Brazil's Albras held flat at 460,000 t/yr across 2016-2017 with a 33-year median of 406,000 t/yr, while China's effective output swung on the 4.9% step-change in 2020 [S5][S6]. Price elasticity: Chinese smelters demonstrably responded to a 12,000→16,000 yuan/t price recovery with a 4.9% production increase, the textbook definition of elastic primary supply [S6]. Downstream integration: mid-tier producers typically ship P1020 only, while Chinese clusters integrate billet, extrusion and finishing — relevant to buyers who want to skip the remelt step.
These four criteria — scale, stability, price-elasticity, downstream integration — are the axes a procurement or design engineer should run country choices against. They also explain why a specifier might split orders: hydro-origin metal for carbon-sensitive end-uses, Chinese integrated supply for cost-driven volume, and a Brazil- or India-origin allocation for geographic diversification. For forming-process buyers, the upstream decision feeds directly into the aluminum window and door profile reference and the aluminum veneer panel reference, where billet and sheet stock specifications begin.
Use Cases: Who Needs Country-Level Capacity Data, and Why
Three buyer profiles lean on this data hard. First, primary-metal traders and large converters who buy P1020, T-bar or sows on annual contracts and need to forecast allocation risk — a 5% swing in Chinese output is a 1.8 Mt market move, large enough to move global LME and SHFE curves [S6]. Second, OEMs serving automotive, packaging or construction under CBAM or similar carbon-border rules, who need to document the carbon intensity of the smelter origin to defend their import declarations. Third, extruders and die casters planning capex — a new casthouse or press line only pencils out if feedstock allocation from one or more primary sources is contractually secured, which is itself a function of national capacity concentration.
Smaller buyers — job-shop fabricators, custom extruders, architectural finishers — generally work downstream of these decisions via their billet or sheet supplier, but they still need to read the same data because their supplier's allocation is governed by it. The gas aluminum melting furnace reference and the aluminum die casting machine reference are the next-link process steps for any of these downstream buyers, and the upstream country decision flows into charge-stock specification for both.
Limitations, Failure Modes, and Sourcing Risks
The single biggest failure mode in country-level capacity data is treating nameplate as deliverable. Brazil's Albras 460,000 t/yr figure is nameplate; the 33-year median of 406,000 t/yr is closer to a realistic run-rate assumption, and the gap between the two is the curtailment and maintenance envelope the buyer should price in [S5]. The second failure mode is ignoring policy: the 2020 Chinese output record came on the back of price signals, but the country has also moved to cap primary capacity at roughly 45 Mt/yr under its dual-control energy policy — a cap that can swing effective output independently of price and outside any LME curve.
A third risk is single-asset concentration. With Albras at 460,000 t/yr nameplate representing a meaningful share of Brazilian output, a forced outage at one potline is a national-level event; a comparable outage in China is absorbed across dozens of smelters [S5][S6]. For buyers running just-in-time contracts, that concentration risk argues for either geographic diversification or for safety-stock thresholds tied to the smelter's typical outage pattern rather than to national output averages.
Sourcing Channels, Standards, and Trackable Signals
Two channels dominate. Direct mill or smelter allocation (annual contract, often with formula pricing tied to LME/SHFE plus regional premium) is used by large converters and traders; the Made-in-China platform and Chinese supplier directories index primary smelters and downstream mills for international buyers [S1]. Indirect sourcing through stockists, regional distributors or trading houses covers smaller buyers and typically carries a premium of 2-6% over direct allocation depending on grade, size and delivery window.
Trackable signals to watch for the next 12 months: Chinese provincial dual-control energy policy implementation, which can cap or release primary capacity outside price signals; CBAM Phase 2 reporting thresholds, which will start pulling hydro-origin premium into EU import declarations; and any restart or curtailment of single-asset producers like Albras that move their national totals materially away from the 460,000 t/yr 2017 baseline [S5][S6]. Each of these is a verifiable input to the next allocation-cycle decision.