The Conversion Cost Did Not Move, and the Margin Was 750 Yuan

The conversion cost did not move

The obvious way to write about glyphosate is demand: transgenic acres, paraquat bans, ninety per cent penetration in soy. The numbers refuse it. IHS Markit put global glyphosate sales at USD 6.15 billion in 2021. The 2025 base is USD 2,743.4 million — 55.39 per cent lower — on greater tonnage. A market cannot lose more than half its value on rising volume and still be a demand story.

What it is instead falls out of two published points in the same year. In March 2025, with glyphosate at roughly RMB 23,000, the glycine-method spread was about RMB 9,000. At the end of August, with the price at RMB 27,300 after a 24 per cent run off the low, the spread was RMB 13,600 and gross profit after depreciation and labour about RMB 3,000.

The conversion cost is stable to within about 2 per cent across a 19 per cent move in product price. The gap between what the herbicide sells for and what it costs to convert the intermediates into it barely breathes: RMB 14,000, then RMB 13,700. Which means the price of the world’s largest herbicide is not being set by what a farmer will pay. It is being set by the cash cost of the marginal route — and at December’s RMB 23,700, the route that 46.1 per cent of world capacity runs on was within about RMB 750 of earning nothing at all.

What one tonne of phosphorus is worth inside one tonne of herbicide

Two calculations make the same point from the phosphorus side.

A tonne of glyphosate contains 18.32 per cent phosphorus. At RMB 24,324 for the element, that phosphorus is worth RMB 4,456 — 17.93 per cent of the herbicide’s price, and 1.49 times the entire gross profit earned per tonne in the good month of August.

And the second, which is stranger. A tonne of glyphosate sells for 1.0216 times the price of a tonne of yellow phosphorus and contains 18.32 per cent of one. All the chemistry between the furnace and the herbicide — the chlorination, the ester, the condensation, the registration file, the twenty-year brand — is worth 2.2 per cent of the input’s unit value.

That is the shape of a market where the differentiated product has been competed back to the cost of the commodity underneath it. The premium is not small. It is gone.

The forecast needs no new furnaces

The base year was built rather than asserted. China’s H1 2025 output of 308.0 kt doubles to 616.0 kt; 810 kt of nameplate at a 76 per cent annualised rate gives 615.6 kt — agreement to within 0.4. Non-China capacity of 368 kt at the same rate gives 279.7 kt. The total is 895.7 kt on 1,178 kt of nameplate, 76.04 per cent. Priced at Xingfa’s four monthly anchors, averaged and de-VATed to RMB 24,850, that is USD 3,062.8 a tonne and USD 2,743.4 million, reaching USD 4,157.1 million by 2033 at 5.33 per cent value and 2.59 per cent volume.

The forecast needs no new capacity at all, and this is not an assumption — new glyphosate plants sit in the restricted category of China’s 2024 industrial catalogue. So the growth comes from utilisation running 76.04 to 93.28 per cent: 203.1 kt of output from zero new furnaces and zero new permits. Every point of utilisation is 11.78 kt.

Value outrunning volume roughly two to one is now the pattern across this whole segment — 5.83 against 3.21 in thermal acid, 6.45 against 2.26 in the chlorides, 5.33 against 2.59 here. That is not house style repeated three times. It is a cross-market finding: in a chain whose volume is capped by catalogue and permit, all of the growth has to come from price and mix.

The chain arithmetic, and a rule that keeps the cluster honest

Both routes need exactly 0.8123 tonnes of PCl₃ per tonne of glyphosate, because 137.33 ÷ 169.07 is the mass ratio of one phosphorus-bearing molecule to another and both routes deliver one phosphorus per herbicide molecule. Route choice moves the intermediates — 0.6509 t of dimethyl phosphite and 0.4440 t of glycine, or 1.3433 t of PMIDA against about 1.5 t practical, an 89.55 per cent oxidation yield. It never moves the phosphorus demand. Methanol, at 0.3790 t in and 0.3790 t out, is a carrier rather than a feedstock.

Applied to China’s 616.0 kt, that is 500.4 kt of PCl₃ — 46.93 per cent of national output. Which states the netting rule explicitly: the chloride market counts the PCl₃ including the half that never leaves the glyphosate site; this market counts the herbicide. The Chinese glyphosate pool alone contains 112.9 kt of elemental phosphorus already counted inside the rock, the furnace and the chloride numbers. Summing the three markets would count the same phosphorus three times.

The margin ladder, per tonne of contained phosphorus

The most useful table in this chain puts each product on the same footing — value per tonne of the phosphorus it contains, against the margin each layer earns.

These are different measures and should be read as order of magnitude, not as a precise ranking. But the direction is unambiguous: a tonne of contained phosphorus earns 18.4 times more as glyphosate than as PCl₃ — and the product with the lowest phosphorus content of the three earns the most per tonne of phosphorus.

PCl₃ sells the phosphorus. Glyphosate sells the registration. In a chain where the element is administered and the volume is capped, that is the only lever that still works — and in December 2025 it was worth about RMB 750 a tonne, which is to say almost nothing.