This is the third and final article in a series examining base case commodity price assumptions in mining feasibility studies versus prevailing spot prices. Part one covered copper, while part two covered gold. The lithium analysis draws on 180 base case price assumptions from studies published between 2011 and 2026. These have been normalized on a lithium carbonate-equivalent basis and benchmarked against an average global CIF lithium carbonate price. Of the three commodities, it presents the most volatile and complex picture, one where the relationship between assumptions and spot prices has swung dramatically in both directions within a decade.

➤ The assumed price in studies has been more volatile and has reached 105% above the spot price, the largest premium in all the commodities in this series.

➤ Anchored assumptions set during high-price periods get stranded when the lithium spot price moves faster than any through-the-cycle framework can absorb.

➤ Average 2025 study assumptions sit at more than double the current spot price, requiring significant adjustment before treating published economics as a guide to value.

Study type composition and caveats

Of the 180 lithium studies in the dataset, 80 (44.4%) are preliminary economic assessments, 36 (20.0%) are prefeasibility studies, 60 (33.3%) are full feasibility studies and just four (2.2%) are mine plans. The high proportion of full feasibility studies in this analysis compared to our copper analysis reflects the wave of advanced-stage lithium project development that occurred between 2017 and 2023, as the battery supply chain race intensified. The small absolute sample size — one to three available studies per year, particularly in the early years — indicates that year-over-year comparisons should be treated with caution. This is because the assumptions of a single large project can materially influence individual year averages.

Feasibility study price assumptions are often set months before publication, meaning published studies can lag turning points in the spot market, and because spot prices are measured using annual averages, some publication-date timing effects are unavoidable, especially in years of rapid price movement. That noted, the dataset tells a compelling story about how hard it can be determine suitable feasibility study prices for a commodity as volatile as lithium.

2011-15: Anticipating the rise

In the early years of the dataset, lithium was still a stable industrial commodity, with lithium carbonate prices ranging between $4,755 per metric ton and $5,899/mt. The few studies from this period — three in 2011, one each in 2012 through 2015 — used assumptions that ran modestly above spot prices, ranging 7%-55% higher. The directional bias toward optimism was consistent with companies already anticipating the structural demand shift from electric vehicles and energy storage, pricing projects to reflect expected long-run equilibrium, rather than a spot price widely viewed as temporarily depressed.

2016-20: First rally and its aftermath

The lithium price surge of 2016-18, driven by rapidly growing EV battery demand and constrained hard rock supply, pushed the spot price to $15,861/mt by 2018. Study assumptions during this period lagged 13%-19%, a moderate level of caution similar to gold's behavior during its bull market, though the sample remains thin relative to the gold and copper datasets. What distinguishes lithium from copper and gold is what happened next. When the spot price fell sharply from its 2018 peak, reaching $10,651/mt in 2019 and collapsing to $6,935/mt in 2020, study assumptions did not follow. Studies in 2019 averaged $14,240/mt (34% above spot) and $12,383/mt in 2020 (79% above spot). This is much larger than any divergence in the copper or gold datasets. This can be explained by anchoring: Companies that had initiated projects during the high-price period were publishing studies with assumptions set before the correction and were apparently unwilling — or unable, given the project cycle — to write down assumptions to reflect a spot price that had more than halved. As the spot price began recovering from its 2020 trough, the anchoring bias persisted: 2021 studies averaged $16,868/mt against a spot average of $13,665/mt, a 23% premium that reflected continued optimism as the market began its next ascent.

2022 spike and 2024-25 correction: From one extreme to another

The 2022 lithium price spike, where lithium carbonate averaged $57,557/mt — nearly four times the 2021 level — produced the largest single-year change in any of the three commodities analyzed. Study assumptions of $21,508/mt represented a 63% discount to spot. The conservatism was rational: No company could credibly embed $57,000/mt into a long-life mine model. But by 2023, as the spot price fell to $38,338/mt, assumptions had risen to $28,511/mt. This was still a 26% discount, reflecting appropriate caution about whether elevated prices were sustainable. The subsequent collapse in lithium prices — $12,385/mt in 2024 and $10,059/mt in 2025 — has created the most extreme optimism in the dataset. Studies published in 2024 used average assumptions of $23,993/mt, 94% above prevailing spot. By 2025, the premium had widened to 105%, with assumptions of $20,594/mt sitting at more than double a spot price of $10,059/mt. Full feasibility studies in this period, where data coverage is better, show a similar pattern. Project economics that looked viable during the 2022-23 high-price environment are now being published into a market that has fundamentally repriced the commodity.

Before attributing the 2025 premium entirely to anchoring, another interpretation deserves consideration: incentive pricing. At $10,059/mt, spot sits below the marginal cost of much of the new supply that consensus demand forecasts require by the early 2030s, particularly higher-cost hard-rock, lepidolite and emerging African production. A base case near $20,000/mt may therefore reflect not only backward-looking inertia but also a forward-looking judgment about the long-run price needed to clear the market and incentivize capacity. For a study modeling a 15-to-20-year mine life, anchoring entirely to a transient supply-driven trough could be less rational than using a higher long-run price.

The distinction matters. Anchoring is a behavioral error; incentive pricing is a defensible market view. In practice, the 2025 premium is likely a blend of both, and the two are difficult to separate empirically. Crucially, however, neither interpretation removes the timing risk. Even if $20,000/mt proves to be the correct long-run incentive price, projects earn market prices, not long-run averages, during their early operating years.

If oversupply persists through construction and into ramp-up, the impact on net present value can be severe, regardless of where prices eventually settle. This risk is increased by the speed of lithium's supply response. As there are many projects but few producing mines — and because lithium mines can be built or ramped up faster than gold or copper operations — the reaction to a price rally is unusually rapid. Therefore the elevated prices needed to incentivize new projects tend to trigger oversupply.

A tentative cross-section by deposit type suggests brine and clay-hosted projects adopted higher base-case assumptions than traditional pegmatite hard-rock projects. This partly reflects timing — brine and clay studies cluster in the more recent, higher-assumption period — but the priced sample by geology is too small to isolate a pure geological effect with confidence.

Lithium vs. gold and copper — a different problem

Comparing lithium to the other two commodities in this series reveals a fundamental difference. Gold and copper assumptions are conservative during bull markets and converge toward spot during stable or declining periods — a pattern that reflects a market where long-run price expectations are well-anchored. Lithium assumptions oscillated more wildly, and with far greater amplitude, to 105% above spot in 2025 from 63% below spot in 2022, a 168-percentage-point swing within three years. By comparison, gold's largest gap was a 27% discount at the 2011 bull-market peak, while copper's widest conservative-to-optimistic reversal — to a 31% premium in 2016 from a 52% discount in 2006 — spanned a decade. Lithium covered a larger range in three years than copper did in 10 years. This is not a failure of industry discipline so much as a reflection of a commodity that moves faster, further and less predictably than any conventional through-the-cycle pricing framework can absorb. For investors, the practical implication is that the gap between study assumption and spot price in lithium can be a source of significant upside as in 2022 and significant downside risk as in 2024-25 within a timescale shorter than the construction period of the projects being assessed.

The 2011-15 experience offers an important counterpoint. Companies that priced above spot in that period anticipated a structural demand shift that had not yet been priced into the market, and thus were vindicated. The current situation is different in a critical respect. The EV demand thesis has already played out in market prices, producing the 2022 spike and the subsequent correction to current spot levels. The 105% premium in the 2025 study assumptions, therefore, does not anticipate an unpriced structural shift; it resists a spot price that has already incorporated it.

Treating lithium feasibility study economics as a reliable guide to project value at prevailing market prices requires far greater adjustment for the assumption than is typically applied in copper or gold analysis. In practice, this means two things. First, one should rerun project net present values at current spot and not the study's base case as the primary valuation anchor and treat the published assumption as an optimistic scenario rather than a central one. Second, one should apply a materially wider price sensitivity band than is standard for copper or gold, spanning at least from the current spot price to the study base case. This would then capture the range of outcomes the dataset has shown as being plausible within a single construction cycle.