Lithium has staged a sharp, supply-driven 2026 rebound — battery-grade carbonate is back to ¥163,750/t (+38% YTD) — but the converter cash cost now sits above spot. The bull case is a genuine 2026 shortage; the brake is a still-ample global resource base waiting to restart.
Lithium enters mid-2026 having staged one of the sharpest rebounds in the metals complex — and, unusually, a supply-driven one. After the 2022 super-spike (battery-grade carbonate above ¥500,000/t in late 2022) and the brutal 2023–2025 collapse (a trough near ¥60,000/t in 2025), as of 9 June 2026 SMM's battery-grade lithium carbonate has recovered to ¥163,750/t (+38% YTD) — having touched a two-year high near ¥200,500/t on 13 May 2026 before pulling back — the GFEX main contract to ¥168,460/t (+38.6% YTD, with the futures running above ¥200k intra-month), and spodumene SC6 (CIF China) to US$2,428/t (+57% YTD). The defining feature is that the marginal catalysts sat on the supply side; demand's strong growth was the continuing backdrop, not the new variable.
Three supply shocks did the work. First, Zimbabwe's concentrate export ban, accelerated to take effect on 25 February 2026 (pulled forward from 2027), removed a meaningful slice of the marginal African feedstock that had become roughly 15% of China's spodumene imports. Second, persistent restart delays at the Jianxiawo lepidolite mine (Yichun) and an industry-wide round of Jiangxi mining-licence renewals tightened domestic raw-material availability. Third — and this is the mechanism that gives the rally its floor — spodumene rose faster than carbonate through the spring, pushing the purchased-spodumene cash cost to ¥167,877/t, above the carbonate spot price, so that non-integrated converters buying ore on the market are now running at a spot loss. SMM's monthly balance flipped to a deficit of −4,891 mt in May 2026, and SMM judges the full-year 2026 market to be in shortage, with May–June destocking.
This report uses SMM's proprietary price, cost, output-by-route, inventory and battery-demand datasets, cross-verified against the exchanges, customs data and the publicly observable market consensus, to make five arguments. (1) The rebound is supply-driven at the margin: the Zimbabwe ban and the Jiangxi disruptions were the catalysts that tipped the balance, with demand's strong growth the continuing backdrop rather than a new variable — which makes the rally more fragile than a demand-led recovery, because the global resource base remains ample and idled Australian capacity can restart. (2) The cost-curve floor is currently binding: with the marginal (purchased-ore) converter under water, loss-driven curtailment of marginal supply is the price's first line of support. (3) Inventory is low — under a month of sales, about three weeks — which amplifies the price reaction to any marginal shock in either direction. (4) Demand is two-speed: EV is decelerating from exceptional rates while energy-storage (ESS) is booming, and LFP chemistry has entrenched its dominance (~85% of cathode tonnage). (5) 2026 is a shortage year, but a thin one: the deficit is small, the futures-spot divergence around the ¥200k threshold is a warning, and the path depends on whether the disruptions persist or the restarts arrive.
For the participants at Singapore International Ferrous Week's Green Energy Metal Forum, lithium is the purest energy-transition metal in the room — and the one whose price has been the most violent. Its supply chain runs from Australian and African hard rock and South American brine, through Chinese chemical conversion, into the LFP and ternary cathodes that power the world's EVs and, increasingly, its grid storage. Singapore and ASEAN sit at a pivotal point in that chain: Indonesia is emerging as a battery and lithium-chemical hub, GFEX has become the world's price-discovery venue, and the green-finance and trade flows that fund the build clear through the region. The chapters that follow lay out the data, the scenarios and the actionable conclusions.
The evidence base is SMM's proprietary lithium and battery data complex: daily battery- and industrial-grade carbonate, hydroxide, spodumene and lepidolite assessments; the GFEX main contract and registered warrants; the SMM purchased-spodumene and purchased-lepidolite cash-cost models; monthly output split by feedstock route (spodumene, lepidolite, brine, recycling); the weekly and monthly inventory survey (smelter, downstream, total); the monthly supply-demand balance; and battery-chain demand series (LFP, ternary, ESS cells, total Li-ion cells, sodium-ion cells, NEV sales) — cross-verified with GFEX, customs data and the publicly observable market consensus. Every headline figure ties to an SMM indicator ID or a traceable public source.
Lithium moves from Resource → Conversion → Cathode → Cell → End-use, fed by three competing resource routes — hard-rock spodumene, Chinese lepidolite and South-American brine — with a growing recycling loop closing the circle. The chemistry split (LFP vs ternary) and the end-use split (EV vs ESS) are where the demand story is decided.
Volumes are SMM monthly figures for May 2026; shares and conversion ratios are SMM indicative estimates. LCE = lithium carbonate equivalent.
Lithium has rebounded almost 40% in 2026, with supply events as the marginal catalyst and strong demand growth as the backdrop. A small deficit, a binding cost floor and low inventory define a market that is tight — but fragile.
The lithium price chart of the last four years is the most violent in the base- and battery-metals complex: a super-spike to above ¥500,000/t for battery-grade carbonate in late 2022, a three-year collapse to a trough near ¥60,000/t in 2025, and now a sharp 2026 rebound. As of 9 June 2026, SMM's battery-grade carbonate has recovered to ¥163,750/t (+38% YTD), the industrial grade to ¥159,750/t, battery-grade hydroxide to ¥150,000/t, and the GFEX main contract to ¥168,460/t (+38.6% YTD). The spot benchmark spiked to a two-year high near ¥200,500/t on 13 May 2026 — with the GFEX futures running above ¥200k — before pulling back toward ¥168k. The crucial point for the forum is the cause: the marginal catalysts of this rebound were supply events, while demand's strong growth is the continuing backdrop rather than a new variable.
The evidence is in the balance. SMM's monthly supply-demand model flipped from a small surplus of +1,968 mt in April to a deficit of −4,891 mt in May 2026, even as total monthly output jumped to a record 113,295 mt (+57% year-on-year) as maintained spodumene lines restarted. In other words, demand grew faster than even a ~3.5× expansion of output could satisfy at the margin, and SMM's assessment of the full-year fundamentals is that 2026 is likely to be a shortage year, with continued destocking through May and June. This is not the 2022 mania — the absolute deficit is small — but it is a genuine turn from the multi-year glut.
Supply-catalysed: the Zimbabwe ban and Jiangxi disruptions were the marginal triggers, against strong continuing demand.
Cost-floored: the purchased-ore converter cash cost now sits above spot.
Low-inventory: under a month of sales cover amplifies every marginal shock.
Two-speed demand: EV decelerating, ESS booming, LFP dominant.
A thin shortage: a real but small 2026 deficit, knife-edge on the restarts.
The demand side is running just as hard. Apparent carbonate demand reached roughly 148,800 mt in May 2026, up about 58% year-on-year — slightly faster than output growth — so even a ~3.5× expansion of output off the trough has yet to catch up with the combined pull of end-use consumption and exchange-warrant accumulation. That is precisely why inventories keep drawing down: the supply shocks were the marginal catalyst, but they landed on a market whose demand was already growing at pace.
The market's internal signature is a futures-spot divergence around the ¥200,000/t threshold. Through the Q1–Q2 rally the GFEX futures led the physical spot higher: in mid-May the futures ran above ¥200k while the spot benchmark touched its own two-year high near ¥200,500 and then lagged. That divergence — speculative far-month positioning versus the slower-moving physical balance — is both a symptom of how thin and sentiment-driven the rebound is, and a warning: a sharp unwind of futures length could correct the price faster than the fundamentals alone would imply. The macro tape is also at work — expectations of an easier Fed path and commodity reflation have amplified allocation-style long participation on the futures leg, a portion of the rally that owes nothing to lithium's own balance.
It helps to frame lithium's last four years as a four-act drama, because each act was driven by a different force and the current act is the least understood. Act one (2021–2022) was a demand-shock mania: EV adoption outran a resource base that takes years to build, and carbonate ran past ¥500,000/t in a textbook commodity super-spike. Act two (2023–2024) was the supply response and demand normalisation: a wave of Australian, African and Chinese capacity arrived, EV growth decelerated from its fastest rates, and the price collapsed roughly 90% to the ¥60,000/t trough — well below the cash cost of much of the industry. Act three (2025) was the washout and curtailment: sustained sub-cost prices finally forced the marginal supply offline and drew down the glut. Act four (2026, the present) is the supply-driven rebound — a recovery built not on a new demand mania but on the withdrawal of marginal supply (Zimbabwe, Jiangxi) into a market that had quietly destocked. Reading 2026 as act four, rather than as a return to act one, is the single most important framing in this report: the volume keeps growing, but the price is governed by the cost curve and the disruptions, not by scarcity.
The GFEX–spot relationship is the live tell of where act four sits. The Guangzhou Futures Exchange carbonate contract, listed only in 2023, has rapidly become the market's primary price-discovery venue, and through the 2026 rally the futures led the physical spot — even as the spot benchmark set its own two-year high near ¥200,500 on 13 May, the GFEX futures ran higher still, a divergence the slower physical market did not fully follow. When futures lead spot in a thin, low-inventory market, the move carries a larger speculative component and a larger reversal risk; when spot leads futures, the move is better grounded in the physical balance. The chapters that follow trace the supply shocks that lit the rally (Chapters 2–3), the cost floor that underpins it (Chapter 4), the inventory that amplifies it (Chapter 5) and the demand that will ultimately decide whether the shortage deepens or dissolves (Chapter 6).
The rebound was lit upstream. Zimbabwe's export ban and Jiangxi's licence and restart disruptions tightened raw-material supply, and spodumene rose faster than the chemical it feeds.
Lithium's 2026 story begins in the rock. Roughly 58% of China's carbonate is made from spodumene (hard-rock lithium concentrate, mostly imported), and through the spring of 2026 the spodumene price ran up faster than the carbonate it produces. The SMM SC6 index (CIF China) is US$2,428/t (+57% YTD), the weekly assessment $2,497.6/t (+75% YTD), and the African and Zimbabwean grades have surged in parallel — the African index to roughly $2,298/dmt (+68% YTD) and Zimbabwe 6% to about $2,300/t (+67% YTD). When the input rises faster than the output, the conversion margin is squeezed — the mechanism behind the cost-curve floor in Chapter 4.
The single biggest catalyst was Zimbabwe's concentrate export ban. Accelerated to take effect on 25 February 2026 — pulled forward from a previously signalled 2027 — the ban is intended to force value-added processing onshore, and it removed a meaningful slice of the marginal feedstock that had grown to roughly 15% of China's spodumene imports. Africa had become the swing supplier: Zimbabwe's mines (Sinomine's Bikita, Huayou's Arcadia/Prospect, Yahua's Kamativi, Chengxin's Sabi Star) together run at roughly 124–134 kt LCE/year (SMM estimates, indicative), with Mali (Kodal/Hainan's Bougouni — a record ~11 kt in March — and Goulamina) ramping behind. The ban's execution pace and any quota carve-outs are the dominant upstream variable to watch: a strict, sustained ban is the clearest bull case; a quota workaround or slow enforcement is the clearest bear.
The second supply shock was domestic. China's lepidolite route (a lower-grade, higher-cost mica feedstock concentrated in Jiangxi) supplies about 12% of carbonate, and it has been disrupted by two forces: persistent restart delays at the Jianxiawo lepidolite mine near Yichun, and an industry-wide round of Jiangxi mining-licence renewals that idled capacity. Lepidolite prices reflect the tightness — the 2.0–2.5% grade at ¥4,975/mtu (+44% YTD) and the 1.5–2.0% grade at ¥3,540/mtu (+54% YTD). Because the Jiangxi lepidolite mines sit high on the cost curve, they are the swing domestic supply: their disruption tightens the balance now, and their resumption is one of the clearest bear triggers for H2 (the June pullback was partly driven by Jiangxi-resumption rumours).
The upstream synthesis is that 2026's tightness is built on two reversible disruptions — an African export ban and a Chinese licence/restart cycle — layered on a resource base that is, in aggregate, ample. That is what makes the rebound supply-driven and, therefore, conditional: the same upstream levers that tightened the market can loosen it. The third resource route, South-American and Chinese brine, plus recycling, is where the structural supply growth is coming from, and it is the subject of the next chapter.
Lithium's geography is the mirror of aluminium's bauxite question and iron ore's Australian–Brazilian concentration: a small number of resource jurisdictions feed a converter base concentrated in one country. Spodumene comes overwhelmingly from Australia (the long-standing low-cost hard-rock base) and, increasingly, from Africa — Zimbabwe first, with Mali (Goulamina, Bougouni), Namibia and others behind. Brine comes from the "lithium triangle" of Chile, Argentina and Bolivia plus the Chinese salt lakes of Qinghai and Tibet. The strategic vulnerability is that the marginal swing supply — African spodumene and the high-cost Jiangxi lepidolite — is precisely the supply most exposed to policy (export bans, licence renewals), which is why 2026's disruptions hit the balance so directly. A producing nation that wants to capture more value (Zimbabwe's ban, Indonesia's downstreaming) can move the global price simply by withholding concentrate, because there is little spare low-cost capacity to absorb the loss in the short run.
This concentration cuts both ways for the price. On the bull side, it means supply is genuinely policy-fragile and a single jurisdiction's decision can tighten the market fast. On the bear side, it means the high prices of 2026 are a powerful incentive for the very restarts and ramps — MinRes Bald Hill and Core Lithium Finniss in Australia, the Mali and Brazil projects, the Argentine brine expansions — that would refill the balance. The race between policy-driven withdrawal and price-driven restart is the upstream expression of the "thin shortage" thesis, and it is why this report treats the disruption-versus-restart status, not any single resource number, as the variable to watch.
The global supply pipeline beyond the 2026 disruptions reinforces the bear-refill side of the ledger. Australia remains the backbone — the Greenbushes, Pilgangoora and Wodgina complexes are among the lowest-cost hard-rock operations in the world and can lift volumes as price allows, while the suspended marginal mines (MinRes's Bald Hill, Core Lithium's Finniss) sit ready to restart. South America's brine expansions in Chile and Argentina add low-cost tonnes on a multi-year cadence, and new hard-rock provinces in Brazil and Canada are scaling. Africa, even with Zimbabwe's export restriction, continues to ramp at Mali's Goulamina and Bougouni and across a pipeline of projects that the high 2026 price actively incentivises. The aggregate message is that the medium-term resource base is not scarce: the projects exist, the orebodies are defined, and most of the question is the price and lead time required to bring them on. That is precisely why this report frames lithium as cost- and disruption-governed rather than scarcity-governed — and why the bull case rests on the timing of supply withdrawal against demand growth, not on any absolute shortage of lithium in the ground.
Chinese carbonate output has risen roughly 3.5× off the 2024 trough to a record 113 kt a month. The route mix is shifting: brine and recycling are the fastest-growing, lower-cost sources of new supply.
The scale of China's lithium-chemical build is easy to understate. Monthly carbonate output troughed near 32,000 mt in February 2024 during the price crash and has since risen roughly 3.5× to a record 113,295 mt in May 2026 (+57% year-on-year, with Jan–May cumulative output up 45%), as maintained spodumene lines restarted into the 2026 price recovery. China is, overwhelmingly, the world's lithium converter: this is where rock and brine from four continents become the carbonate and hydroxide that the cathode industry consumes. The route mix for May 2026 is instructive: spodumene 65,495 mt (58%), brine/salt-lake 20,740 mt (18%), lepidolite 13,730 mt (12%) and recycling 13,330 mt (12%).
The shift within that mix matters as much as the total. Year-on-year, recycling (+114%) and brine (+66%) are compounding fastest as structural, lower-cost supply; spodumene (+85%) rebounded hard off a maintenance-depressed year-ago base; and lepidolite (−24%) is the only shrinking route — the direct imprint of the Jiangxi disruptions. This is the structural supply story underneath the cyclical disruption: the marginal new tonne of Chinese lithium is increasingly a low-cost brine tonne (from Qinghai and Tibet salt lakes, and from imported South American carbonate) or a recycled tonne, not a high-cost hard-rock tonne. Over time that re-shapes the cost curve downward — a structural headwind to price even as the 2026 disruptions support it.
China's converters cannot run on domestic resource alone. Carbonate imports — largely finished chemical from South American brine — reached 32,650 mt in April 2026 (Jan–Apr cumulative ~116,000 mt, +47% year-on-year), with Chile supplying 21,385 mt in April (January-April cumulative imports up ~34% year-on-year) and accounting for roughly 61% of the Jan–Apr cumulative total, Argentina ~28% and Indonesia an emerging ~7%. Spodumene imports ran at ~758,000 physical mt in April (~63,000 mt LCE), with Australia alone above 350,000 mt (+38.9% month-on-month) as maintained mines restarted. The picture is of a conversion industry pulling hard on every available feedstock — a tightness signal that corroborates the deficit, but also a reminder that the resource is there to be mobilised when price justifies it.
The conversion stage is where China's grip on the lithium chain is tightest. The country hosts the overwhelming majority of the world's lithium-chemical conversion capacity, and the 2024–2026 restart cycle — output up roughly 3.5× off the trough — demonstrates how quickly that capacity can flex when price justifies it. But the industry is split by a divide that the 2026 cost squeeze has made decisive: integrated producers, who own their own spodumene, lepidolite or brine, versus non-integrated converters, who buy ore on the open market. Integrated producers have a far lower, more stable cash cost and can run through the cycle; non-integrated converters are exposed to the spodumene price and, in 2026, are the ones losing money (Chapter 4). The cost-curve floor is therefore set by the marginal non-integrated converter, and the strategic response across the industry — vertical integration into resource — is reshaping who survives the next down-cycle. For a "secure supply chains" forum, the lesson is that processing dominance and resource ownership are now being recombined under one roof.
| Feedstock route | May-2026 (mt) | Share | YoY & note |
|---|---|---|---|
| Spodumene (hard rock) | 65,495 | 58% | +85% · restart rebound, imported AU/Africa |
| Brine / salt-lake | 20,740 | 18% | +66% · structural, low-cost |
| Lepidolite (mica) | 13,730 | 12% | −24% · Jiangxi, high-cost swing |
| Recycling | 13,330 | 12% | +114% · circular, structural |
| Total | 113,295 | 100% | +57% · ~3.5× the 2024 trough |
The single most important support under the price is that the marginal converter is losing money: the purchased-spodumene cash cost now sits above the carbonate spot, forcing marginal curtailment.
If the disruptions lit the rebound, the cost-curve floor is what holds it up. The decisive fact of mid-2026 is that the purchased-spodumene cash cost — what a non-integrated converter pays to turn market-bought ore into carbonate — has risen to ¥167,877/t (+42% YTD), which is above the carbonate spot price of ¥163,750/t. In SMM's words, "spodumene gains exceeded carbonate, non-integrated enterprises suffer losses, spot profitability in deficit." The marginal producer is under water.
This is the textbook mechanism of a cost-supported price. When the marginal converter loses money on every tonne, the rational response is to cut run-rates or idle, which removes supply and supports the price — a self-correcting floor at roughly the marginal cash cost (currently ~¥165,000–170,000/t for the purchased-spodumene route). The lepidolite route, by contrast, retained a thin margin as processing fees rose with prices (purchased- lepidolite cash cost ¥150,688/t), but it is the high-cost Jiangxi mica capacity that is most exposed to any price relapse. SMM cites exactly this cost-curve squeeze — alongside the export ban and the Jiangxi disruptions — as the core reason prices have "held up well" with "limited downside."
The cost floor is a powerful medium-term support, but it is not a hard barrier — as 2025 proved, when prices crashed to ¥60,000/t, far below most converters' cash cost, before supply finally curtailed. A floor at marginal cash cost holds only if producers actually cut when they lose money; integrated producers (who own their own spodumene or brine) have far lower costs and can keep running, so the floor is set by the marginal purchased-ore converter, not the average. The cost-spot gap — how far spot sits below the purchased-spodumene cash cost — is therefore the single cleanest gauge of how much marginal curtailment the price is relying on.
The strategic read for the forum is that lithium in 2026 is a cost-supported, disruption-tightened market, not a demand-squeeze. That makes the floor durable as long as spodumene stays expensive and the disruptions persist — but it also means the floor moves down as the low-cost brine and recycling routes grow (Chapter 3) and as Australian and African supply restarts. A forecaster must track the cost base, the disruption status and the restart pipeline together; the flat price is the residual of those three.
It is worth picturing the cost curve itself, because its shape explains the price behaviour. At the low end sit the South American brine operations and the best Australian hard-rock mines, with integrated cash costs that can be well below ¥80,000–100,000/t LCE — these run regardless of price. In the middle sit the integrated spodumene converters. At the high end — the marginal supply that sets the clearing price — sit the non-integrated purchased-ore converters and the high-cost Jiangxi lepidolite, whose cash cost is the ¥150,000–168,000/t band that the 2026 price is now testing. The curve is steep at the top, which is why the price found support so quickly once it rose back into that band: a relatively small amount of marginal supply turns uneconomic over a narrow price range, and curtailing it tightens the balance fast. The same steepness works in reverse on the way down, which is why a price relapse can be violent once the marginal tonnes capitulate.
The 2025 episode is the cautionary precedent. Prices fell to ¥60,000/t — far below the cash cost of most of the industry — and stayed there for months before supply finally curtailed, because producers chase market share, honour offtake contracts, defer maintenance and hope for a rebound long before they idle. The lesson is that the cost floor is a medium-term attractor, not an intraday backstop: it tells you where the price gravitates once supply rationalises, not where it cannot go in a sentiment-driven overshoot. In 2026 the floor is doing its work because the disruptions have done the curtailing for the market; if those disruptions reverse, the floor will be tested again — and the brine-and-recycling cost-curve shift means the floor it is tested against will be lower than today's.
Lithium carries less than a month of inventory — roughly three weeks — and it has been destocking. A thin buffer magnifies the price reaction to any marginal supply-demand shock — in both directions.
The third force shaping 2026 is the inventory structure. SMM's weekly total inventory (in-factory plus GFEX warrants, excluding bonded and port stock) is 98,786 mt (−25% year-on-year), and the sample total in early May was around 103,000 mt against an inventory-to-sales ratio of under one month — about three weeks. That is a low-inventory structure, and SMM has warned explicitly that it amplifies the price elasticity to any marginal supply-demand shock: with under a month of buffer, a small change in the balance forces a large change in price to clear the market. The violent 2026 moves — both the rally and the ¥200k-then-pullback — are partly a function of this thin buffer.
The most striking inventory signal is the surge in GFEX registered warrants to 55,544 lots — a 9 June 2026 print, up ~66% on the nearest comparable year-ago observation (33,457 lots on 30 May 2025). Upstream plants have been hedging into the futures rally and registering physical material onto the exchange — a sign that, even as spot circulation tightened, producers were willing to deliver metal into the futures market to lock in the high prices. The warrant surge is thus a double-edged signal: it confirms physical availability flowing to the exchange (a bearish hint that the squeeze is not absolute), while also reflecting the producer hedging that a sharp, fast rally invites.
Disaggregating the monthly inventory shows where the metal sits — and why the May print needs careful reading. The monthly total rose month-on-month to 97,139 mt, but the jump is almost entirely a coverage effect: the survey added a trader/in-transit category (roughly 37,000 mt) to the sample from May. On a like-for-like basis, smelter stock (17,925 mt) plus downstream cathode-maker stock (42,018 mt) actually edged lower on the month; the headline total is broadly level with a year earlier — itself a coverage artefact. The structurally more informative number is smelter inventory, down about 68% year-on-year (55,391 → 17,925 mt): the destock is concentrated upstream — a truer tightness signal than the headline total. Taken together with the sustained weekly destock, it corroborates SMM's shortage read. Either way, the buffer is thin relative to a ~149,000 mt/month consumption rate. For a trading desk, the practical implication is that lithium will remain a high-beta, headline-sensitive market in 2026 — the thin inventory is the amplifier, the disruptions and restarts are the signal, and the cost floor is the backstop.
It is worth asking why lithium's inventory buffer is structurally thin, because the answer explains the volatility. Unlike a base metal such as aluminium — where exchange warehouses, bonded stock and a deep merchant trade hold months of cover — lithium chemicals are perishable in a commercial sense: battery-grade carbonate degrades in storage if not carefully handled, the chemistry specifications are tight, and the just-in-time cathode-and-cell supply chain is designed to minimise working capital rather than to hold strategic stock. The result is a chain that runs lean by design, with under a month of cover from converter to cathode-maker, and little of the speculative warehouse inventory that cushions other commodities. The arrival of the GFEX contract has begun to change this — registered warrants are, in effect, a new pool of exchange-visible stock — but the roughly two-thirds warrant surge is better read as producer hedging into a price rally than as a durable strategic buffer.
The implication is that lithium's price will keep over-shooting in both directions relative to the modest size of its supply-demand imbalances, because there is no inventory shock-absorber to smooth the adjustment. A deficit of a few thousand tonnes a month — trivial against a base metal's stock cover — forces a large price move in lithium precisely because there is no buffer to draw down. For risk managers, the takeaway is that lithium should be sized and hedged as a high-volatility exposure whose realised moves will routinely exceed what the fundamentals alone seem to justify; the thin inventory is not a temporary feature of 2026 but a structural property of a young, fast-growing, specification-tight chemical market.
Demand is firm but rotating. EV growth is decelerating from exceptional rates while energy-storage is booming, and LFP chemistry has entrenched a dominance that defines lithium demand for the decade.
Lithium demand in 2026 is firm but two-speed. The mature pillar — electric vehicles — is decelerating, not reversing: China NEV sales were 1,619,000 units in May 2026, up roughly 24% year-on-year — the apparent ~5% decline is an illusion of comparison against December 2025's seasonal peak-month high base (Jan–May cumulative sales are up 5.7% year-on-year) — with new-energy penetration around 63% of China's auto sales. That is still strong growth, but off a vast base and at a slowing rate. The emerging pillar — energy storage (ESS) — is booming: ESS cell output reached 79.32 GWh in May 2026 (+117% year-on-year), up roughly 6.6× from about 12 GWh in January 2023, within a total Li-ion cell output of 234.45 GWh (+65% year-on-year).
The EV deceleration deserves to be understood precisely, because it is easy to misread a slowing growth rate as a weakening market. China's new-energy penetration at roughly 63% of auto sales is extraordinarily high — there is simply less room for the percentage to keep doubling, so growth naturally moderates from the triple-digit rates of the early 2020s toward the high-teens-to-twenties that a 1.6-million-vehicle monthly run-rate (up ~24% year-on-year) still represents. That is maturation, not contraction. The lithium-relevant nuance is that each new vehicle is also, on average, carrying a larger battery as ranges extend and as LFP enables bigger, cheaper packs — so lithium demand per vehicle is rising even as unit growth slows. Layer the global EV market (Europe, North America, the rest of Asia) on top of China's, and the EV pillar remains a large, growing — if no longer explosive — source of carbonate demand. The deceleration matters for the rate of demand growth, which is what determines whether the thin 2026 deficit widens or closes; it does not threaten the level of demand.
ESS is the structural growth story the forum should focus on. Grid-scale storage is the natural complement to the solar and wind build, and — crucially — a large and growing share of ESS demand is ex-China and far less price-sensitive than the domestic market. SMM notes that above roughly ¥200,000/t some China ESS projects turn wait-and-see on internal-rate-of-return grounds, but ex-China ESS demand stays positive across a wider price range. That makes storage both a volume engine and a price stabiliser: it broadens the demand base and underwrites the upside that EV deceleration alone would not.
The single most important structural fact in lithium demand is the entrenchment of LFP (lithium iron phosphate) chemistry. LFP cathode output reached 489,100 mt in May 2026 (+74% year-on-year), roughly 5.5× the ternary (NCM/NCA) output of 88,950 mt (+38% year-on-year) — so LFP is about 85% of cathode tonnage and growing faster. Both power batteries and ESS are LFP-heavy, and the cost, safety and cycle-life advantages of LFP have made it the default for all but the highest-energy-density applications. This matters for lithium because LFP consumes carbonate (not hydroxide, the ternary feedstock), which is why the carbonate price and the GFEX carbonate contract — not hydroxide — are now the market's true benchmark.
The deeper significance of the ESS boom is that it changes the character of lithium demand, not just its level. EV demand is tied to the consumer auto cycle, to subsidies and to the China penetration curve that is now past 60% and flattening. Grid storage, by contrast, is tied to the build-out of solar and wind and to the economics of firming intermittent generation — a multi-decade investment programme with a different, more utility-like demand driver. Critically, ESS demand has a different price-elasticity: SMM observes that above roughly ¥200,000/t, some Chinese ESS projects turn wait-and-see because the higher cell cost pressures the project's internal rate of return, whereas ex-China ESS demand — where storage often competes against more expensive alternatives and is driven by reliability and policy rather than marginal cost — stays firm across a wider price band. The practical consequence is that the global ESS bid puts a higher, stickier floor under lithium demand than the China-only picture suggests, and it is the single best reason to believe the 2026 floor is durable even as EV growth slows.
The chemistry choice reinforces this. Because both power and storage are now overwhelmingly LFP, the marginal tonne of battery demand is a marginal tonne of carbonate demand — and LFP's cost and cycle-life advantages mean its share is unlikely to reverse. That is why the carbonate price, and the GFEX carbonate contract, have displaced hydroxide as the market's reference: the high-nickel ternary chemistries that consume hydroxide are now a minority, premium niche (long-range performance vehicles), while the volume of the energy transition runs on iron-phosphate and carbonate. A forecaster modelling lithium demand should therefore weight LFP/ESS far more heavily than the headline EV-unit number, which understates the structural pull.
The demand verdict for the forum: lithium demand is not in question — it is growing across both EV and ESS — but its composition is shifting in ways that favour carbonate over hydroxide and broaden the buyer base beyond the China EV cycle. The risk is not a demand collapse but a demand deceleration meeting a supply base that can grow quickly once price justifies it. That tension — firm, rotating demand against an ample, mobilisable resource — is what makes the 2026 shortage thin and the price path so dependent on the supply-side disruptions.
Sodium-ion is the substitution threat that is not yet a threat; recycling is the circular supply that is quietly becoming structural. Both shape the long-run lithium balance more than the 2026 print.
Two forces will shape lithium's longer-run balance more than any single year's disruption: the substitution threat from sodium-ion, and the recycling loop that returns lithium to the chain. Take substitution first. Sodium-ion batteries — which use no lithium — have attracted enormous attention as a potential lithium-demand cap, and their percentage growth is dramatic: cathode and anode output rose 90–154% year-on-year in spring 2026. But the absolute base is tiny. Sodium-ion cell output was just 0.6 GWh in May 2026, up from 0.01 GWh in January 2023 — which is under 0.3% of the 234 GWh of Li-ion cell output in the same month.
The honest read is that sodium-ion is not a near-term lithium-demand threat. Its upstream electrolyte salts (NaPF₆, NaFSI) are capacity-constrained, its energy density still trails LFP, and its economics only compete when lithium is expensive — so it acts as a soft ceiling on extreme lithium prices rather than a volume substitute at today's levels. Within sodium-ion, the polyanion cathode route is gaining over layered oxide. For the forum, the right framing is that sodium-ion is a genuine long-run hedge against lithium-price spikes and a niche for low-cost, low-energy-density storage — but it does not change the 2026 or 2027 lithium balance, and treating it as an imminent demand cap would be a forecasting error.
Recycling is the quieter, more consequential story. Carbonate from recycling reached 13,330 mt in May 2026 — about 12% of total output and more than doubling year-on-year (+114%) — making it the fastest-growing supply route outright: recycling's structural climb sits alongside brine's, while spodumene's +85% is a low-base rebound (Chapter 3). As the first large cohort of EV batteries reaches end-of-life and as LFP-scrap and black-mass processing scale, recycling becomes a structural, price-insensitive source of lithium that is also the lowest-carbon route — the lithium analogue of the scrap story in steel and aluminium. Over the decade, recycling plus brine re-shape the cost curve downward and reduce China's dependence on imported hard rock, which is both a supply-security gain and a structural headwind to the long-run price.
Substitution (sodium-ion) caps the upside at extreme prices; recycling and brine lower the floor over time. Between those two, the structural trajectory of lithium is toward a lower, less volatile price than the 2022 mania — which is exactly why a supply-driven 2026 rebound, rather than a demand-led one, is the right way to read this cycle. The energy transition guarantees the volume; it does not guarantee the price.
The recycling supply chain is worth understanding in more detail because it is the route most likely to surprise forecasters. Spent batteries and manufacturing scrap are mechanically processed into "black mass" — a concentrate of the cathode metals — which is then hydrometallurgically refined back into battery-grade carbonate or hydroxide. Two tailwinds are scaling it fast. First, the manufacturing-scrap stream alone is large and growing simply because cell production is growing: a fixed scrap rate on 234 GWh/month of cell output is a substantial, rising feed even before a single EV battery retires. Second, the first big cohort of EV packs sold in the 2018–2021 boom is now beginning to reach end-of-life, adding the post-consumer stream. Because LFP scrap is lower-value than nickel-rich ternary scrap, LFP-recycling economics are tighter, but policy support and the sheer volume of LFP in the fleet are pulling capacity in regardless. The strategic point is that recycling is a domestic, price-insensitive, low-carbon source of lithium that grows mechanically with the installed base — and over the decade it structurally reduces China's import dependence and lowers the cost curve, exactly as scrap reshaped the steel and aluminium balances.
On sodium-ion, the realistic roadmap is one of niche complementarity rather than displacement. The chemistry's advantages — abundant, cheap sodium feedstock and good cold-weather and safety performance — make it attractive for low-cost stationary storage and entry-level short-range vehicles, exactly the applications least sensitive to energy density. But its lower energy density rules it out of the mainstream long-range EV and most high-value applications, and its supply chain (electrolyte salts, hard-carbon anode) is still being built. The right expectation is that sodium-ion captures a slice of the cheapest storage and micro-mobility demand, growing fast in percentage terms from a tiny base, while LFP-lithium retains the overwhelming majority of the power and grid-storage market through the forecast horizon. Sodium-ion is best modelled as a price-capping option — it becomes economically compelling precisely when lithium spikes — rather than as a base-case volume threat.
SMM sees 2026 as a shortage year with limited downside; the market consensus is more cautious, and this report takes SMM's view as final. The path turns on the disruption-vs-restart race, with the cost floor below and the ¥200k threshold above.
The forecast debate is unusually two-sided, because the 2026 shortage is real but thin. SMM's view is that the full-year market is in shortage with continued destocking and prices should "maintain a relatively strong trend" on cost support, the tone softening by early June to "in the doldrums but with limited downside" as Jiangxi-resumption rumours circulated. The market consensus is more cautious: some public base cases still see a small nominal surplus for 2026, flipping to deficit only if the Jiangxi disruptions persist (aggregated public forecasts, unattributed). The divergence turns on how durable the disruptions are judged to be — this report takes SMM's view as primary (reasons below).
Two features of this divergence matter. First, SMM (a deficit) and parts of the market (a small nominal surplus) disagree on the sign of the 2026 balance — the market is genuinely on a knife-edge, hinging on the Jiangxi disruptions. Second, both agree the imbalance is small: the implied surplus on a multi-hundred-kt market is close to balance and flips on a single large disruption. This report takes SMM's deficit read as the final view — grounded in observed May destocking, the binding cost support and bottom-up route-by-route output data — with the anonymous surplus expectation used only as a bear cross-check: it is small enough that the SMM-observed disruptions plausibly close it.
Zimbabwe ban holds and quota execution stays slow; Jianxiawo + the Jiangxi licence shutdowns extend into Q3 (a potential −55 kt LCE); ESS demand (esp. ex-China) stays strong; low inventory amplifies the squeeze. A disruption-led re-test of the ¥200k high.
The cost floor and the thin deficit hold prices in a range; disruptions and restarts roughly offset; ESS growth absorbs decelerating EV. Spot oscillates around the purchased-spodumene cash cost; the futures-spot gap normalises.
Australian restarts (MinRes Bald Hill, Core Lithium Finniss) and the African/Mali ramp refill supply; Jiangxi resumes; the nominal surplus reasserts; China ESS softens above ¥200k on IRR pressure; a futures-length unwind overshoots the move lower, briefly through the cost floor as in 2025.
The base case breaks up if the Jianxiawo/Jiangxi disruptions extend through Q3 while ESS demand stays firm — a genuine, sustained deficit into a market holding under a month of inventory would re-test ¥200k. It breaks down if the Australian and African restarts arrive on schedule and Jiangxi resumes, restoring the nominal surplus — in which case the cost floor is the first line of defence, and a futures-length unwind could briefly overshoot below it as it did in 2025. The distinctive feature of lithium is that its low inventory makes it high-beta to the marginal tonne: a forecaster must track the disruption status, the restart pipeline and the cost-spot gap far more closely than any single price target.
Distilled to a single dashboard, five indicators carry most of the forward signal for the next twelve months. (1) The Jiangxi disruption status — the swing: confirmation that Jianxiawo and the licence-renewal mines stay offline through Q3 is the clearest bull trigger; a confirmed resumption is the clearest bear trigger. (2) The cost–spot gap — the floor's pulse: how far the carbonate spot sits below the purchased-spodumene cash cost (¥167,877) tells you how much marginal curtailment the price is relying on; a gap that closes (spot rising to meet cost, or cost falling) signals a healthier balance. (3) Spodumene SC6 — the input: a sustained move back below ~US$2,000/t would relieve the converter squeeze and remove the cost support. (4) Weekly inventory and GFEX warrants — the amplifier: a sustained faster-than-seasonal draw toward the ~90 kt range confirms the deficit is biting, while a continued warrant surge hints at physical availability flowing to the exchange. (5) ESS cell output, especially ex-China — the durable demand: continued double-digit growth underwrites the floor regardless of EV deceleration. A reader who tracks these five — disruption status, cost-spot gap, spodumene, inventory/warrants, and ESS — will understand the lithium cycle better than one watching the flat GFEX print alone. That, ultimately, is the analytical message of this report: lithium in 2026 is a market to be read through its supply disruptions and its cost curve, not its headline price.
Lithium is the purest energy-transition metal — and its supply chain increasingly runs through ASEAN. Indonesia is becoming a battery and chemical hub, GFEX sets the price, and Singapore finances the build.
The structural forces in this report converge on Singapore and its region. China dominates lithium conversion and battery manufacturing today, but the next phase of the supply chain is migrating toward ASEAN — and Indonesia in particular. Already an emerging ~7% of China's carbonate imports, Indonesia is leveraging its nickel-battery ecosystem (the cluster of leading Chinese and Korean battery investments around its industrial parks) to move into cells and, increasingly, lithium chemicals. The battery supply chain — from cathode to cell to pack to recycling — is becoming an ASEAN story, and the capital, the offtake and the trade clear through Singapore.
Lithium's import map underscores both the opportunity and the dependence. China — the converter to the world — relies on Chile (~61% of carbonate imports) and Argentina (~28%) for finished brine chemical and on Australia and Africa for spodumene. That concentration is a supply-security question of exactly the kind a "secure supply chains" forum is convened to address, and it is the reason resource-holders (Zimbabwe's ban, Indonesia's downstreaming push) are exercising more pricing and processing leverage. ASEAN's role is to become the neutral, well-financed midstream that diversifies and re-aggregates these flows.
Three Singapore franchises matter. First, price discovery and hedging: the GFEX lithium carbonate contract has become the world's price-discovery venue — the two-year high near ¥200,500 and the futures-spot divergence (with the futures leading) were centred on GFEX — and the international market increasingly needs Asian-hours access, a CME/LME hydroxide relationship, and the hedging infrastructure that a high-beta market with under a month of inventory demands. Second, green and trade finance: funding the ASEAN battery and chemical build, the recycling capacity, and the brine/spodumene offtake that underpins it. Third, the recycling and circular-economy supply chain, as end-of-life batteries become a strategic, lower-carbon resource across the region. For SMM, whose carbonate, hydroxide, spodumene and cost assessments increasingly underpin physical and financial contracts, Singapore is where the data becomes the infrastructure of the trade.
The timing is deliberate. Singapore International Ferrous Week convenes these conversations — across the Green Energy Metal Forum, the green-finance track and the supply-chain sessions — at the moment lithium is transitioning from a China-centric, EV-driven, hard-rock market to a more diversified, ESS-inclusive, recycling-aware and ASEAN-distributed one. A transparent benchmark relationship between GFEX and an international contract, a financing structure for ASEAN battery capacity, and a certification regime for low-carbon, recycled lithium are all products waiting to be built — and each is a natural fit for a neutral, deeply-capitalised hub between the Chinese, ASEAN and resource-holder poles of the market.
Indonesia deserves particular attention because it is the clearest example of how the battery supply chain is migrating. Having built a dominant position in nickel and nickel-based battery materials on the back of its ore reserves and an aggressive downstreaming policy, Indonesia is now extending the same playbook toward cells and lithium chemicals — anchored by the leading Chinese- and Korean-linked battery investments clustered around its industrial parks and by its emerging ~7% share of China's carbonate imports. The strategic logic mirrors the aluminium-to-ASEAN and iron-ore-to-Simandou migrations elsewhere in this conference: capacity and value-added processing move toward lower-cost, policy-supported geographies that sit closer to the demand centres of the future. For lithium specifically, an ASEAN midstream offers China's converters and the world's automakers a diversification away from a single-country processing chokepoint — precisely the supply-security objective the forum is built around — while giving resource-holders and host governments the downstreaming they increasingly demand.
The internationalisation of price discovery is the financial counterpart to that physical migration. GFEX has won the battle to be lithium's reference price, but a contract priced in renminbi and settled in China does not fully serve a global market that increasingly transacts across ASEAN, Australia, South America and Europe. The gap between a dominant onshore benchmark and a globalising physical trade is exactly the space a Singapore franchise can fill — through an Asian-hours, internationally-accessible relationship to GFEX, a hydroxide or LCE contract that complements it, and the clearing and hedging infrastructure a high-beta market with under a month of inventory demands. Whoever builds the transparent benchmark and the financing rails for the diversifying lithium trade captures a disproportionate share of its value, and the most credible candidate to host that infrastructure is Singapore.
The circular-economy dimension is the third leg of the regional opportunity, and the one most aligned with the forum's green agenda. As the global battery fleet grows, the end-of-life and manufacturing-scrap streams become a strategic, lower-carbon source of lithium, nickel and cobalt — and the logistics, processing and certification of that black-mass trade are a natural fit for a well-regulated regional hub. ASEAN's proximity to the cell and pack manufacturing migrating into the region, combined with Singapore's strength in trade finance, commodity trading and standards, positions it to host the recycling-and-certification layer that the low-carbon battery economy will require. A credible, transparent regime for certifying recycled content and carbon intensity — the lithium analogue of the green-aluminium premium discussed elsewhere at this conference — would let the region capture not just the volume of the battery build but the value of its sustainability. That convergence of physical migration, price discovery, green finance and circular-economy infrastructure is why lithium, more than any other metal at the forum, is an ASEAN-and-Singapore story in the making.
The 2026 lesson is that the marginal purchased-ore converter loses money: integration (owning spodumene, brine or recycling feed) is the route to surviving the cost-curve squeeze. Resource-holders gain pricing leverage; converters must secure low-cost, diversified feed and watch the disruption-vs-restart race.
Low inventory and the futures-spot divergence make lithium the highest-beta battery metal. The alpha is in the GFEX curve, the cost-spot gap and the disruption headlines — and in the basis between GFEX and any international contract. Singapore is where Asian-hours risk clears.
LFP dominance makes carbonate (not hydroxide) the key feed; ESS is the durable demand pillar, especially ex-China where it is less price-sensitive. Securing carbonate offtake and recycled content is now a commercial and ESG decision, and ASEAN is the emerging manufacturing base.
The ASEAN battery and chemical build, the recycling capacity and the brine/spodumene offtake all need capital — a Singapore green- and trade-finance franchise. The policy agenda is supply-chain diversification, a credible recycling/low-carbon certification regime, and transparent price discovery linking GFEX to the world.
Lithium in 2026 is a supply-driven rebound: a genuine but thin shortage, lit by Zimbabwe's export ban and the Jiangxi disruptions, floored by a binding purchased-ore cash cost, amplified by under a month of inventory, and pulled by a two-speed demand base where EV decelerates and ESS booms under entrenched LFP dominance. It is tighter than the glut years but more fragile than the 2022 mania, because the resource base is ample and the restarts are coming. The growth and the diversification migrate to ASEAN; the price is discovered on GFEX; and the finance, the recycling and the data clear through Singapore. For the Green Energy Metal Forum, lithium is the metal where the energy transition's volume is most certain — and its price the least.
| Indicator | Latest | Unit | Date | SMM / source ID |
|---|---|---|---|---|
| Battery-grade carbonate | 163,750 | ¥/t | 2026-06-09 | s20000222 |
| Industrial-grade carbonate | 159,750 | ¥/t | 2026-06-09 | s20021036 |
| Battery-grade hydroxide | 150,000 | ¥/t | 2026-06-09 | s20000218 |
| BG–IG carbonate spread | 4,000 | ¥/t | 2026-06-09 | j00123038 |
| GFEX main contract | 168,460 | ¥/t | 2026-06-09 | a10172596 |
| Spodumene SC6 (CIF China) | 2,428 | $/t | 2026-06-09 | s20021032 |
| African spodumene (high) | 2,298 | $/dmt | 2026-06-05 | s22802388 |
| Zimbabwe spodumene 6% (low) | 2,300 | $/t | 2026-06-05 | s22790662 |
| Lepidolite 2.0–2.5% | 4,975 | ¥/mtu | 2026-06-09 | s20157111 |
| Purchased-spodumene cash cost | 167,877 | ¥/t | 2026-06-09 | a10173890 |
| Purchased-lepidolite cash cost | 150,688 | ¥/t | 2026-06-09 | a10173889 |
| Monthly carbonate output | 113,295 | mt | 2026-05 | a10006555 |
| — Spodumene route | 65,495 | mt | 2026-05 | a10097731 |
| — Brine / salt-lake route | 20,740 | mt | 2026-05 | a10097733 |
| — Lepidolite route | 13,730 | mt | 2026-05 | a10097732 |
| — Recycling route | 13,330 | mt | 2026-05 | a10146287 |
| Supply-demand balance | -4,891 | mt | 2026-05 | a10006559 |
| Weekly total inventory | 98,786 | mt | 2026-06-04 | a10172022 |
| GFEX registered warrants | 55,544 | lots | 2026-06-09 | a12715139 |
| LFP cathode output | 489,100 | mt | 2026-05 | a10001859 |
| Ternary cathode output | 88,950 | mt | 2026-05 | a10001956 |
| ESS cell output | 79.32 | GWh | 2026-05 | a12715571 |
| Total Li-ion cell output | 234.45 | GWh | 2026-05 | a10179173 |
| Sodium-ion cell output | 0.6 | GWh | 2026-05 | a12763981 |
| China NEV sales | 1,619,000 | veh | 2026-05 | a10150318 |
| Chile carbonate imports | 21,385 | mt | 2026-04 | a10009660 |
| SMM scenario | 2026 carbonate range | Conditions |
|---|---|---|
| Bull (~30%) | ¥190,000–225,000/t | Jiangxi outage persists through Q3; destocking sustained faster than seasonal norms; overseas storage demand stays strong |
| Base (~50%) | ¥150,000–185,000/t | Disruptions and restarts roughly offset; spot oscillates around the bought-ore cash cost; the futures–spot basis normalises |
| Bear (~20%) | ¥120,000–150,000/t | Jiangxi restart confirmed; Australian/African restarts deliver; spodumene in sustained retreat below the ~US$2,000 band |
Volume series (output, inventory, demand, cathode) are converted from SMM's native metric tonnes (mt) to thousand tonnes (kt) for readable chart axes; battery output is in GWh. Daily price, cost and futures series were downsampled to one observation per month plus the latest ~45 days of daily prints for file size; the underlying SMM series are full-resolution. The import-source split and the price-band scenarios are SMM indicative figures; the band is illustrative.
| Lithium carbonate | The primary lithium chemical for LFP cathodes; SMM's battery-grade carbonate is the domestic benchmark (¥/t). |
| Lithium hydroxide | The lithium chemical for high-nickel ternary cathodes; less in demand as LFP dominates. |
| Spodumene | Hard-rock lithium concentrate (SC6 = 6% Li₂O grade); the largest feedstock route, mostly imported. |
| Lepidolite | A lower-grade lithium mica, concentrated in Jiangxi; a high-cost swing feedstock. Priced in ¥/mtu (metric tonne unit). |
| Brine | Salt-lake lithium (Chile, Argentina, Qinghai, Tibet); a low-cost route and the source of most carbonate imports. |
| LFP vs ternary | LFP (lithium iron phosphate) uses carbonate, dominates (~85%); ternary (NCM/NCA) uses hydroxide, high energy density. |
| ESS | Energy-storage systems (grid/stationary batteries); the fastest-growing demand pillar, LFP-heavy. |
| GFEX | Guangzhou Futures Exchange; its lithium-carbonate contract is the world's price-discovery benchmark. |
| LCE | Lithium carbonate equivalent; the standard unit for comparing lithium across chemical forms. |
| Sodium-ion | A lithium-free battery chemistry; a long-run price hedge, but under 0.3% of cell volume today. |
| Cash-cost floor | The marginal (purchased-ore) converter's cash cost; the price's first line of support when spot falls below it. |