Lithium Price Forecast
and Market Trend Analysis

English-language sellside coverage of this was thin. Bernstein gave it a paragraph. CRU did a real analysis. Goldman may not have touched the permitting mechanics at all. The reason is probably that the story lives inside Chinese local administrative law, which is not where London commodities desks spend their time. Fair enough. But 17% of global supply is not a footnote-level topic, and the gap between what CRU published and what the rest of the sellside published is, frankly, an embarrassment for an industry that charges what it charges for research.

The permit mechanism is the whole story. Without understanding it, the 17% number is just a number. With understanding it, the implications for supply elasticity extend years beyond the current disruption.

Lithium is classified as a strategic mineral under Chinese law. Extraction of strategic minerals goes through the Ministry of Natural Resources in Beijing. Slow process, rigorous, comparable in timeline to what Western jurisdictions require. Galaxy's James Bay project in Quebec took six years of permitting. Capex ran around $10,500 per tonne of annual LCE capacity.

There was a regulatory ambiguity. Ore grading below 0.4% Li₂O could be classified as a non-strategic industrial mineral. Things like kaolin, or what the permits actually said, "lithium-bearing ceramic stone." These permits come from local bureaus, not the Ministry. Yichun lepidolite runs around 0.3% Li₂O. Below the threshold. CRU was explicit that the companies were not breaking the law. They were operating within a genuine ambiguity in the regulatory framework.

Jianxiawo got its local permit in about two years. Capex around $5,500/t annual LCE. Two years instead of six, $5,500 instead of $10,500, and you can see why Yichun exploded in 2022 when lithium carbonate hit 600,000 yuan per tonne. CRU describes "hundreds" of EV supply chain companies entering the area. CATL. Gotion High-Tech. A long tail of smaller operators, many of whom had never been in the mining business before.

Nobody mines lepidolite when lithium is cheap. The mineral is low grade, 0.2-0.5% Li₂O, compared to 6% for the spodumene concentrate coming out of Australia. The extraction process is ugly. BMO's analysts were harsh about Yichun: dust covering the hillsides permanently, waste infrastructure grossly inadequate. Roughly 18 tonnes of concentrate input per tonne of lithium carbonate output. Roughly 23 tonnes of solid waste. Roughly 13 kg of hydrofluoric acid. When lithium was at $10/kg none of this penciled. At $70/kg everything penciled, including the environmental damage.

On July 1, 2025, the revised Mineral Resources Law took effect. All strategic mineral approvals centralized under the Ministry. The ambiguity that had allowed the local-permit pathway ceased to exist.

The Yichun bureau told eight lepidolite operators to resubmit resource verification reports because the minerals on their permits didn't match what was being extracted. Combined capacity of those eight: 160,000 tonnes of lithium carbonate per year. Then Jianxiawo's permit expired August 9 and wasn't renewed. Bank of America estimated that single mine at 5-6% of global production. Then the 27 cancellations in December.

Among the 27: one permit held by Jiangxi Special Electric Motor, who filed an objection. Four held by Gao'an Mining Development, a subsidiary of Yichun Mining. Yichun Mining is the state-owned entity that the municipal government set up in 2013 to consolidate the city's lithium resources. The government's own vehicle had permits being pulled. That detail says something about how broad the regulatory sweep is.

And during the transition, which could take years, every individual permit renegotiation plays out at the municipal level in Yichun, subject to local political dynamics that are opaque to Chinese market participants in Shanghai, let alone to anyone reading a CRU PDF in London. The risk premium from this uncertainty doesn't resolve quickly.

The connection to the GFEX futures market is where the Yichun story turns into a price story.

GFEX launched lithium carbonate futures in July 2023. Before the contract, lithium pricing was bilateral negotiation between converters in Jiangxi and Sichuan and their cathode customers. Information traveled slowly through the supply chain. After the launch, the whole market watches one screen, and the speed of price discovery collapsed from weeks to seconds.

Fastmarkets projects about 1,500 tonnes LCE deficit for 2026. That number means almost nothing for price forecasting because of the GFEX amplification. A small physical deficit passing through a futures market where financial positioning dwarfs physical flow can produce a price response appropriate for a deficit ten times larger. Or it can sit dormant for months and reprice in a single week.

The Yichun overhang and the GFEX mechanism interact. The 17% in regulatory limbo is a large risk of uncertain magnitude and uncertain timing. That is exactly the kind of risk that financial markets handle badly, and exactly the kind of risk that gets amplified rather than dampened by a futures market trading against a small physical base.

Australian spodumene mines. Several that have been operating seven to ten years show grade drift from 1.8-2.0% Li₂O down toward 1.2-1.5%. Sellside models hold mine output flat until closure. The reality is slow erosion of output per tonne of ore processed, slow increase in unit cost. Not dramatic enough to make a headline, significant enough over a decade to matter for aggregate supply.

Storage demand. Grew 71% year-on-year in 2025 per Benchmark. Storage is already roughly a quarter of global battery demand, maybe 35-40% in the US within a few years. EVs are still 72-80% of total lithium consumption, that hasn't changed and won't soon. What matters for price dynamics is that storage procurement arrives in pulses. Provincial grid tenders dump large volumes of demand into the spot market within single quarters. In the second half of 2025 multiple Chinese provinces launched standalone storage projects simultaneously, and a single quarter's incremental demand exceeded Europe's entire annual storage lithium consumption. A 100 MWh station uses 60-80 tonnes of lithium carbonate. The spot market's oscillation between weeks of tightness and sudden availability maps onto these pulses. EV demand by contrast is smooth and plannable months in advance.

AI data centers need 2-4 GWh of battery backup per 500 MW campus. Growing fast, undercounted in models.

Sodium-ion is commercial. CATL's Naxtra line at 30 GWh, global capacity past 50 GWh by year-end 2026. Cell cost $55-70/kWh, well below LFP. Energy density is the constraint. At 150-175 Wh/kg sodium-ion handles storage and two-wheelers fine but cannot do a 600 km passenger car that needs 250 plus. So the substitution ceiling on lithium pricing exists only in storage and low-end transport. When carbonate stays above roughly $25,000-30,000/t, those segments start migrating. Long-range EVs and aviation have no alternative at the required energy density and pay whatever lithium costs.

Bernstein's $15,000 long-term number implicitly reflects the sodium ceiling on the storage side.

Zimbabwe suspended all lithium concentrate exports in February 2025 after shipping 1.128 million tonnes the prior year. China's battery export VAT rebates drop from 9% to 6% in April 2026, full elimination January 2027, adding 6-13% to export costs. EU Battery Regulation mandates carbon footprint disclosure from 2027 and caps from 2030. Roughly 65% of lithium salt processing is in coal-powered China. Tesla's Texas refinery is operational. All of these push lithium salt production costs up. Global lithium resources exceed 100 million tonnes LCE. The rock in the ground is not scarce. Processing capacity, concentrated in China, is the actual constraint, and Tesla and Stellantis investing in refining directly is a signal about where competitive advantage in this industry is heading.

Spent battery volumes are projected to go from 1.4 GWh in 2020 to 112.3 GWh by 2030. China allowed black mass imports from August 2025. A large share of retired batteries get diverted to second-life applications because the margins are better than recycling. Retired battery volume and recycling feedstock volume are different numbers, and confusing them inflates recycling supply projections. The EU classifies black mass as hazardous waste from November 2026, banning export to non-OECD countries. DLE at roughly 11% of current production, under 3% of projected demand through 2035, lab recovery rates 90%+ dropping to 70-80% in industrial operation. Not a supply factor in the 2026-2028 window.