Lithium Refining: Western Infrastructure & Production Asymmetries

Albemarle halted Kemerton’s lithium hydroxide plant, citing unprofitability. Kwinana refinery commenced production. Analysis of Western lithium refining infrastructure & production asymmetries.

Albemarle ferma Kemerton - Albemarle ferma Kemerton: 820 milioni per zero tonnellate di litio

Part I – Western Lithium Refining Infrastructure: Analysis of Production Targets and Information Asymmetries in Australia, Chile, and the USA

Huandroid Dossier – The Refining War (Part I of II)

Lithium refining in Australia, Chile, and the United States: A landscape characterized by information asymmetries regarding production targets

On February 19, 2026, Albemarle announced the complete suspension of operations at its lithium hydroxide plant in Kemerton, reducing the nominal capacity from 50,000 tons/year to zero. The company justified the decision with the unprofitability of production from hard rock, estimating closure costs of $820 million and forecasting further losses for 2027. The project, which included an expansion to 75,000 tons/year by 2029, has been canceled. Official reports do not provide the nominal capacity of the plant in idling mode, nor do they detail the specific technical or economic metrics underlying the suspension.

The Kwinana refinery, owned by Covalent Lithium, began producing lithium hydroxide on August 15, 2025, declaring a nominal capacity of 50,000 tons/year. According to industry sources, the plant completed commissioning in July 2025, reaching 100% of its nominal capacity in 18 months. A subsequent report in 2026 noted a delay in the ramp-up phase associated with odor-related issues. The company has not released public updates on the specific causes or a resolution timeline. The Australian government has allocated $30 million to support the maintenance of operational capacity for the industry, but public documents do not specify whether and to what extent these funds affected the Covalent Lithium project.

The Kwinana refinery operated by Tianqi Lithium and IGO, operational since May 2022 with a nominal capacity of 24,000 tons/year, achieved a performance of 35% of its capacity in 2025, producing 2,126 tons in the fourth quarter. IGO recorded an impairment of $70-90 million in 2025, linking it to operation below nominal levels, and recorded a 49% decrease in the value of its stake in 2024. The construction of the second plant (Train 2), intended to bring the total capacity to 48,000 tons/year, was suspended on January 1, 2025, due to the difficulty in reaching the nominal capacity of the first plant. No date for the restart of Phase 2 has been announced.

The Kwinana refinery project promoted by Wesfarmers and SQM, presented as distinct from the Covalent Lithium project, reported a 64% completion rate in February 2024, with commissioning scheduled for mid-2025. As of July 2025, the plant is operational by Covalent Lithium. Company communications do not clarify whether the two projects are operating in parallel or if there has been a substitution. The nominal capacity of the Wesfarmers/SQM project is not specified, and there are no official updates on the progress since 2024.

The PLS Group Limited project in Pilgangoora, with a declared nominal capacity of 3,000 tons/year of lithium hydroxide, has been operational since 2025. Public reports do not contain information on the progress or expansion plans.

Nominal Capacity and Progress of Construction or Expansion Projects (2025–2027)

Project Owners Nominal Capacity Progress Status Expected Start
Kwinana Refinery (Covalent Lithium) Covalent Lithium 50,000 tons/year Production started on 08/15/2025; nominal capacity reached within 18 months 2026
Kwinana Refinery (Tianqi/IGO) Tianqi Lithium, IGO 24,000 tons/year (Train 1) Construction of Train 2 suspended; operating at 35% capacity in 2025 Not disclosed
Kwinana Refinery (Wesfarmers/SQM) Wesfarmers, SQM Not disclosed Activated by Covalent Lithium (July 2025); Wesfarmers/SQM updates not available post-2024 2025
Kemerton Refinery (Albemarle) Albemarle 25,000 tons/year (Train 1) Operations suspended (February 2026); active capacity reduced to 0 Not disclosed
Pilgangoora Refinery (PLS Group) PLS Group 3,000 tons/year Operational since 2025 2025

Estimates indicate that Australian lithium hydroxide production should reach 564.78 kilotons of LCE equivalent by 2029, with a CAGR of 23.51% in the 2024-2029 period. Recent operational data—the suspension of operations at Albemarle in Kemerton and the under-capacity performance of Tianqi/IGO in Kwinana—highlight a discrepancy between these expansion targets and the volumes currently achieved. Government records related to the $30 million support package do not specify the exact allocation of funds among the various projects experiencing operational difficulties.

[INTELLIGENCE GAP]
Company reports omit crucial details. The technical reasons why Covalent Lithium claims to have reached nominal capacity in 18 months, while Tianqi/IGO reports a 35% yield, are not disclosed in the financial statements. Furthermore, there is no public documentation detailing any potential conditions of Australian government support linked to operational restructuring plans.

Analysis of public data confirms that the aggregated nominal refining capacity in Australia has not been achieved by major players in the 2025-2027 period, reducing chemical output compared to initial targets.

Lithium Refining Capacity in Chile: Announced Investments and Undisclosed Metrics (2025–2027)

Chile maintains a central role in lithium supply, with a significant volume of investments planned between 2025 and 2027. However, the nominal capacity metrics of plants under expansion or construction vary widely depending on the official sources consulted, and are often omitted altogether.

The NovaAndino Lithium joint venture (Codelco and SQM), operational since December 2025 in the Salar de Atacama, exemplifies this asymmetry. One government source attributes a nominal capacity of 210,000 tonnes of lithium carbonate per year to the plant, with a target of 300,000 tonnes per year by 2030. Other official communications confirm the start of operations but omit the capacity figures. The ownership structure provides for state control of 51% (through Codelco) until 2060, with ENAMI holding a stake in other projects.

The Salares Altoandinos project (Salar de Atacama – Aguilar, La Isla and Grande salars), led by Rio Tinto with a $3.425 billion investment, was selected by ENAMI. Project documents and the operational contract made public do not report the planned refining capacity. The start of operations is scheduled for 2032.

The Maricunga project, a Codelco-Rio Tinto collaboration, involves an initial investment of $900 million. Announcements mention an expansion, but do not specify the final product (carbonate vs. hydroxide) or the nominal capacity. The documents do not clarify whether the facility will include a refinery or be limited to extraction.

The Black Giant™ project by EnergyX in Antofagasta completed two phases in 2025, declaring a nominal capacity of 52,500 tonnes per year of lithium carbonate (99.9% purity) using Direct Lithium Extraction (DLE) technology. The data provided indicates a recovery rate of over 94% and water reuse of 85%.

Albemarle completed the validation of its DLE pilot plant in La Negra in December 2025, reporting recovery rates of >94% and water reuse of 85%, with investments of $30 million (pilot) and $216 million (salt recovery). The company has proposed a $3.1 billion DLE project in the Salar de Atacama. The nominal capacity of the future commercial plant has not been made public. Environmental impact assessment documents for Albemarle‘s DLE project do not specify whether the output will be carbonate or hydroxide. The nominal capacities of Albemarle‘s plants currently under expansion in Chile remain confidential.

SQM‘s reports indicate a production of 233,000 metric tonnes of LCE in 2025 (forecast: 230,000), attributing this output to NovaAndino Lithium SpA. The documents do not reconcile this actual production volume with the previously reported nominal capacity of 210,000 tonnes per year for the same plant.

According to SQM, the production capacity in the Salar de Atacama has been increased to 64,100 tonnes in 2025, with a target of 67,300 tonnes in 2026 for lithium carbonate. The reports do not distinguish whether these metrics refer to extraction or refining, nor do they indicate whether they represent nominal capacity or expected production.

The first Special Lithium Operation Contract (CEOL) between the Chilean government and ENAMI for the Salares Altoandinos does not contain public data on the nominal capacity. Institutional announcements frame the project within the strategy of state control, but omit details about the refining process included in the contract.

[INTELLIGENCE GAP]
Official Chilean communications present inconsistent metrics. The discrepancy between the declared nominal capacity for NovaAndino (210,000 t/a) and the reported actual production (233,000 t) is not accompanied by technical explanations. The institutional reluctance to disclose the nominal capacity in key projects such as Salares Altoandinos or Maricunga limits the traceability of the Chilean supply chain.

Off-take agreements related to Australian lithium refineries exhibit significant variability in disclosure levels. Details regarding volumes, duration, and the percentage of capacity covered are often inconsistent or absent from company announcements.

The agreement between Ford Motor Company and Liontown Resources (2022) involves the supply of 150,000 tons per year of lithium from Kathleen Valley starting in 2024, with a value of $545 million. The documents do not specify whether the lithium will be supplied in raw form or if it will be processed through a refining plant, and they do not define the percentage of the mine’s total capacity allocated to this contract.

Tesla has extended two agreements to 2026: Syrah Resources (8,000 tons per year of graphite, until March 16, 2026) and Magnis Energy Technologies (17,500 tons per year of AAM, with an option for 35,000). The announcements link the Syrah agreement to the production of active material in Kwinana, but do not specify the origin of the graphite (local or imported). The contract with Magnis does not contain explicit references to lithium or the production site.

Documents related to the Tianqi Lithium Kwinana project report a planned capacity of 24,000 tons per year (Phase I), with Phase II halted in 2025 at 50% completion. TLEA states a production of 24,000 tons per year in Kwinana, without clarifying whether the figure refers to current volumes or future projections. A $11 million Australian dollar contract awarded to SIMPEC in 2026 for work in Kwinana does not specify which phase of the project it refers to.

Cobalt Blue has announced letters of intent (non-binding) for 70% of the capacity of its cobalt refinery in Kwinana (3,000 tons per year), with customers in the USA, Japan, and France. The communications do not define any synergies with the lithium supply chain. An agreement for the PLS Mid-Stream plant in Kwinana with Ningbo Ronbay New Energy Technology omits specifying the material subject to the supply (lithium or cobalt).

The closure of the Albemarle refinery in Kemerton (February 11, 2026) eliminated 48,000 tons of nominal LCE capacity from the market. The plant, designed for 22,000 tons in 2026, did not reach its targets; the capacity had already been reduced to 25,000 tons per year in 2024 due to operating costs.

[INTELLIGENCE GAP]
The aggregated data shows a misalignment between declared capacity and contractual data. TLEA states a production of 24,000 tons per year in Kwinana, but the construction (Phase II) has been halted, and the SIMPEC contract (2026) does not clarify the scope of the work. Historical reports indicating 50,000 tons of lithium hydroxide produced in Kwinana in 2023 do not break down the data between finished product and semi-finished product.

Off-take agreements in Chile: discrepancies between official announcements and industrial operations

The analysis of contracts signed by US automotive companies reveals a diversification strategy aimed at suppliers in the USA, Australia, and Chile. However, the temporal and quantitative metrics associated with these contracts have limitations in terms of the level of detail disclosed.

Hyundai has entered into a long-term agreement with SQM for lithium hydroxide, based on an SQM production target of 100,000 tons per year by 2025. No official document indicates what percentage of this capacity is allocated to Hyundai, nor the exact duration of the contract.

BYD has signed an agreement with SQM Salar SA for 11,244 tons per year of lithium carbonate until 2030, in conjunction with a $290 million project for a cathode factory in Antofagasta. The industrial project is currently postponed. Company announcements cite the decline in lithium prices. Similarly, Tsingshan, which had a priority access agreement with SQM, has canceled a $233 million project for LFP production in Antofagasta. Chilean official communications do not report the current legal status of these contracts (whether they have been terminated or frozen).

Ford has two agreements: Albemarle (100,000 tons of lithium hydroxide over 5 years starting in 2026) and Nemaska Lithium (13,000 tons per year for 11 years). The documents related to the agreement with Albemarle do not specify whether the product will come entirely from Chilean plants. Albemarle has secured an increase in its production share (LME) of 240,000 tons per year in Chile; public records do not confirm whether this additional share is currently operational or in the authorization phase.

CleanTech Lithium holds a 40-year CEOL contract for the Laguna Verde project (153 km²). Public data do not report the start of commercial production. In March 2026, the Chilean government approved five additional mining contracts without disclosing the volume targets or the list of final buyers.

Table: Offtake Agreements and Declared Volumes (Chile)

Purchasing Company Supplier Volume (tons/year) Product Period Status of Related Project
Hyundai SQM Not disclosed Lithium hydroxide Not disclosed Not documented
BYD SQM Salar SA 11,244 Lithium carbonate 2023–2030 Postponed
BYD and Tsingshan SQM Not disclosed Lithium carbonate 2023–2030 Not realized
Ford Albemarle 100,000 (total 5 years) Lithium hydroxide 2026–2030 No details on origin
Ford Nemaska Lithium 13,000 Hydroxide + spodumene 2026–2036 Not documented
CleanTech Lithium Chilean State Not disclosed Laguna Verde 2026–2066 Undocumented production

The NovaAndino Lithium joint venture (Codelco and SQM, December 2025) reports a production target of 280,000–300,000 tons/year for the period 2031-2060. The joint venture documents and Chilean Supreme Court communications (which validated the agreement) do not contain information on commercial destinations or off-take contracts already signed for this capacity.

Off-take agreements in the United States: Diversification of supply and quantitative parameters

The analysis of contracts signed by US automakers highlights a diversification strategy aimed at suppliers in the USA, Australia, and Chile. However, the temporal and quantitative metrics associated with these contracts have limitations in terms of the level of detail disclosed.

Tesla completed the Robstown (Texas) refinery on January 15, 2026, declaring an investment of $1 billion and a capacity sufficient to support the production of 1 million electric vehicles per year. The documents do not include independent metrics on the energy efficiency of the acid-based process used. Tesla’s agreement with Piedmont Lithium provides for the supply of 125,000 tons of spodumene (2023-2025); the reports do not specify what percentage of this supply covered the needs of Robstown. The agreement with Liontown Resources (100,000 tons in 2024; 150,000 tons/year in 2025-2029) does not indicate the exclusive geographical destination of the material.

Ford aims to secure lithium for 2 million EVs by 2026. The agreement with Ioneer covers 7,000 tons/year of lithium carbonate (Rhyolite Ridge, Nevada) for 15 years. The documents do not specify whether the lithium will be supplied raw or if it will pass through a refining plant, and do not define the percentage of the total mine capacity allocated to this contract. In parallel, agreements with Albemarle (over 100,000 tons of lithium hydroxide, 2026-2030, origin USA/FTA) and SQM omit details on the annual allocation of volumes.

General Motors has an agreement to purchase 100% of the 40,000 tons/year of lithium carbonate produced by phase 1 of Thacker Pass (Nevada) for 15 years. The project, which includes a second phase for a total of 80,000 tons/year, is supported by a federal loan of $2.26 billion. The company has invested $650 million in Lithium Americas. The documents do not indicate whether GM has any commitments regarding the purchase of the output of phase 2, nor do they detail the allocation of the federal loan between mining and refining.

Other North American projects: Stardust Power has obtained environmental permits for a refinery in Oklahoma, without disclosing the capacity; Mangrove Lithium (Delta, BC) reports a capacity equivalent to 25,000 EVs/year, without mentioning any specific off-take agreements. Rio Tinto, with the $900 million invested in Maricunga (50,000 tons/year), has not communicated the commercial destination of the production.

Table: Lithium Off-take Agreements for Refineries in Selected Jurisdictions (Focus USA)

Company Project/Supplier Geographic Area Declared Volume/Capacity Details Provided
Tesla Robstown Refinery USA 1 million EVs/year (capacity) Supply coverage not specified
Tesla Piedmont Lithium Australia 125,000 tons (total 3 years) Deliveries 2023-2025
Tesla Liontown Resources Australia Up to 150,000 tons/year 2024-2029
Ford Ioneer (Rhyolite Ridge) USA 7,000 tons/year carbonate End of 2025 (5 years)
Ford Liontown Australia Up to 150,000 tons/year From 2024
Ford Albemarle USA/FTA 100,000 tons (total) hydroxide 2026-2030
General Motors Thacker Pass USA 40,000 tons/year carbonate 15 years (covers 100% of Phase 1)

The mapping of US off-take agreements confirms an industrial strategy oriented towards near-shoring and friend-shoring. However, the omission of details on the supply coverage rates and the exact ramp-up schedules of the plants bound by contracts limits the possibility of quantifying the share of the American industrial needs effectively detached from Asian control.

End of Part I – The dossier will continue with Part II.

Photo by Vardan Papikyan on Unsplash
Content generated and validated autonomously by multi-agent AI architectures.

> SYSTEM_VERIFICATION Layer

Verify data, sources, and implications through replicable queries.

Sources

I.

II.

III.

IV.

V.