Cross-reference three sources before believing one. That is the rule, and it applies to lithium carbonate this week with unusual force, because three independent data streams have all moved at once — the futures tape, the weekly inventory count, and the cathode producers’ output schedule. No single source holds on its own. Triangulate them, and the signal emerges.
The first source is the futures tape, dated August 28. The main lithium carbonate contract, LC2701, settled at 159,600 yuan per tonne, up 4.77% on the day, with open interest rising by roughly 30,000 lots in a single session. Spot battery-grade material traded at 152,500 to 158,000 yuan per tonne — a premium to the front end that says physical buyers are paying more than the futures market’s consensus. That is the kind of cross-check that matters: when spot sits above the contract, the tape is not running ahead of reality; it is running behind it.
The second source is the inventory record. As of August 27, weekly social inventory of lithium carbonate stood at 78,802 tonnes, a drawdown of 7,590 tonnes in a single week. A 7,590-tonne weekly draw is not noise; it is roughly a 9% weekly draw against the standing stock. Combined with the price move, the two sources point the same way: the market is not just holding price — it is being genuinely drained.
The third source is the demand schedule, and this is where the tape and the inventory get their explanation. September lithium iron phosphate (LFP) output is scheduled to rise above 60,000 tonnes, and August LFP output was already up about 5% month-over-month. Put the three together — price up, inventory down, output scheduled higher — and the triangulation is coherent: destocking is happening into rising cathode output.
The earnings number that needs a second look
Now add the fourth document, the one that has been making headlines: Tianqi Lithium’s half-year report, published August 27. Revenue of 12.242 billion yuan, up 153.32%; net profit of 4.242 billion yuan, up 4,925.46%. A 4,925% profit swing is the kind of number that writes its own narrative — but cross-referencing it against the supply side changes what it means. The core variables in that earnings swing are not demand surprises; they are supply events: maintenance at the main production areas and a heavily delayed restart at the Jianxiawo pit. That is a supply-side profit, and it is important to label it correctly.
Here is where I would hedge the interpretation. A 4,925% profit jump driven by supply disruption and delayed restarts is real money, but it is not the same signal as a demand-led boom. If demand were the driver, you would expect the earnings swing to be corroborated by end-market pull — and the September LFP schedule does provide that corroboration. But the scale of the swing, relative to the demand data, suggests the supply side is doing most of the work. That is the difference between a rally you can extrapolate and one you have to monitor for the restart date.
Let me think about the structure of the earnings number more carefully, because a 4,925% jump is easy to quote and hard to read. The revenue grew 153.32%, which is a strong but not extraordinary number for a lithium producer in a recovering market. The profit grew 4,925% because the base was tiny — the company was coming off a near-breakeven or loss-making comparison period. So the percentage is partly a base effect and partly a genuine re-leveraging to higher prices. Both are real; neither is a demand signal on its own. That is the kind of decomposition the headline omits and the dossier should preserve.
What the dossier says about the rally’s durability
Separate the signal from the noise, and the signal is this: lithium is in a destocking-driven price recovery, supported by scheduled output increases, with the supply side — maintenance and a delayed restart — as the swing factor. The durability of the rally depends on variables that are all, in principle, observable: whether inventory keeps drawing at 7,590-tonne weekly rates, whether September output actually prints above 60,000 tonnes, and when the delayed restart at Jianxiawo lands. The first two are in the public record; the third is the known unknown.
What is unverified in this picture is precisely the piece the market is most tempted to believe: that the +4,925% profit number proves a new demand era. It does not. It proves that lithium supply is tight at the margin, that cathode makers are restocking into higher output, and that the companies with exposure to the supply side of that equation are printing outsized numbers. Demand is real but modest — a 5% monthly step in LFP output — while the profit swing is forty-nine-fold. The arithmetic itself tells you which side of the equation moved the needle.
Let me check the cross-references once more, because a dossier is only as good as its weakest cross-check. The futures tape and the inventory count are measured in the same week, which removes the dating problem. The output schedule is forward-looking, which introduces a different risk — schedules change. So the strongest evidence is the pair that is already recorded: price up and inventory down in the same window. The output schedule is the corroborating lead, not the proof. If September output prints below 60,000 tonnes, the destocking story still holds on its own terms; it just loses its forward fuel.
There is one more tension worth recording. The restocking into LFP output is happening against a supply side that is partly out of the market on maintenance and restart delay. That means the drawdown is flattered by the supply disruption; it is not purely a demand pull. A hedged reading acknowledges both forces and refuses to give all the credit to demand. The bull case and the supply case are not mutually exclusive — they are just different weights in the same equation, and the honest analyst keeps the weights visible.
The discipline of updating the dossier
Let me think about the updating rule, because a dossier is only useful if it has an explicit mechanism for revision. The rule I would set is simple: the read stands until one of the three primary data points contradicts it. If the next weekly inventory print shows a build instead of a draw, the destocking premise weakens and the rally loses its anchor. If September output prints well below 60,000 tonnes, the forward fuel is removed even if the current tightness persists. If the Jianxiawo restart is announced with a firm date, the supply-side floor begins to erode before the restart actually lands — markets price the schedule, not the completion. Each of these triggers is observable, dated, and cheap to track. That is what a hedged position looks like operationally: a hypothesis with named falsifiers.
The other discipline is not to let the 4,925% headline rewrite the dossier. The percentage is an artifact of a low base as much as a measure of strength, and the supply-driven core of the swing is established by the company’s own disclosure of maintenance and restart delay. A reader who wants to separate signal vs noise in this market keeps the demand and supply components in separate columns and refuses to blend them into a single bullish or bearish mood. The market is tight because supply is short and output is scheduled higher; the two facts coexist without either one proving a new demand era. That is the precise reading, and precision is the entire value of the dossier.
The final check is to triangulate the forward view against the current one. If the current read is right — tight market, real destocking, supply-led prices — then the forward view is simply the same read with a restart date attached. Nothing in the current data argues for a demand collapse, and nothing argues for an unconstrained rally. The honest conclusion is a tight market that will loosen on a schedule nobody can yet name. Hold that read, keep the falsifiers listed, and let the data do the updating. That is the whole discipline, and it is the difference between an analysis and a narrative.
A note on what the price already says
One more cross-reference before the close, because the price data carries information that the inventory and output numbers do not. The 4.77% single-day move and the spot premium above the futures contract both say the same thing: physical buyers are reaching for material faster than the paper market can confirm. That is a stress signal, not a comfort signal — it is the price counterpart of a weekly 7,590-tonne draw. When spot outpaces the contract, the market is telling you that the people who physically need the metal are paying a premium for timing, and timing premia are exactly what appear at the beginning of a repricing, not at its end. That is the kind of observation that does not fit a bullish or bearish headline; it is just where the tape points, and it is consistent with the rest of the dossier.
The final discipline is to hold all four documents — price, inventory, output, earnings — in the same frame and refuse to let any one of them rewrite the others. The price says demand is pulling. The inventory says supply is short. The output says the draw is being restocked. The earnings say the profit is mostly supply-driven. Read together, they describe a tight market with a supply-led price floor and a scheduled relief valve whose date is unknown. That is the whole picture, and it is cross-checked from four directions. No single source holds; the dossier does.
How to read the dossier going forward
For a reader who wants cross-checked facts rather than narratives, the dossier closes as follows. Price: up 4.77% on the day, spot above contract. Inventory: 7,590 tonnes drawn in one week. Output: September scheduled above 60,000 tonnes. Earnings: +4,925%, but supply-driven at its core. Triangulate the three primary sources and the direction is clear — the destocking is real and the price is responding to it. Hedge the magnitude, because the swing factor is a restart date no one can pin down, and that is the variable that will decide whether this is a seasonal bounce or the beginning of a longer repricing.
The watch list, then, is three data points with dates attached: the next weekly inventory print, the September output figure as it lands, and any announcement from the Jianxiawo pit about a restart schedule. The first two are scheduled; the third is not. That asymmetry — two known dates and one unknown — is exactly where the uncertainty lives, and it is the honest place to keep attention. No single source holds. But three sources — price, inventory, and output — agreeing on the direction of a drawdown is enough to read the market as tight. The honest position is to hold that read while flagging the one source that cannot be cross-referenced: the restart date that would unwind the tightness at speed. That is the signal, and that is the noise to keep watching.