Markets

Supply-Demand Models

For most commodities, supply and demand balance on a timescale of months. For critical minerals, the timescale is years or decades - a mine takes 10–17 years from discovery to production, while EV adoption can accelerate far faster than that. Modelling this asymmetry is one of the central analytical challenges in commodity economics today.

Avg mine development timeline

16 yrs

Discovery to first production, USGS data

IEA lithium demand range by 2040

3–10×

Current demand, across NZE vs STEPS scenarios

Projects that miss initial schedule

~60%

Historical analysis of large mining projects

China share of Li-ion processing

>70%

Binding midstream constraint in most models

Lithium Market Scenario Explorer

Adjust the key variables that drive lithium supply-demand models and see how the projected 2030 market balance shifts. This is a simplified illustrative model - real analyst models contain hundreds of asset-level inputs, but these four drivers explain most of the variance.

Adjust scenario inputs

40%
20% (slow)70% (fast)
45%
20% (NMC dominant)70% (LFP dominant)

LFP uses ~40% more Li per kWh than NMC but eliminates Co/Ni

60%
30% (delays/slippage)90% (on schedule)

Reflects the % of announced projects that deliver on time

5%
0% (not deployed)20% (transformative)

Direct lithium extraction from brine - nascent but watched closely

Projected 2030 market balance

Surplus / Deficit

- kt LCE

Balance signal

-

Deep deficit Balanced Large surplus

Projected balance 2025–2030 (kt LCE)

2030 demand

-

kt LCE

2030 supply

-

kt LCE

Incentive price signal

-

Illustrative model only. Base values calibrated to approximate 2024 analyst consensus (IEA APS, BMI base case). All figures in thousand tonnes lithium carbonate equivalent (kt LCE). Not for investment or commercial use.

Reading a Cost Curve

The supply cost curve is one of the most powerful tools in commodity economics. It ranks every source of supply by the cost per tonne of producing it - and the intersection with demand determines the marginal (price-setting) producer.

Stylised lithium cost curve (illustrative)

Hard rock (spodumene) Brine Clay / other
$30k $22k $14k $6k Demand Marginal cost ~$21k/t Cumulative supply volume →

Cost ($/t LCE)

Infra-marginal producers

High-quality brines (Atacama, Salar) earn supernormal margins. They supply even at prices well below today's incentive price.

Marginal producers

The intersection of the demand line with the cost curve sets the market-clearing price. These producers earn near-zero economic rent.

High-cost / unbuilt

Clay and undeveloped projects need higher prices to be viable. They become "incentive" supply if demand grows past current capacity.

The Six-Stage Modelling Process

How professional analysts build a supply-demand model from first principles, and the two stages where most forecast errors originate.

01

End-use decomposition

Demand

Break total demand into application segments (EVs, grid storage, consumer electronics, industrials). Each segment is modelled separately with its own adoption curve.

02

Technology & intensity assumptions

Demand

Apply mineral intensity per unit (kg Li per kWh, kg NdPr per motor). Model chemistry mix evolution - e.g. NMC→LFP shift reduces Co/Ni, thrifting lowers kg/kWh over time.

03

Asset-level supply database

Supply Common error source

Catalogue every producing mine and credible project with nameplate capacity, ramp schedule, probability weight, and unit cost. Historical data shows systematic optimism bias in ramp timelines.

04

Processing & midstream

Supply Common error source

Mine output ≠ market supply. Track conversion capacity (e.g. spodumene → Li hydroxide, cobalt hydroxide → sulfate) separately - often the binding constraint on effective supply.

05

Calculate market balance

Balance

Subtract total supply from total demand year-by-year. Positive = surplus (price pressure downward). Negative = deficit (price pressure upward, risk of shortfall).

06

Cost curve & incentive price

Price

Rank supply by marginal cost. Long-run price floors at the cost of the marginal tonne needed to balance supply. Deficits signal price must rise to incentivise new capacity.

Why models consistently overestimate supply

Analysis of historical project forecasts across mining shows a consistent pattern: announced project timelines are over-optimistic by an average of 2–4 years, and capital costs underestimated by 30–60%. Modellers must apply probability weights and schedule adjustments to avoid "paper supply" - capacity that exists in a database but not in the ground. Projects in early-stage permitting typically receive 10–30% probability weights in rigorous models; those with completed feasibility studies and secured financing receive 70–90%.

Who Builds These Models

Six institutions produce the most widely cited supply-demand models for critical minerals, each with different methodologies, coverage, and access models.

IEA International Energy Agency
Intergovernmental

Focus minerals

Energy transition minerals: Li, Co, Ni, Cu, REEs, graphite

Flagship product

Critical Minerals Market Review (annual)

Methodology

Top-down scenario + bottom-up asset database

Scenarios published

NZEAPSSTEPS

Strength

Policy credibility, broad scenario range, free access

Limitation

Annual cadence; less granular on specific assets

BMI Benchmark Mineral Intelligence
Private consultancy

Focus minerals

Lithium-ion battery supply chain: Li, Co, Ni, graphite, Mn

Flagship product

Lithium-ion battery & EV forecast

Methodology

Bottom-up cell-to-mine flow model

Scenarios published

BaseBullBear

Strength

Highest granularity on battery chemistry & cell demand

Limitation

Subscription only; battery-focused rather than all minerals

Wood Mac Wood Mackenzie (incl. Roskill)
Private consultancy

Focus minerals

Broad metals & mining coverage; all critical minerals

Flagship product

Metals & Mining Research Suite

Methodology

Asset-level databases + integrated price models

Scenarios published

BaseAcceleratedDelayed transition

Strength

Widest asset coverage; long track record via Roskill

Limitation

Premium pricing; some minerals less granular than BMI

USGS U.S. Geological Survey
Government agency

Focus minerals

All minerals on U.S. critical minerals list

Flagship product

Mineral Commodity Summaries (annual)

Methodology

Production statistics + basic balance estimates

Scenarios published

None - primarily historical statistics

Strength

Free, authoritative production/reserve data

Limitation

No forward projections; 1-year lag on data

S&P Global S&P Global Commodity Insights
Private data & analytics

Focus minerals

Metals, mining & battery materials

Flagship product

Battery Raw Materials Service

Methodology

Asset-level supply + sector demand modelling

Scenarios published

BaseHigh EVPolicy stress

Strength

Integration with price assessments and financial data

Limitation

Heavy enterprise focus; expensive

CRU CRU Group
Private consultancy

Focus minerals

Steel, aluminium, base metals, battery materials

Flagship product

CRU Market Outlook series

Methodology

Bottom-up cost curve + balance models

Scenarios published

BaseHighLow

Strength

Strong cost-curve methodology; steel/alloys depth

Limitation

Less coverage of specialty REEs and minor metals

Scenario Analysis and the Uncertainty Problem

The IEA's projections for lithium demand in 2040 vary by a factor of three or more depending on which scenario is used. This is not a failure of modelling - it is an honest representation of genuine uncertainty. The energy transition's speed depends on policy continuity, cost trajectories, consumer behaviour, and geopolitical conditions that no model can reliably forecast a decade out.

This is why responsible supply-demand modelling focuses less on the central forecast and more on the key uncertainties. Which variables drive the widest range of outcomes? What conditions would need to be true for a supply deficit to emerge by 2028? What is the fastest-plausible path for DLE technology to relieve brine supply constraints? Structuring analysis around these questions produces more decision-relevant insights than a single point forecast.

For market participants, the practical implication is that any supplier of critical mineral analysis claiming narrow forecast confidence bands should be viewed with scepticism. See Benchmarks and PRAs for how price reporting data feeds into these models, and Trade Flows and Customs Codes for how trade statistics are used to calibrate and validate supply estimates.