A resources table tells you where lithium sits in the ground. It tells you almost nothing about who makes money. Five lenses do that work — lithium is two products, not one; a cost curve decides who survives a price crash; the demand bull-case hides two real risks; China's dominance faces a slow contest; and a handful of companies occupy the whole structure. This is how to read them.
01 — One element, two markets
"The lithium price" is a simplification. There are two battery-grade chemicals — carbonate and hydroxide — and which one a battery needs is set by its cathode. As the cathode mix shifted, the two prices split apart. This is the distinction most overviews collapse, and it explains a price divergence that otherwise looks like noise.
LFP overtook nickel-based chemistries to take more than half of global EV battery deployments in 2025 — up from under 10% in 2020 — and stationary storage is now ~95% LFP. Both run on carbonate. So the growth in lithium demand has been overwhelmingly carbonate-shaped.
Hydroxide, the salt assumed to command a structural premium, instead collapsed from over $80/kg to roughly $8/kg between late 2022 and 2025 — a >90% fall — as its addressable market (high-nickel NMC) lost share to LFP. Since mid-2023, carbonate has traded above hydroxide, reversing years of the opposite.
The chemistry shift deepens China dependence rather than diluting it: over 98% of LFP cathode and cell production sits in China, a tighter concentration than the nickel-based supply chain it replaced. "The market moved to a cheaper, safer chemistry" and "the market moved further inside China's grip" are the same sentence.
02 — The chart to internalise
A commodity lives and dies on its cost curve. But "cost" is three different numbers doing three different jobs, and conflating them is the most common mistake in reading a miner. Pulled apart, the boom-bust stops being mysterious.
The shut-down test is cash cost (C1) — reagents, energy, labour, royalties. Once capital is sunk, a mine keeps running while price clears cash cost. The build test is the incentive price — cash cost plus sustaining capex, the upfront build capital, and a return on it. It is the price needed to justify a new project. Between the two sits AISC (cash cost plus sustaining capital). The gap between the bottom line and the top line is where the cycle is born.
The two lines explain everything. At the June 2025 trough of ~$8,200, the right-hand third of the curve sat above the price — its cost exceeding what it could sell for — so it mothballed — Australian mines idled, and in August 2025 China's CATL suspended its Jianxiawo lepidolite mine (~65kt LCE/yr) when its permit lapsed. But across that same trough, no new project anywhere cleared the ~$18–22k incentive price, so a wave of greenfield projects was shelved as unbankable. Existing supply survives the bust; new supply doesn't get sanctioned — the pipeline thins, a deficit forms, and price spikes (above $22k by mid-2026, helped by a Zimbabwe export ban). The gap between survival and incentive is the engine that turns every bust into the next boom.
The cost stack ranks operating cost. Capital cost ranks almost the other way, which is the real brine-versus-rock trade-off — and the reason the cheapest-to-run option is not the easiest to build.
| Route | Cash cost | Capex intensity | Build / ramp | Native product |
|---|---|---|---|---|
| Brine — evaporation ponds | low · ~$4–6.5k | high | slow · 4–7 yrs | Carbonate |
| Hard-rock mine — concentrate | low · ~$2.5–5.5k* | moderate | fast · ~2–3 yrs | Concentrate |
| Integrated hard rock — mine + refinery | mid · ~$6–9k | very high | slow | Hydroxide |
| DLE — direct extraction | low · ~$3–6.5k | very high · ~$50–80k/t | unproven at scale | Carbonate |
*Hard-rock mine cash cost looks lowest because it makes only concentrate, a low-value intermediate — not the finished chemical. That is the trap the next panel resolves.
A spodumene mine's headline cash cost is for concentrate, which sells at roughly half the carbonate price. To become battery-grade, that concentrate must be roasted, acid-leached and crystallised into carbonate or hydroxide — a separate step, located overwhelmingly in China. So the cost stack above is only honest on a delivered chemical basis; the conversion layer is the hidden second half, and it is where China's leverage lives.
| Item | Indicative figure | What it tells you |
|---|---|---|
| Spodumene → chemical, China | ~$2–4k | The cheap, dominant conversion step — China's structural advantage. |
| Spodumene → chemical, Western refinery | >$10k (IGO, 2025) | Why refining stays in China despite friend-shoring money. |
| Technical → battery grade upgrade | ~$1.5–2.5k | Plus a 3–5% yield loss; the purity premium. |
| Conversion margin (SC6 → LCE spread) | ~$3.9k now · $13–16k in 2022 | When it turns negative, refineries curtail — the earliest supply-cut signal. |
| Hydroxide → carbonate switch | ~$1.4k | Cheap enough that producers are flipping lines toward carbonate. |
The two deposit types have native products. Brine makes carbonate directly and cheaply; reaching hydroxide needs an extra step. Spodumene reaches hydroxide relatively directly (over 85% of high-nickel hydroxide comes from spodumene); carbonate is equally easy. So the LFP-driven swing to carbonate plays straight to brine's strength — and strands the hydroxide capacity built for a high-nickel future that lost share. Cost curve, chemistry, and the price split are one connected system.
Ask three questions. Where does it sit on the cost stack (survives at $8k, or only at $20k)? Does it merely mine, or also convert — i.e. who captures the conversion margin? And which salt does its geology natively make, carbonate or hydroxide? Those three answers locate almost any lithium name on the map.
03 — Stress-testing the bull case
The consensus points one way — the IEA's ninefold rise to 2040, the industry's tripling, "no substitute." Probably right on volume. But a defensible view names what would falsify it. Here is the honest ledger, with a verdict on each.
| Threat | Effect on Li demand | Timing | Verdict |
|---|---|---|---|
| Sodium-ion | caps upside | Commercial 2026 | The real one — but a price ceiling, not a volume killer. CATL's Naxtra reached mass production in 2026 (first passenger EV: Changan, mid-2026), ~30% cheaper than LFP, superb in cold. Yet it stays lower-energy-density and mostly stationary for now. Crucially it becomes competitive precisely when lithium is expensive — a self-correcting brake that flattens lithium's spikes rather than erasing demand. The "30–40% of the market" target is aggressive and self-interested. |
| LFP mix shift | neutral on volume | Now | Lithium-neutral overall, but it re-routes demand to carbonate and guts hydroxide (see §1). The risk it carries is not "less lithium" — it is "wrong lithium product" for anyone long hydroxide. |
| Falling intensity | mild headwind | Ongoing | Smaller packs, cell-to-pack efficiency and right-sized ranges trim grams of lithium per vehicle. Real but gradual; swamped by unit growth. |
| Recycling | secondary supply | Material ~2035+ | Urban mining becomes a genuine supply source late next decade and a partial hedge against ore concentration — though China leads recycling too, so it is not automatically a Western escape hatch. |
| Solid-state | more lithium | Late decade | Counter-intuitively bullish: most designs use a lithium-metal anode, raising lithium per cell. If it scales, it deepens demand rather than threatening it. |
The bull case is robust on volume and genuinely fragile on two narrower things: the carbonate-versus-hydroxide mix, and the price ceiling sodium-ion now imposes at the low end. The question that matters is not "will lithium be replaced" — it will not, near-term — but "which lithium product, and capped at what price."
04 — The other half of the China story
China's chokehold on refining is well documented. The less-told half is the counter-move — the friend-shoring and resource-nationalism actively reshaping who controls what. The map is not static dominance; it is a live contest.
Neither side is winning cleanly. Friend-shoring is real but slow and expensive — refining is the hardest link to replicate, as the >$10k Western conversion cost in §2 shows — and resource nationalism cuts both ways, raising costs for everyone. The useful model is not "China dominates, full stop," but "China holds the midstream while a coalition spends heavily, and clumsily, to build a parallel one." Where any deposit or company sits in that contest is now a first-order driver of its value.
05 — Who occupies the structure
Organised by position on the value chain, with each name's place on the cost stack (§2) flagged — because that tag says more about who survives a downturn than any country label. A mid-2026 snapshot of an unusually fast-moving layer.
— reading it all together
Four dials carry the whole picture, and the country tables show none of them.
Carbonate or hydroxide — set by cathode chemistry, and diverging hard in carbonate's favour.
Survival vs incentive. Bottom of the stack earns through any price; the top mothballs first, and the gap between them drives the cycle.
Volume is safe; the live risks are the carbonate/hydroxide mix and the price ceiling sodium-ion now imposes.
China holds the midstream; a coalition is building a parallel one. Where an asset sits in that fight drives its value.
Turn those four dials and every company in §5 falls into place. The atoms are an inventory; the sector lives in the product split, the cost stack, the demand mix, and the contest.