Editorial Article
Sixty Per Cent of the Fee Is One Precursor
The invoice has three numbers and only one of them belongs to the cathode maker
A tonne of lithium iron phosphate cathode, priced in April 2026, contains two commodities and one service. The commodities are lithium carbonate and iron phosphate. Both are quoted, both are traded, and both move for reasons that have nothing to do with a kiln in Guizhou. The service is the conversion itself — the calcination, the atmosphere control, the milling, the yield loss and the specification that a cell maker will actually accept.
Take the tonne apart. Lithium carbonate at RMB 55,800 a tonne, at 0.2342 tonnes of it, is RMB 13,068. Iron phosphate at RMB 11,650, at 0.956 tonnes, is RMB 11,137. The converted fee the cathode maker invoices — the only number in the transaction it negotiates — is RMB 18,000, and that fee includes the iron phosphate. The invoice is therefore RMB 31,068.
Read the fee against its own contents. Of the RMB 18,000, iron phosphate is RMB 11,137 — 61.87 per cent of the fee. What is left for energy, labour, depreciation, yield loss and profit is RMB 6,863 a tonne, or 22.09 per cent of what the customer pays.
That is the LFP cathode industry stated as a single ratio. The fee is 57.94 per cent of the invoice, and more than half of the fee is a precursor. The business is a tolling operation with a rock attached to it, and the industry has spent five years being valued as though it were a materials company.
Two defensible market sizes, and only one of them is this industry’s
On a conversion basis the market is USD 8.30 billion in 2025, reaching USD 26.20 billion by 2033 at a 15.45 per cent baseline CAGR. On an invoiced basis — the full price on the purchase order, lithium included — the 2025 figure is USD 15.99 billion. Both are arithmetically correct and they are not the same market.
The larger number is a lithium derivative with a fixed 0.2342 multiplier. Size this market on the invoice and it doubles in 2026 for reasons originating in a mine, not in a plant — and the people reading it will conclude that the cathode industry had a good year when what actually happened is that lithium re-priced.
The ratio itself is the proof. The fee is 57.94 per cent of the invoice on the April 2026 quote, and 51.91 per cent on 2025’s weighted fee of RMB 15,806. The fee rose 44 per cent between early 2025 and April 2026. Nothing about any plant changed in that window. What changed was the price of the metal passing through it.
The volume line is one constant: 2,551 tonnes per gigawatt-hour
LFP cell demand in 2025 was 1,477.86 GWh. Of that, 937.86 GWh went to EV and power cells — 1,421 GWh of global power battery output at 66 per cent LFP — and 540 GWh went to stationary storage, which grew about 73 per cent in a year and is now 36.54 per cent of the chemistry’s demand.
LFP consumes 2,551 tonnes of cathode per GWh. That is how 1.48 TWh becomes 3.77 million tonnes, and how 2033 becomes 9.90 million tonnes at a 12.83 per cent volume CAGR.
The storage pool is what changed the year. It is why roughly twenty Chinese cathode plants ran at or above nameplate in the second half of 2025, and why monthly output peaked above 410,000 tonnes. It is also why the growth shape of this market is unlike almost every other materials market in the chain: the peak annual growth year is 2026, the first year of the window, at +51.19 per cent in volume and +54.71 per cent in value. Everything after 2026 is deceleration into capacity that has already been poured.
And the constant is not constant. Higher compaction and the higher voltage of LMFP both reduce tonnes consumed per unit of energy delivered. The volume forecast depends on gigawatt-hour growth outrunning chemistry efficiency. It has, decisively, for three years. It is still an assumption, and it is the one nobody in the industry argues about, because arguing about it is bad for orders.
The global leader earns 9.15 per cent gross, and that is the argument
Hunan Yuneng sold 1,137,100 tonnes in 2025, up 60.03 per cent, at a realised RMB 30,450.06 a tonne. In 2024 it sold 710,554 tonnes at RMB 31,804.10. Volume rose 60 per cent; price per tonne fell 4.26 per cent.
Revenue rose 53.22 per cent to RMB 34.62 billion and net profit rose 115.18 per cent to RMB 1.277 billion — a 3.69 per cent net margin, or RMB 1,123 a tonne. Gross margin was 9.15 per cent: RMB 2,786 of gross profit on a tonne that sold for RMB 30,450.
This is the most integrated producer in the world. It self-supplies iron phosphate and it has its own lithium carbonate conversion. Which means that of the RMB 11,137 of iron phosphate sitting inside the fee, a meaningful share is Yuneng’s own margin rather than a supplier’s — and that is the only reason the gross line exists at all. A cathode maker buying precursor on the merchant market at RMB 11,650 is buying the same tonne of rock and the same calcination, minus the spread.
That is why the defensible investment in this chain is precursor capacity, not cathode capacity. Whoever owns the precursor owns the only part of the fee that cannot be quoted away by a tier-three entrant, and Yuneng’s 70,000 tonnes a year of lithium carbonate conversion is now the template every tier-two plant is copying.
Seventy-eight per cent of the tonnage is sold at a fee anyone can quote
The 2025 generation mix was 44.8 per cent third-generation material at 2.50 g/cm³ compaction, 33.5 per cent second-generation at 2.40, 11.5 per cent third-and-a-half at 2.55, and 10.3 per cent fourth-generation at 2.60.
The first two together are 78.3 per cent of the market’s tonnage, and they are sold at a fee any of a hundred plants can quote. Fourth-generation powder is made by perhaps four to six companies in the world — and the market pays it RMB 18,800 against RMB 15,500 for second-generation, a premium of 21.29 per cent, for a route that requires a second calcination and roughly doubles kiln time per tonne.
A technical frontier rewarded at a fifth of a premium is a frontier being asked to fund itself out of somebody else’s cost base. It also distorts capital: because a double calcination doubles kiln time, capital per tonne of qualified fourth-generation output is materially above capital per tonne of nameplate. The industry reports nameplate. The return is earned on qualified output.
Meanwhile Chinese nameplate reached 6.88 million tonnes in 2025 with 46 per cent added in a single year. 2026 is forecast at 9.50 million tonnes nominal and 9.00 million effective, with 2.50 million tonnes a year of new build landing against demand growth of 1.90 million tonnes. Newbuild is running ahead of demand at exactly the moment the fee is recovering — the classic shape of a margin that gets given back rather than kept.
The demand got a fire test and the supply got a fence
On 15 July 2025 China’s Ministry of Commerce and Ministry of Science and Technology added battery cathode material preparation technology to the restricted-export catalogue — Announcement No. 28 of 2025, item 252604X. The control point is written as a specification: LFP with powder compaction density of at least 2.58 g/cm³ at 220 MPa, 0.1C capacity of at least 160 mAh/g and first-cycle coulombic efficiency of at least 97 per cent. The precursor processes are named too, including iron phosphate with tap density above 1.2 g/cc and magnetic foreign matter below 10 ppb.
Fourth-generation material is specified at 2.60 g/cm³. The control line therefore sits exactly where the commercially valuable grades begin, and the catalogue lists iron phosphate as a controlled precursor process. That cuts two ways, which is rarely said out loud: it is a supply risk to every customer outside China and a competitive risk to every producer outside China, simultaneously and for the same reason.
It also explains why Spain and Malaysia keep appearing in capacity announcements. Neither is cheaper than Guizhou on any operating line that matters. They are capital spent to buy market access in a decade when access is being rationed by rule rather than by price — and investors in this chain are now underwriting a fee and a jurisdiction at once, with only one of those priced in the model.
What breaks it
Sodium-ion. It uses no iron phosphate and no lithium carbonate and none of this fee structure. It is currently a small share of storage, which is precisely where energy density matters least — and storage is 36.54 per cent of LFP demand. Every ten percentage points of storage that migrates to sodium removes roughly 1.4 million tonnes of LFP demand at 2026 volumes, which is 37 per cent of the 2025 base. That is the largest single discrete threat in this market, and it has a two-year lag and a five-year probability.
LMFP. The more likely of the two, because it is the same plants, the same precursor logic and the same customers. Good for the incumbents’ position; bad for this report’s volume line.
The denominator. If 2,551 tonnes per GWh falls by a tenth, a tenth of this market disappears at unchanged cell output.
The one line to watch
Not the cathode fee. The iron phosphate price. At RMB 11,650 it is 61.87 per cent of the fee, and the fee is a residual claim on it. Everything else in this chain is a negotiation. That one is arithmetic — and it is the reason a market that grows its tonnage 2.6 times over eight years only manages a 2.33 per cent CAGR on the fee per tonne.
Value compounds at 15.45 per cent against 12.83 per cent for volume. The gap is not a growth story. It is the fee recovering to where it was, and staying there only if the precursor stays scarce.
Frequently Asked Questions
Why is the LFP cathode market sized at USD 8.30 billion rather than USD 15.99 billion?
Because the larger figure includes the lithium. A tonne of LFP cathode is 0.2342 tonnes of lithium carbonate plus 0.956 tonnes of iron phosphate plus a conversion fee. On the April 2026 numbers the invoice is RMB 31,068, of which the fee is 57.94 per cent. Lithium is a separately quoted commodity, so counting it here would make this market double every time a lithium contract re-prices, for reasons originating in a mine rather than a cathode plant. USD 8.30 billion is the conversion pool the cathode industry actually bills; USD 15.99 billion is the invoiced total, disclosed separately so any figure built on the full invoice price can be reconciled.
How much of the LFP processing fee is actually the precursor?
Roughly sixty per cent. Iron phosphate at 0.956 tonnes times RMB 11,650 is RMB 11,137, which is 61.87 per cent of the RMB 18,000 fee and 35.85 per cent of the RMB 31,068 invoice. That leaves RMB 6,863 a tonne — 22.09 per cent of what the customer pays — to cover energy, labour, depreciation, yield loss and profit. The fee share rises with the fee, so the ratio was 51.91 per cent on 2025's weighted fee of RMB 15,806 rather than 57.94 per cent on the April 2026 quote.
Is LFP cathode manufacturing technology export-controlled?
Yes. Announcement No. 28 of 2025, item 252604X, added battery cathode material preparation technology to China's restricted-export catalogue on 15 July 2025. The control point is a specification rather than a product: powder compaction density of at least 2.58 g/cm3 at 220 MPa, 0.1C capacity of at least 160 mAh/g and first-cycle coulombic efficiency of at least 97 per cent. Precursor processes are named too, including iron phosphate above 1.2 g/cc tap density with magnetic foreign matter below 10 ppb. Fourth-generation material is specified at 2.60 g/cm3, so the control line sits where the commercially valuable grades begin.
What is the biggest risk to the LFP cathode forecast?
Substitution, and it arrives through storage rather than through cars. Stationary storage is 36.54 per cent of LFP demand and is the segment where energy density matters least, which makes it the natural first market for sodium-ion. Every ten percentage points of storage that migrates removes roughly 1.4 million tonnes of LFP demand at 2026 volumes — 37 per cent of the 2025 base. LMFP cannibalisation is the more likely of the two threats because it runs through the same plants, the same precursor logic and the same customers.