Editorial Article
The Cheaper Salt That Costs More
The salt is 53% cheaper and the electrolyte is 1.82% dearer, in the same month
The case for sodium-ion rests on a salt. Sodium hexafluorophosphate is chemically the lithium salt’s twin — a monovalent cation sitting on the same hexafluorophosphate anion — and it has been dramatically cheaper for as long as anyone has quoted it. In January 2026 NaPF6 was quoted at RMB 79,500 a tonne against RMB 168,500 for LiPF6. That is 52.82% cheaper by weight.
Correct for the mass of the charge carrier and it gets better. Sodium is lighter than lithium per mole of the working ion, so per mole of charge carrier the two salts come to RMB 187.95 for sodium against RMB 151.90 for lithium — 61.87% cheaper. On the line item that everyone quotes, the sodium chemistry wins by a distance that looks decisive.
Then the cell drinks it. A sodium-ion cell absorbs 1,300 tonnes of electrolyte per gigawatt-hour. The lithium chain runs at 1,000. At RMB 28,000 a tonne for sodium electrolyte, that is RMB 0.0364 per watt-hour. At RMB 35,750 a tonne for LFP electrolyte, the lithium cell pays RMB 0.03575. The cheaper salt produces an electrolyte that is 1.82% more expensive per kilowatt-hour, and the dosage cancels the discount almost exactly.
That is the whole finding, and it is arithmetic rather than judgement: 30% more material at 21.7% less price per tonne is a premium, not a saving.
The 1,300 is not an estimate, and it is corroborated twice from two different materials
Dosage is the number the entire comparison turns on, so it is worth showing where it comes from. The cleanest source is not a survey. It is a procurement contract: CATL’s 2026 electrolyte contracts are for 770,000 tonnes against 770 GWh of cells. One line, one division, and it yields 1,000 tonnes per GWh for the lithium chain — not a modelled assumption but the rate at which a cell maker actually buys the material.
The sodium rate holds up under the same treatment. Reported electrolyte shipments of 4,818 tonnes in the first half of 2025 against 3.7 GWh of cells imply 1,302 tonnes per GWh. Reported shipments of 5,060 tonnes in 2024 against 3.6 GWh imply 1,406. Two years, two independent pairs of numbers, and both land within 8% of the 1,300 the model uses.
Those two figures also say something the industry has not absorbed: dosage is falling. 1,406 to 1,302 is 7.4% in twelve months, as formulations improve and cell design gets more efficient. That is simultaneously this market’s tailwind — cheaper cells, which is the whole point of the chemistry — and its headwind, because every point of dosage improvement removes tonnage per gigawatt-hour sold. The forecast runs dosage from 1,300 to 950 by 2033 and says so in the risk chapter, where it belongs.
The deflation script broke, and the break is the story
Sodium electrolyte fell from RMB 180,000 a tonne in 2021 to RMB 28,000 in the first half of 2025 — 84.4%. That decline is the chemistry’s entire commercial promise in one series, and on the strength of it SMM expected NaPF6 to approach RMB 50,000 to 60,000 in the second half of 2025.
It quoted RMB 79,500 in January 2026, above the top of that range. The reason is not sodium demand. It is lithium: LiPF6 bottomed near RMB 49,000 in July 2025 and reached roughly RMB 180,000 by November, and the lithium electrolyte price pulled the sodium salt up with it, because both are made from the same PCl5, the same anhydrous HF and the same fluoride. Do-Fluoride’s own 2025 filing names the three inputs for the sodium salt, and they are the lithium salt’s inputs.
So the input cost of the electrolyte that is supposed to displace lithium is set by the lithium market. This is the same structure as the thermal phosphorus furnace priced by a power bill: a small-volume product manufactured on a shared line from shared intermediates, whose price is therefore decided by the larger market it is meant to attack. Sodium electrolyte is not a price maker. It is a residual claimant on a lithium chain.
And the fluorine has a second exposure with nothing to do with batteries at all. HF is fluorspar plus sulfuric acid, so the anion under both salts is levered to a sulphur market — the same sulphur market that closed the Strait of Hormuz in February 2026 and repriced purified phosphoric acid in this segment’s own report. A sodium-ion cell is a bet on sulphur, whether or not its buyer knows it.
The value went to the formulator, not to the salt or the cell
Per tonne of finished sodium electrolyte at January 2026 prices:
- NaPF6, 140 kg — RMB 11,130, 39.75% of cost
- NaFSI, 15 kg at RMB 150,000 — RMB 2,250, 8.04% of cost from 1.5% of mass
- Solvents, 845 kg — roughly RMB 5,070
- Raw material total — RMB 18,450, 65.89%
- Gross spread — RMB 9,550, 34.11%
Two things follow. The first is that the additive line — 1.5% of the mass for 8% of the cost — is where the technical differentiation sits, and it is where the margin is defended. The second is that when the salt eased to RMB 70,000 by September 2026, the formulator’s spread widened. Cheaper salt does not reach the cell maker. It reaches whoever buys the salt and sells the formulation.
It also matters that this market has no forward curve. SMM describes public quotes, no unified pricing, pulse demand, produce-to-order and no inventory. Nothing in this chain can be hedged, which means the 34.11% spread is not a stable rent — it is a position taken every time a customer order arrives.
The most-quoted cost structure in the chemistry is stale, and I can show which half is wrong
The published split of sodium-ion cell material cost puts the cathode at 26% and the electrolyte at 26% — a parity that has been repeated so often it functions as a premise. Reproduce it at 2025 prices: the cathode comes back at 25.75% of the material pool, essentially unchanged. The electrolyte comes back at 13.66%.
The parity was real when electrolyte sold at six figures a tonne. It is not real now. That matters beyond bookkeeping, because “electrolyte is co-equal with the cathode” is the argument for building dedicated electrolyte capacity ahead of demand — and the argument was constructed at 2021 prices by people who had not yet seen the price collapse.
Capacity is not the barrier, and that is the barrier
Tinci’s Jiujiang approval is RMB 542 million for 500,000 tonnes a year of electrolyte capacity — 450,000 lithium and 50,000 sodium — which is RMB 1,084 per tonne of nameplate. At that cost of entry, two announced sodium projects totalling 100,000 tonnes represent 9.43 times the entire 2025 market and 87.57% of the 2030 volume this report forecasts.
The salt tells the same story from the other side. Do-Fluoride runs 65,000 tonnes of LiPF6 capacity at 70% to 80% utilisation against a sodium line it describes in its annual report as kilotonne-scale — a ratio of roughly 65 to 1. Nobody is short of a plant. What nobody can shortcut is the formulation file, which varies by more than 30% between customers on the same nominal chemistry. That is the moat: not capacity, and not the salt, but a recipe book that has to be rewritten for every cell maker.
Regulation pulls 2026 forward and leaves 2027 exposed
MOF and STA Announcement No. 2 of 2026, dated 8 January 2026, cuts the battery VAT export rebate from 9% to 6% from 1 April 2026 and abolishes it from 1 January 2027, across 271 product codes. The rate that applies is set by the customs export date, not the contract date — which means every tonne that can be shipped before the deadline will be, and 2026 becomes the strongest growth year in the forecast for administrative rather than commercial reasons.
On export control the direction reverses again. Announcement No. 28 of 2025 set thresholds for LFP cathodes at 2.58 g/cm³ and LMFP at 2.38, and published none for polyanion sodium cathodes and none at all for electrolytes. The technology that is fenced in is the cathode. The electrolyte is exportable as written. That is an inference from the published text rather than a statement of intent, and the report flags it as one.
Risk one is a company that stopped, and it stopped for capital
Natron Energy received a WARN notice dated 28 August 2025, and both of its plants — Holland, Michigan and Santa Clara — closed permanently on 3 September 2025, with 95 jobs lost. The company had raised a follow-on round, a Series B, a convertible note and a management buyout, and all four failed.
That is the honest first risk in this report, because Natron was not defeated by chemistry. It made Prussian-blue cathodes, sat outside the Chinese cathode specification, and had a real product. It ran out of capital in a market where the incumbent’s cost curve is still falling and where no customer will sign a long-term contract for a material with no forward curve. The same sentence applies to every sodium electrolyte start-up: the technology works, the arithmetic is thin, and the clock is set by someone else’s balance sheet.
The fourth risk is a fourfold disagreement about 2030
Published forecasts put sodium-ion at 435 GWh by 2030. This report’s path arrives at 2030 with about 114,000 tonnes of electrolyte and roughly 104 GWh of cells — a factor of four apart, and the entire gap sits inside the next four years.
The disagreement is not about the chemistry or the price. It is about what a forecast is counting. The higher figure describes announced capacity; the lower one describes materials that shipped. In a market with two 50,000-tonne projects scheduled against a 10,608-tonne base, the difference between those two definitions is the whole argument. This report does not dismiss the higher path — it names it, states its basis, and declines to adopt it.
The market, and the perimeter it does not cross
Pheonix Research sizes the global sodium-ion battery electrolyte and NaPF6 market at USD 42.58 million in 2025, reaching USD 850.6 million by 2033 at 45.40% a year. Volume runs from 10,608 tonnes to 475,000 tonnes at 60.84%, so revenue per tonne falls from USD 4,014 to USD 1,791 — 9.60% a year. Fifteen points of divergence between tonnage and value is the price of the chemistry’s own promise.
This market sizes the formulated electrolyte. It excludes the cell, the cathode, the hard carbon anode, the merchant solvent and any salt sold outside an electrolyte — the contained NaPF6 pool appears as a memo line and is never added to the total. Cathode and anode are sized in the companion report at USD 162.6 million in 2025, and the bridge is arithmetic rather than assumption: 3,579 tonnes of active material per GWh and 500.0 GWh in 2033, which is exactly the gigawatt-hour base this report’s 2033 tonnage is built on at 950 tonnes per GWh. No tonne is counted twice.
The market’s headline is a 45.40% compound rate. The finding underneath it is narrower and more useful: the cheaper salt costs more per kilowatt-hour, the discount is decided by the lithium market it is meant to displace, and the only durable advantage in the chain belongs to whoever owns the formulation file.
Frequently Asked Questions
Is NaPF6 cheaper than LiPF6?
By weight, yes: NaPF6 was quoted at RMB 79,500 a tonne in January 2026 against RMB 168,500 for LiPF6, 52.82% cheaper. Per mole of charge carrier the gap widens to 61.87%, RMB 187.95 against RMB 151.90, because sodium is lighter per mole of working ion. But a sodium-ion cell absorbs 1,300 tonnes of electrolyte per GWh against 1,000 for lithium, so at RMB 28,000 and RMB 35,750 a tonne respectively the sodium electrolyte costs RMB 0.0364 per Wh against RMB 0.03575 — 1.82% dearer per kilowatt-hour.
Why does a sodium-ion cell need so much more electrolyte than a lithium one?
Sodium-ion absorbs about 1,300 tonnes of electrolyte per gigawatt-hour against 1,000 for the lithium chain, and the 1,000 is not a model assumption — it falls out of CATL's 2026 contracts, 770,000 tonnes against 770 GWh. The higher sodium figure is corroborated twice: 4,818 tonnes of reported H1 2025 shipments against 3.7 GWh implies 1,302, and 5,060 tonnes in 2024 against 3.6 GWh implies 1,406. Dosage is also falling, down 7.4% in twelve months, and the forecast takes it from 1,300 to 950 by 2033.
Who makes sodium-ion electrolyte, and is capacity the barrier?
Tinci's Jiujiang approval is RMB 542 million for 500,000 tonnes a year — 450,000 lithium and 50,000 sodium — or RMB 1,084 per tonne of nameplate. Do-Fluoride runs 65,000 tonnes of LiPF6 at 70–80% utilisation against a sodium line it describes as kilotonne-scale, roughly 65 to 1. Capacity is therefore not the constraint: the moat is the formulation file, which varies by more than 30% between customers on the same nominal chemistry.
What happened to Natron Energy, and what does it say about the market?
Natron Energy received a WARN notice dated 28 August 2025 and both plants — Holland, Michigan and Santa Clara — closed permanently on 3 September 2025, with 95 jobs lost. It was a Prussian-blue cathode manufacturer outside the Chinese specification with a working product. A follow-on round, a Series B, a convertible note and a management buyout all failed: it ran out of capital, not chemistry, which is the risk every sodium-materials entrant shares.