By-product credits, and why companies don't tell you about them.

The first time we ran the numbers, they didn't agree

A junior miner's investor deck claimed an All-In Sustaining Cost of US$597 per tonne of finished product. Modelled honestly on a polymetallic basis — same orebody, same metallurgy, same mine plan — the same project ran as low as US$283 per tonne. A 53% reduction.

The difference was not an error. It was a disclosure choice. The published number treated a polymetallic project as a single-commodity operation; the integrated number applied by-product credits. Both are mathematically defensible. They describe different things.

This is a routine disclosure choice in mining. It is also one of the most consequential pieces of fine print in resource investing, and it explains more about the gap between published valuations and intrinsic value than anything else we have studied.

This piece is about why that gap exists, why it persists, and what it tells you about how to read a developer's economics.

What a by-product credit actually does

When a single project produces two or more saleable products, an accounting question arises: how do you allocate the cost of mining and processing between them?

There are two conventions.

The first is co-product accounting. You allocate cost to each commodity in proportion to its share of revenue. If lithium accounts for 80% of project revenue and tantalum 20%, then 80% of mining cost is allocated to lithium and 20% to tantalum. Both commodities show their full unit cost. Conservative. Symmetric. Used when neither product is plainly secondary to the other.

The second is by-product credit accounting. You designate one commodity as the primary product and treat the others as by-products. Revenue from the by-products is subtracted directly from the primary product's cost line. The primary product reports a net cost — its full mining and processing cost, minus the value the by-products contribute. Used when one commodity is clearly the project's reason for existing and the others ride along.

Neither method is wrong. They produce different reported numbers because they describe different things — co-product allocation tells you what each commodity costs on its own merits; by-product credit tells you what the primary commodity costs given that the others come along anyway.

The economic substance is identical. The reported AISC is not.

For a polymetallic developer with a clearly dominant primary product and meaningful secondary streams, the second method describes operational reality more honestly. A mine that is going to be built for lithium, and that will produce tantalum and caesium because the geology requires it, faces lithium-specific decisions at lithium-specific costs. The right question for the operator — "what does it cost to deliver a tonne of lithium concentrate, given that we are also selling tantalum?" — is the question by-product credit accounting answers.

The reason this matters: companies frequently choose to publish only the first number. The cost-on-its-own-merits number. The conservative one.

That choice has a habit of looking very different the day it changes.

Six reasons companies don't publish integrated economics

Every developer with material by-product streams faces the same choice in their disclosure documents: publish lithium-only AISC, or publish lithium AISC net of by-product credits, or both. Most publish the first and are careful to bury the second.

The reasons are not what most retail readers assume. The story is rarely the company is hiding the upside — it is almost always something more procedural, more sequenced, more strategic. Six reasons in particular show up repeatedly:

One. NI 43-101 and JORC compliance. The Canadian and Australian disclosure standards are explicit about what counts as a Mineral Reserve. A reserve must be supported by an economic study that meets specific criteria, including formal classification of the resource that backs it. A developer can have an enormous tantalum or caesium resource sitting alongside its lithium reserve, but if the secondary mineral has not been classified as a Mineral Reserve in its own right, it cannot legally be used to flow through the project's reported economics. The company is not refusing to publish — it is forbidden from doing so until the engineering study sequencing supports it.

Two. Engineering studies are sequenced for a reason. A pre-feasibility study covers the primary product first. A definitive feasibility study refines those numbers and may or may not include co-product economics. Updated studies — issued every two or three years as the project develops — are where co-product economics finally land in formal reserve-grade form. This sequencing is best practice, not a delay tactic. Skipping ahead to publish integrated economics undermines the credibility of the underlying study; mining capital markets are conservative about this for good reasons learned over a century of disasters.

Three. The Brinsden / Pilbara playbook. The most successful developer-to-producer transition of the last decade — Pilbara Minerals under Ken Brinsden — was built on a deliberate strategy of under-promising at each engineering stage and over-delivering at the next. The company published conservative economics, raised capital against those, then delivered project performance ahead of guidance, then published better economics, then raised again at higher prices. Each successive raise priced off the credibility built in the previous one. Junior developers watching this playbook explicitly choose conservative disclosure as a corporate strategy.

Four. Negotiating leverage. The day a company publishes its full integrated economics is the day its offtake partners, project finance lenders, and prospective acquirers anchor every conversation around that number. Publishing high integrated NPV early sets a floor for capital costs that is hard to negotiate down later. Most developers find it strategically rational to under-disclose during the financing window and reveal upside after the most expensive capital is raised.

Five. Caesium and other small-market commodities are awkward to value publicly. Caesium pricing is bilateral and opaque. There is no public benchmark. Publishing a price implicitly sets one — and alerts every competitor to your assumed monetisation. For tantalum, the situation is similar: spot pricing exists, but the largest contracts are bilateral and the published number is rarely what the largest producers are actually receiving. Companies prefer to keep these markets opaque because opacity is in their commercial interest.

Six. Conservative communication compounds credibility, and credibility raises capital cheaply. This is the most under-appreciated reason. A company that publishes conservative numbers, then beats them, then publishes again, builds a reputation that compresses its cost of equity. Every subsequent capital raise prices off that credibility. A company that publishes aggressive numbers, then misses them, finds itself diluting at deeper discounts when the gap to delivery becomes visible. Over a decade-long development cycle, the difference between these two reputational paths is hundreds of basis points of cost of capital.

These reasons are not exhaustive, and they overlap. Together they explain why almost every developer's published AISC is the lithium-only number, the conservative number, the under-disclosed number — even when the operator privately models the project on full integrated economics every day.

A worked example, with real numbers

To make the above concrete, here is the AISC stack for a real polymetallic lithium developer we have been studying — PMET Resources, listed as PMT.AX in Australia and PMET on the TSX.

The published lithium-only AISC, drawn from the October 2025 Feasibility Study and signed off by qualified persons under NI 43-101, sits at US$597 per tonne of SC5.5 (spodumene concentrate at 5.5% Li₂O grade). This is the number that appears on the cover slide. It is the number quoted by analysts. It is the number that flows through into virtually every published valuation of the company.

The same orebody contains a substantial tantalum resource. Tantalum is mined alongside lithium with the same ore body, the same haul road, the same crushing circuit, and the same labour. Industry planning ranges for tantalum pentoxide put conservative long-term prices around US$100 per pound, mid-cycle pricing around US$150 per pound, and sustained-spot scenarios at US$200 per pound under tight supply.

The conversion of these tantalum prices into per-tonne credits against lithium concentrate output is calculated from PMET's own resource ratios — the tantalum tonnage produced per tonne of SC5.5 spodumene is project-specific. The arithmetic that follows uses these calculated credits.

Three observations.

First, even the most conservative case — tantalum at the low end of long-term planning prices — moves net AISC by 26%, from US$597 to US$441. This is not a marginal effect. It is a structural recalibration of the project's competitive position.

Second, at mid-cycle pricing the net AISC of US$360 places the project at a level comparable to the second quartile of the global lithium cost curve. At sustained-spot pricing of US$283, the project sits at a level comparable to the bottom-decile producers — alongside Greenbushes and the Atacama brine operations, the assets the rest of the industry uses as a benchmark of long-term durability.

Third, none of this counts caesium. Caesium markets are too thin to defensibly model in a public note, but the company's resource is one of the largest in the Americas and the cost-curve compression numbers above would compress further once formally included.

The total effect, before any caesium contribution, is to move a project from "second-quartile producer at low commodity prices, breakeven at the cycle bottom" to "bottom-decile producer at low prices, structurally insulated through the cycle." These are entirely different economic propositions. They sit inside the same orebody, the same engineering study, the same management team. They differ only in the convention used to allocate cost.

The market currently prices the company off the first number.

What this changes for readers

Three takeaways, for anyone reading published economics on polymetallic developers:

The first: when you read a junior miner's published AISC, ask explicitly "what has been excluded for compliance reasons?" If the company's resource statement names secondary minerals — tantalum, caesium, beryllium, lithium hydroxide concentrate, gallium, anything — those are likely sitting outside the cost line for sequencing reasons rather than economic ones. The published economics are conservative by construction. The company knows this. The market prices off the published number anyway.

The second: identify the moment in the project's engineering study sequence where integrated economics are scheduled to land. For most developers this is the broader-project PEA or the updated FS for lithium that incorporates by-product credits formally. Once that study lands, the integrated number becomes citable in disclosure documents, then in broker reports, then in retail dashboards. The market reprices off the new number. The window between today and that scheduled disclosure is the window in which a careful reader can hold the integrated view privately while the market still trades the lithium-only view.

The third: treat the discount-to-fair-value as structural, not anomalous. Pre-production developers consistently trade at 30–50% of disclosed NAV. The discount is rational — it reflects financing risk, permitting risk, capex inflation, execution risk over multi-year construction periods. None of that goes away. But the baseline against which the discount is taken is the published number, which we have just demonstrated is conservative. A 35% discount to a conservative NAV is a much smaller discount to an honest one. Some of what looks like a discount is actually the under-disclosure compounding in the reader's blind spot.

None of this is a recommendation. Every individual developer has different by-product circumstances. The arithmetic above can run in either direction — a project where the by-product credits are generous because the secondary mineral pricing is unsustainable will look better on paper than it deserves to, and the same framework applied honestly will reveal that. The point is not that polymetallic developers are systematically undervalued. The point is that the published economics are not the economics that matter, and reading research carefully is mostly the work of unpacking the second from the first.

What we don't know

A piece that ends here would be cheerleading. The honest section comes next.

Caesium pricing held at spot is the weakest link in any polymetallic NPV. These markets are bilateral and opaque, the published prices do not reflect what large contracts actually clear at, and the largest resources — including the one we have referenced above — are large enough that selling at sustained spot pricing is structurally unrealistic at scale. Any model that holds caesium at spot prices is overstating the credit. The honest scenario has caesium contributing meaningfully but not at full spot multiples.

By-product credit accounting is also conventionally aggressive in its treatment of allocation. A genuinely co-product project, modelled honestly, sits somewhere between the two methods. Companies that dispense with the lithium-only framing entirely and publish co-product economics tend to report numbers that are more conservative than full-credit, more generous than lithium-only. The "right" answer for any given project is somewhere in this band, and we should not pretend the bottom of the band is automatically correct.

Finally, by-product credits do not eliminate the binary risks every developer carries — permitting decisions, financing windows, capex inflation, commodity-price collapse during construction. A bottom-decile producer in a US$700/t lithium environment is still a producer, but only after construction. Until first production, the project remains a long-duration call option on the operator's ability to execute against the published plan. By-product credits change the steady-state economics; they do not change the path-dependence of getting there.

Reading research carefully means holding all of this at once.

Closing

The numbers in published decks are facts, audited by qualified persons, signed off under disclosure law. The framing of those numbers — whether to apply by-product credits, whether to publish integrated economics, whether to lead with lithium-only AISC — is a decision. It is a decision shaped by compliance, sequencing, negotiating leverage, communication strategy, and the structural opacity of certain commodity markets.

Reading research carefully is mostly the work of unpacking the second from the first.