The Kazakh Plant Lindian Is Buying Files Its Rare-Earths Line at 600 Tonnes of Oxide a Year. Its Feed Plan Delivers 6,875.

The claim#

On 3 March 2026 Lindian Resources announced that a joint venture — Lindian 51%, an “in-country partner” 49% — would acquire 100% of the SARECO hydrometallurgical plant in Stepnogorsk, Kazakhstan, for US$15m. The announcement describes the asset in the present tense: the facility “is already constructed, operational and retains its core processing infrastructure”. And again, two pages later: “It is fully operational with existing workforce transitioning to the JV entity, permits, logistics, and supply chains established.”

The plant is the downstream half of the company’s story. Stage 1 of the Kangankunde mine in Malawi is to supply “~12,500 tonnes per annum Monazite Concentrate” to the JV, which will convert it to mixed rare-earth carbonate — a higher-value product than concentrate — at “approximately 92% TREO and ~97% NdPr” recovery, with the plant “fully operational within ~9 months (Q4 2026)”. The announcement prices the alternative at “over half a billion dollars typically required” and multi-year construction. The purchase structure, as announced in March, was US$3m initially and US$12m deferred until three months after “efficient plant operations at the Processing Facility (commercially producing MREC)” — a milestone the binding-terms table dates “~1H 2027” — while the announcement’s title promises “High Value MREC Production By Q4 2026”. Read together, the company’s timeline is: production beginning in Q4 2026, and commercial operation in the half-year after. Both dates matter here, because each sits on the near side of the permitting question this paper is about.

Twenty-nine days after the announcement, Lindian raised A$100m — 133.33 million new shares at A$0.75, benchmarked in the announcement four ways (a 3.2% discount to the 10-day VWAP through an 18.0% premium to the 45-day) — with the stated uses of proceeds including the plant’s “advancement of downstream processing capability, operational readiness”. The placement announcement describes the raise as establishing “a fully funded, debt-free pathway to first production and cash flow” across Stage 1 and the plant.

Then, on 10 August 2026, the transaction changed shape. Lindian announced a definitive Sale and Purchase Agreement for 100% of the plant, in place of the 51% joint-venture interest announced in March, “[f]ollowing comprehensive legal, tax, technical and environmental due diligence”. The consideration changed with it: the seller receives the Tenge equivalent of up to US$20m in cash — against the US$15m indicative price of March, the difference attributed to additional warehouses, buildings and land identified in due diligence — and RA Group, which in March was to fund US$7.35m of the purchase for its 49%, instead exits for up to US$22m in Lindian equity: US$15m in shares on completion and title transfer, US$5m in performance rights vesting after one full year of commercial production from plant start-up, and US$2m after two full years, all at a deemed issue price of A$0.775. The announcement keeps the Q4 2026 clock — “MREC processing targeted next quarter Q4 2026”, “targeting first processing at SARECO in Q4 2026” — and softens the feed plan by one adverb: the ~12,500 tonnes of Kangankunde concentrate is now “ultimately expected to feed the Facility”, while the company “is also evaluating alternative feedstock available in Kazakhstan” for “the timing and source of initial feed ahead of the Kangankunde ramp-up”.

The market has paid for this. The stock, which last traded at A$0.6975 before the 5 August trading halt discussed below, was reinstated on 10 August and closed at A$0.825 — up 18.3% on the day. At that close the company’s 1,849,396,405 quoted shares are worth about A$1.53bn — for a company with no producing asset yet on either continent. In late July the ASX queried a price run from a $0.72 close on 24 July to an intraday high of $0.845 on 27 July; the company answered that it was aware of no undisclosed information — a routine price query, which the company answered; nothing in this paper suggests that answer was other than correct.

The company’s own website, refreshed since the 10 August announcement and re-read on 11 August 2026, now headlines “100% Lindian Ownership” and “Q4 2026 Target MREC Production”, says “[a]pproximately 12,500 tonnes per annum of monazite concentrate from Stage 1 of the Kangankunde Rare Earths Project is expected to feed the facility”, and rests the pathway on “a prior operating history and established permitting framework”.

What none of these documents states — the March announcement, two quarterlies, the placement, and now the twelve-page definitive-agreement announcement issued after completed due diligence — is a rated or design capacity for the plant: no nameplate, no design capacity, no tonnage the plant is rated to take. The nearest the 10 August announcement comes is a class statement rather than a figure: SARECO is “the only constructed and operational MREC/CLP Processing Facility of commercial scale outside of China, MP Materials Inc., Serra Verde and Lynas Rare Earths Ltd.” It places the plant among named producers of scale; it attaches no tonnage to it. None of the documents reconciles the 12,500-tonne expectation with the figure the plant’s operator has filed.

The constraint#

The plant’s operator files public documents with Kazakhstan’s environmental regulator. Under the Kazakh Environmental Code, projects with emissions consequences pass through public hearings, and the applications — emissions projects, waste-management programmes, environmental-impact assessments — are published on the Ministry of Ecology’s National Databank on Public Hearings, where anyone can download them. The filings are in Russian and Kazakh. They are made by the plant’s operator under statutory obligation, they predate Lindian’s transaction by years, and Lindian controls no part of them. They are the plant’s own account of itself, given to a regulator, at a time when no Australian investor was listening.

The operator — the same “Summit Atom Rare Earth Company” LLP named as seller in Lindian’s announcement, BIN 100540004010 — filed nine public hearings between December 2021 and February 2025. The most recent, registration 24373011001, was published on 9 January 2025, fourteen months before Lindian’s announcement. Its centrepiece is a draft emissions-limits project (Проект НДВ) that describes the plant’s modernisation programme in specific, internally consistent numbers:

• The filing says three things about the rare-earths line’s scale, in three places. Its opening states that the project «позволит увеличить объемы производства готовой продукции с 180 до ~ 600 тонн ТRЕО» — will raise finished-product output from 180 to ~600 tonnes of TREO of inorganic rare-earth concentrates. A separate bullet states the line’s production capacity flatly: «Производственная мощность НКТМО … 600 тонн в год» — production capacity, 600 tonnes a year. A third gives the same figure as the uprate target: «Увеличение производительности по выпуску НКТМО до ~600 тонн TREO». The document then describes the programme as bringing the enterprise to its design capacity («выйти на проектную мощность») — without stating that design capacity as a number.
• The declared programme spans three product lines: 600 t/y of rare-earth concentrates, 24,000 t/y of sulfoammophos fertiliser, and 15,000 t/y of high-purity manganese sulphate monohydrate.
• Daily rate 1.81 tonnes; hourly rate 0.075 tonnes; 330 working days a year. The arithmetic closes on itself: 0.075 × 24 = 1.80 t/day; 1.81 × 330 = 597 t/y against the stated ~600.
• Feedstock: technogenic mineral formations — uranium-era processing residue — stockpiled at the former Prikaspiysky mining-chemical combine at Aktau, grading 5.13% rare-earth oxides (sum of REO + Y₂O₃).
• A workforce of 125, and a sulphation furnace in the equipment list. The non-technical summary in the same hearing package names solid and liquid radioactive waste as the principal hazardous factors of environmental impact from the plant’s operations.

The same package, carrying the same figure, was noticed twice — for a December 2024 hearing and again for the February 2025 hearing — so the figure appears in two consecutive notices of one application, not two independent applications. Independent corroboration of the 600-tonne figure from a different document in the package remains open work.

Third-party records frame the plant’s history, and they cut both ways. The founding joint venture — Sumitomo Corporation with Kazatomprom — opened the plant in November 2012, and Sumitomo’s own release set “an annual output target of 1,500 tons during the initial years”, from uranium-ore residue feed. The release’s own product description is “mixed rare earth carbonate”, so its natural reading is gross carbonate tonnes — meaningfully less in contained-oxide terms — but Kazakh trade press later rendered the figure as “1,500 tonnes TREO (the sum of rare-earth oxides)”, and this paper takes that most generous reading, on which the Stage 1 feed plan still carries about 4.6 times it. The same 2016 analysis carries larger numbers, and they are stated here because they cut against this paper: it reports that the then state owner’s own website had projected the plant’s capacity rising to 3,000 tonnes of TREO a year, and to 5,000–6,000 tonnes by 2017 («предполагалось, что производственная мощность предприятия вырастет до 3 тыс. тонн TREO в год, а к 2017 году – до 5-6 тыс. тонн TREO в год»). The upper of those projections sits within about 15% of the 6,875 tonnes this feed plan carries — the strongest evidence in the public record that the site was once conceived at something near Stage 1 scale. What the same article records is that none of it happened: the plant’s capacity was reported as having fallen to 700 tonnes («производственная мощность сократилась до 700 тонн»), 2015 output at about 4.5% of possible volume — a ratio the article derives by setting concentrate tonnages against the 1,500-tonne capacity, so indicative rather than like-for-like — and quarterly output at 1.4 tonnes by mid-2015. The bull reading of that nadir is the fairer one to state: on Lindian’s own account the collapse traces to “constraints in procuring high quality feedstock” — a feedstock failure, not a plant failure — and a buyer that owns a 55%-grade concentrate stream removes exactly that constraint. Nothing in the 2016 record describes the circuit as it stands after the 2021–25 modernisation. A February 2017 report marked the restart under Kazatomprom after Sumitomo’s exit; the operator’s own most recent filing puts the line’s production capacity at ~600 tonnes, alongside fertiliser and manganese co-production. Of every number ever attached to this plant, the only one filed with a regulator in the last decade is ~600 — and of the figures that were ever declared rather than projected (1,500 aimed, 700 reported, 180 before the current uprate, ~600 filed), not one reaches a quarter of the contained oxides in the buyer’s feed plan.

Grey: the operator's own January 2025 regulator filing — output before the current uprate, and the stated production capacity the modernisation works toward. Blue: the contained rare-earth oxides in the ~12,500 tpa of 55%-grade concentrate the buyer's Stage 1 plan supplies (before any netting for the standing 6,000 tpa Iluka concentrate offtake discussed in the text). A higher historical figure — the founders' 1,500-tonne target — is discussed in the text and left off this axis because whether it counts gross carbonate or contained oxide is unsettled. Log scales flatter gaps like this one; the axis is linear from zero.

The gap#

Lindian’s own two numbers fix the feed: ~12,500 tonnes a year of concentrate, at “a grade of approximately 55% TREO” on the ANSTO head assays quoted in the same announcement. That is 6,875 tonnes of contained rare-earth oxides a year delivered to the plant.

One qualification on that figure, from the company’s own record. The 30 July 2026 quarterly states that the fifteen-year Iluka Resources offtake “remains on foot with a commitment for 6,000 tonnes per annum of Stage 1 Monazite Concentrate” — concentrate sold as concentrate, not fed to Stepnogorsk. (The same quarterly records another offtake’s termination as “enabling earlier supply of Monazite Concentrate to the SARECO MREC Processing Facility”.) The company has not reconciled the Iluka commitment with the ~12,500 tpa it says is “ultimately expected to feed the Facility”: either Stage 1 output is larger than 12,500 tpa, or a substantial part of it is already contracted elsewhere. On the second reading the contained oxides actually reaching the plant at Stage 1 would be nearer 3,600 t/y and the gap nearer six times the filed figure rather than eleven. This paper uses the company’s own 12,500 tpa because that is the number the company puts against the Facility, and states the alternative here so the reader can discount it.

Set against the plant’s filed production capacity of ~600 tonnes a year:

Measure	Tonnes TREO/y	Multiple of filed capacity
Stage 1 contained TREO in feed	6,875	11.5×
Recovered into product at the 3 March announcement’s overall estimate, 85–90%	5,844–6,188	9.7–10.3×
Recovered at the 96% TREY extraction restated on 10 August	6,600	11.0×
Feed net of the 6,000 tpa Iluka offtake, contained	3,575	~6×
Stage 2 additional feed (April quarterly’s stated goal, 100,000 tpa)	55,000	~92×

However the recovery assumption is set, the plan requires the plant to process roughly ten to eleven times the rare-earth throughput its operator described to its own regulator in January 2025. Stage 2 volumes would be about ninety-two times the filed figure; the announcement calls the plant “[s]calable” to Stage 2 production but puts no number on that claim, and the quarterly’s own text frames the Stage 2 tonnage as a goal contingent on a drilling programme still converting resource.

The feedstock compounds the arithmetic. The permitted programme is built around 5.13%-grade residue trucked from Aktau; Kangankunde concentrate grades about 55%. This is not more of the same material — it is a different mineral, ten times the grade, imported rather than domestic. (On radioactivity the company’s record runs the other way and is stated here in fairness: ANSTO found uranium and thorium in the MREC “below analytical detection limits”, the concentrate is exempt from radioactive transport classification, and the 15 July 2026 announcement is titled in part “Non-Radioactive MREC”.) A change of feedstock and an elevenfold change in rare-earth throughput are each the kind of change an emissions-limits programme describes.

Share of the feed material that is the target oxide, per cent. Grey: the residue feed in the operator's January 2025 filing. Blue: the concentrate grade quoted in the buyer's own acquisition announcement.

Then there is what the register does not contain. The operator’s nine public hearings, December 2021 to February 2025, are the stages of one modernisation project’s permitting — environmental-impact assessment, then state ecological expertise — and the register shows earlier gaps of over a year within that sequence, so an interval of silence is not, by itself, anomalous. The narrower observation is this: nothing on the public-hearings databank, as at the morning of 11 August 2026, describes an uprated rare-earths programme, a change of feedstock, the buyer entity, or Lindian — and the plan requires the first two and involves the second two. The sweep queried the register by customer name across every name variant of the operator, the buyer entities and the counterparty, in both casings of the case-sensitive filter, was control-tested against operators known to file, and was then confirmed independently of the name filter altogether: the register’s entire index since January 2025 — several thousand hearings — was pulled unfiltered and searched, and contains one entry by the operator (the January 2025 application this paper quotes) and none touching the plant, the buyer, or the feedstock. Two fair caveats bound this: completion of the purchase remains pending, so the entity that would file for the new programme may simply not be in a position to file yet; and the databank records matters that pass through public hearings, which the desk has not established is every permit instrument a feedstock change would require. This pillar is therefore a dated observation and a monitoring point, not a conclusion.

Meanwhile the clock is running, and every date on it is the company’s own — though the reader should be clear what the near dates do and do not require. The 3 March announcement set “High Value MREC Production By Q4 2026” in its title and dated its deferred-consideration milestone — payable three months after the plant is “commercially producing MREC” — at “~1H 2027”; the 10 August announcement targets “MREC processing … next quarter”; the company’s website, re-read on 11 August, targets “first MREC production in Q4 2026”. The definitive agreements the June-quarter report targeted for Q3 2026 were executed on 10 August, within that window; completion — the title transfer that triggers the first US$15m of RA Group’s equity — remains subject to conditions. First processing in Q4 2026 does not require the uprate: on in-country feedstock it could run inside the filed programme, lawfully and without a new hearing, and a first-MREC announcement in Q4 2026 would not answer the question this paper raises and should not be read as answering it. The question falls due as Kangankunde concentrate arrives at Stage 1 rate — which on any mine ramp-up is later. At that point a plant filed at 600 tonnes must be permitted, recommissioned and uprated to roughly ten to eleven times its filed rare-earth throughput. The constraint this paper describes is rate, not capability: the plant has already run Kangankunde material — the 3 March announcement records a 10 kilogram sample “successfully produced … through MREC Facility” meeting product specifications — and nothing here suggests the circuit cannot treat the material. What the regulator’s public-hearings file, re-swept on the morning of publication, contains no application for is the rate.

The desk’s analysis is the join: the feed arithmetic from the company’s announcement, the capacity from the operator’s filing, the silence from the register. Each input is a primary document; the inference that they are in tension is this desk’s.

Corroboration#

Secondary signals, offered as secondary.

The company describes the plant as ready and as needing recommissioning in the same documents. The April quarterly carries the line “The SARECO facility is ready.” — in a passage contrasting the purchase with “the half a billion dollars or more it would cost to build one from scratch” — and, in the same document, lists “completion of key SARECO due diligence and recommissioning” among the June-quarter priorities. On the ordinary reading those are compatible: a plant can be ready in the sense of already built and still need recommissioning, and this paper does not suggest otherwise. The company’s website today describes “a structured recommissioning and optimisation program” and “a streamlined pathway to recommissioning and commercial operation” on the strength of an “established permitting framework” — the framework whose most recent public-hearings entry describes 600 tonnes on residue feed. The documents do not state what the recommissioning involves, what it is designed to reach, or what it will cost. The record shows both words; what neither states is the throughput the recommissioned plant is designed to reach.

The whole country’s export record is an order of magnitude below the plan. UN Comtrade records Kazakhstan’s exports of rare-earth compounds (HS 2846) to the world peaking at 1,047.6 tonnes gross in 2023 — the best year in the series. Three caveats bind this hard: HS 2846 also captures scandium compounds; gross product weight is not contained TREO; and exports are not production. It is order-of-magnitude context only. But the order of magnitude is the point: at the company’s own 48.6 wt% TREO+Y product specification, 6,875 t of contained oxide is roughly 14,000 t of gross MREC from one plant — better than ten times the country’s best reported export year.

Annual exports to the world, gross product weight, from the UN trade database. The database reports no data for 2017, 2019, or any year after 2023. Gross weight is not contained oxide, the class includes other compounds, and exports are not production — the comparison is order-of-magnitude only.

The counterparty was named three ways in one document, and has now changed sides of the table. The partner that was to fund US$7.35m of the US$15m purchase for 49% of the JV appeared in the March announcement as “RA-Group LLP” in the opening paragraph, “RA Group LLP” in the body, and “RA Group LTD” in the binding-terms Parties row: two different corporate forms across three spellings, including in the row whose purpose is legal identity. The 10 August announcement settles the drafting — “RA Group Ltd” in the Subscription Agreement’s Parties row — and settles the seller question the same way: the executed SPA names the seller as the plant’s registered operator at BIN 100540004010, without the stray “Arctic” of the March opening paragraph, so that discrepancy resolves as clerical, exactly as this paper’s field assignment anticipated it might. What the August announcement does not do is locate RA Group in a register: “Ltd” is not a Kazakh corporate form, no jurisdiction is stated, and the Kazakh state-register routes reachable without payment (the open-data portal, the statistics bureau’s search, and a commercial aggregator) return nothing in any rendering — a search whose reach is limited, and stated as such in the methodology note. Under the restructure it no longer invests US$7.35m; it receives up to US$22m in Lindian shares and performance rights, the first US$15m tranche on the assets’ title transfer. The announcement states the basis for it: the equity is payable “in connection with the Transaction and delivery of unencumbered titles to the assets”, and the move to full ownership “provides Lindian with 100% exposure to SARECO’s production and margins, exclusive marketing rights over MREC produced from the Facility and full control over future investment and development decisions”. Its stated rationale for the structure is quoted here in fairness: “The performance-linked component aligns RA Group with the successful transition of SARECO to Lindian ownership and its sustained commercial operation, while preserving Lindian’s cash position.” The register extract remains a field assignment, not a finding. No adverse inference is drawn, and none should be read into this paragraph.

The buying entity resolves in the register — as a trading company, registered a year ago. The SPA names the buyer as “Silkway Metals LLP (BIN 250840006533), a Lindian Resources Limited affiliate / nominee”. That entity does resolve in free tiers of Kazakhstan’s public company records: registered on 7 August 2025 — just under seven months before the March announcement — with a registered activity of wholesale trade in rare, rare-earth and non-ferrous metals, and an Astana address. None of this is adverse: an acquisition vehicle formed ahead of a deal, with no operations yet, is ordinary, and its early date shows the Kazakh structure predates the March announcement by months rather than following it. Its founders are not visible on the free tiers, and “affiliate / nominee” is not further defined in the announcement; the register extract that would show its ownership is a field assignment. It has no filing on the environmental regulator’s public-hearings databank.

The stock was halted while this paper was in preparation — over something else. On 5 August 2026 the company requested a trading halt “pending an announcement regarding a response to media speculation”; the halt became a suspension, and the stock was reinstated on 10 August when the company responded. The response concerned the Malawi project — Government statements in the National Assembly on 7 August that the mining licence “remains valid and in good standing” and that inspections “found no evidence of unauthorised extraction or illegal mining” — and does not address the Kazakh plant’s capacity or permitting, so the kill criterion this paper set against it did not fire. One note for completeness: independent press accounts of the same sitting report that Parliament also ordered a committee investigation of the project. The company’s announcement, which is itself a response to commentary it describes as “inaccurate and misleading”, addresses that process once, saying it is “not aware of any matter arising from the recent media reports or the parliamentary committee process that is expected to materially affect the Kangankunde Project, the validity of mining licence MML0290/22, the Company’s ability to continue operating under that licence, or its previously announced target of first production in Q4 2026”. The Malawi licensing question is outside this paper’s scope, no view is taken on it here, and nothing in this paper suggests that statement is other than accurate. This paper was held, unpublished, until that response had been released and read; the publication-timing note in the methodology box records how.

What would explain this instead#

Four explanations, argued properly. This paper alleges no wrongdoing of any kind by anyone, and any of these four may simply be the answer.

1. The filed 600 tonnes is a programme, not a ceiling. Kazakh emissions-limits projects are written around a declared production programme. On this reading, the plant’s physical circuit — reactors, sulphation furnace, filter presses, calciner — could take more, and the new owner will simply file a new programme when its plans firm. The 10 August announcement adds a genuine strand to this steelman: the company says it is “evaluating alternative feedstock available in Kazakhstan” for “the timing and source of initial feed ahead of the Kangankunde ramp-up”, with “potential in-country feedstock sources and stockpiles” able to bring processing forward. One specific domestic feed is what the current permit envelope describes — the Aktau residue, tabulated by composition in the filing. If that stockpile is the alternative feedstock the company means, an initial period of processing could begin inside the filed programme, lawfully and without any new hearing, and the absence of a new filing today would be unremarkable. The company names no source, so this reading is open to it but not established. What this strand cannot do is close the Stage 1 gap: it defers the question, which falls due whenever the ~12,500 tonnes of 55%-grade concentrate the plan is built on — “ultimately expected to feed the Facility”, in the announcement’s words — arrives at rate. Parts of the programme-not-ceiling reading are supported by the desk’s own arithmetic and by the filing itself: at 5.13% feed grade, 600 t/y of output implies the plant moves roughly 11,700 t/y of solids on the rare-earths line alone at full recovery of the contained oxide — about 14,600 t/y at an 80% recovery — and the filing’s thermal-treatment source data declares 40,515 t/y of material transferred, more than three times the planned 12,500 tpa of concentrate. Materials handling is not the constraint, and this paper does not claim it is. The filing states 600 t as the declared programme’s production capacity, and says the works will let the enterprise reach its design capacity — without stating what that design capacity is. The strongest reading against this desk is that it is not 600 t at all: this paper’s own Kazakh sources put the plant’s founding design rating at 1,500 t TREO («Завод рассчитан на производство 1500 тонн TREO», February 2017), two and a half times the current filing. On that reading a 180-to-600 t step is a stage in a recovery toward an existing rating, not the rating itself, and the uprate precedent is the plant’s own history. And what the desk cannot say is where the rare-earth-specific stages bind. Leach, purification and precipitation vessels are sized by solution volume, and 55%-grade feed carries roughly ten times the oxide per unit of volume that 5.13% residue does — ANSTO’s own pregnant leach solution reached 21.1 g/L TREY, the highest tenor in its programme. On that arithmetic a volumetrically limited circuit could carry far more contained oxide than the filed programme declares. The desk has not sized the acid-bake and leach circuit and does not claim to have; that is a field assignment below, and it is the assignment that could retire this pillar. What would settle it: a new filing on the register declaring a Stage 1-scale programme; or an engineering review of the filing’s per-unit emissions annexes showing the acid-bake and leach circuit sized far above the declared programme.

2. MREC is an earlier-stage product than the concentrate the permit describes, so the binding stage may differ. The plant’s filed flowsheet includes purification stages added after the ownership changes; if the new owner stops at a coarser mixed carbonate, the constraining unit operation could be different from the one the 600-tonne figure describes. This is technically literate and partly supported — but the filing’s own product description already includes rare-earth concentrate in carbonate form, which is substantially what MREC is, and the figure the filing attaches to it is 600. What would settle it: a flowsheet-level capacity statement for the acid-bake and leach circuit specifically — obtainable from the filing’s annexes under specialist review, or from the company simply stating it.

3. The company never claimed a capacity, so there is nothing to correct. This is the company’s strongest defence, and the desk’s own grep proves the factual half of it: neither the 3 March announcement nor the 10 August definitive-agreement announcement contains any occurrence of “nameplate”, “design capacity”, “rated capacity”, “plant capacity” or “throughput” — and the only tonnes-per-annum figures in the 10 August announcement are the feed expectation and the third-party acid supplier’s plant, whose capacity is quoted to the tonne. The market was told the feedstock Lindian would supply, and that the plant exists, works, and is permitted — all defensible statements individually. The March structure’s deferred consideration priced ramp-up risk, and the August structure keeps a performance-linked element: US$7m of RA Group’s equity vests only across two years of commercial production. On the March reading, ramp-up disclosure would come when the definitive agreements completed. The definitive agreements have now been signed, after due diligence the company describes as comprehensive across “technical” and “environmental” aspects — and the twelve pages announcing them still state no capacity.

There is an affirmative half to this defence, and it is strong. The 30 July quarterly says the due-diligence programme “covered process flowsheet, circuit capability, reagent and utilities supply, environmental compliance and legal structure”; the 10 August announcement says it was completed by “reputable global firms/ consultants” across technical and environmental aspects, and that it “reinforced Lindian’s confidence in the Facility”. Its Executive Director is quoted that the programme “reinforced our confidence in the condition, capability and long-term potential of the SARECO Facility” and that “[t]he existing cracking, leaching and precipitation circuits and supporting infrastructure provide a strong platform for MREC production”. Having looked at circuit capability with site access, Lindian then raised its commitment from a 51% interest toward US$7.65m of a US$15m price to 100% for up to US$20m cash plus up to US$22m in scrip. That is a costly signal from the best-informed party, and it is evidence this desk cannot match. What it is not is a number: none of it tells a shareholder what the circuit is rated to take, and the paper’s claim is confined to that. What no reader of any ASX release since 3 March — the announcement, two quarterlies, the placement, the definitive-agreement announcement — could learn is the size of what remains: the capacity the plant is filed for, or that the feed plan implies roughly ten to eleven times it, or that the permitting framework described as “established” was established for a different feedstock at a fraction of the throughput. The claim this paper makes is that gap, and only that gap. What would settle it: any company disclosure stating the plant’s current rated capacity and the uprate plan — including the 3 March webcast, which this desk could not retrieve and has flagged as the first field assignment to run.

4. The state wants this plant running, and permitting will follow. The announcement itself carries a statement from Kazakhstan’s Minister of Industry and Construction that, given the project’s strategic importance, it “will receive all necessary support on the ground from the Ministry to secure its successful implementation”, against the backdrop of the United States–Kazakhstan critical-minerals cooperation the same announcement describes. Political backing of that kind is real and can compress timelines. What it does not do is substitute for the filings themselves: the Ministry of Industry is not the environmental regulator, public hearings remain part of the permitting pathway, and the monitoring criterion — a new filing appearing on the register — is unchanged. What would settle it: the filings appearing, at which point the thesis dies on its own kill criteria, as designed.

A fifth, on the names: the three renderings of the counterparty were probably drafting variance, and the stray “Arctic” has since resolved as exactly that — the executed SPA carries the seller’s registered name. The prior held for the seller; the register extract that would confirm it for RA Group is a field assignment.

Kill criteria#

Written before the conclusion, and specific.

1. A new filing on the Kazakh register — by the operator, the buyer entity, the counterparty or Lindian, lodged after 3 March 2026 — declaring a production programme at or near Stage 1 throughput. The capacity strand dies the day it appears. The register is free and public; the desk checks it monthly.
2. Any company disclosure stating the plant’s current rated capacity and quantifying the uprate — in an announcement, presentation, or the 3 March webcast. The omission strand dies, and the paper reduces to the permitting-timeline question.
3. Engineering evidence that the physical circuit is sized well above the declared programme — from the filing’s own annexes or a company technical release. The thesis narrows from capacity to timing.
4. Register extracts resolving the counterparty and seller identities innocently — an established, capitalised partner and a clerical “Arctic”. The counterparty limb closes. Status, 11 August: half met. The executed SPA names the seller at BIN 100540004010 without “Arctic”, closing the seller half exactly as anticipated; RA Group Ltd remains unresolved in any register the desk can reach.
5. Execution and completion of definitive agreements accompanied by a Kazakh regulatory approval package covering Stage 1 volumes. The thesis is spent in full. Status, 11 August: the definitive SPA was executed on 10 August; no approval package covering Stage 1 volumes was disclosed with it, and completion remains conditional. Not met.
6. The company’s response to the 5 August halt addressing the plant’s capacity or permitting. If the market learns the numbers above from the company before it learns them from anyone else, this paper’s disclosure strand is moot — which is the outcome the paper argues for. Status, 11 August: the response, published 10 August, concerned the Malawi licence and does not mention the plant’s capacity or permitting. Not met.

Right of reply#

The company was not contacted before publication. Every claim above is drawn from its own lodged announcements, its own website and its own quoted words, each linked in full, together with filings made by the plant’s operator to a foreign regulator under statutory obligation, and independent press and trade-press reports, all cited. Where the company has addressed a point — its rationale for moving to full ownership, its stated basis for the equity consideration payable to its counterparty, its “Non-Radioactive MREC” testwork, the state support quoted in its own announcement — its position is quoted in its own words and engaged with rather than dismissed. Nothing here alleges any breach of law, of the ASX Listing Rules, or of any regulator’s requirements, by the company, by any of its officers, or by any counterparty named. The argument is that a capacity figure material to the plan has never been published, and that the only such figure filed with a regulator in the last decade is an order of magnitude below what the plan carries.

Methodology#