In Q4 2025, 31% of NEM intervals priced negatively (NEM: Australia's National Electricity Market) — but the surplus concentrates by region and by hour, not uniformly across the system. The usual reading treats negative prices as an inefficiency to smooth away with storage and transmission. Inari's analysis instead reads them as feedstock, and asks where that feedstock is most capturable. On these conditions, the framework surfaces dispatch-flexible (able to ramp output up or down to match available power) rare-earth magnet processing — an industry that grows more sensitive to electricity the further downstream you go — as a leading candidate. This is one coherent read: the NEM is the surplus problem, and the analysis traces where surplus turns capturable.
Australia is a nation with grid friction: renewable build has run ahead of flexible demand, storage and transmission, so daytime wholesale prices increasingly collapse to zero or below — 31% of NEM intervals negative in Q4 2025, but concentrated by region (Pilbara, SA, QLD) and by hour. Marginal Loss Factors (MLF — charges that vary by grid location) and effective delivered price (p_eff) rewrite bankability (how financeable a project looks to lenders) node-by-node, so industrial siting beats national averages. The analysis ranks dispatch-flexible downstream processing highest on energy-leverage — and the rare-earth magnet chain is the clearest candidate the framework surfaces: mining/oxide can't arbitrage curtailment (surplus power that gets thrown away), NdPr alloy is the inflection, NdFeB magnet preforms are where leverage concentrates. The binding constraint is capital coordination, not technology. Figures are ordering only.
In Q4 2025, 31% of NEM intervals priced negatively (AEMO, the Australian Energy Market Operator). But negative pricing does not arrive evenly — it concentrates in particular regions and particular hours. A national waterfall is the wrong instrument at continental scale; the regional one is the right one. Pilbara, South Australia, and Queensland each present a structurally distinct productive-surplus profile, and the bankability picture rewrites at the node, not the nation.
Two physical variables drive that rewrite. Marginal Loss Factors scale the value of energy by where it is delivered, and the effective delivered price (p_eff) — what an industrial load actually pays after losses and shape — is the number that decides whether a process is sited or shelved. The consequence is blunt: industrial siting matters more than national averages. A modelled-scenario read on the NEM is a map of where and when surplus is real, not a single headline figure.
Negative-priced intervals are not an accounting nuisance to be smoothed away — they are time-localised clean power with no productive home. Storage and transmission shift that energy across time and space, but their value saturates as spreads collapse. Industrial conversion does not: it can absorb large volumes of curtailed power without erasing the price signal. The open question the analysis addresses is which industries can absorb it — and in what order.
This section is Inari's own techno-economic analysis — first-principles, not a survey of outside evidence. The framework and the ordering are the high-confidence core; every magnitude is illustrative and qualitative only. The load-bearing finding: energy-leverage rises non-linearly downstream along the rare-earth value chain, and the ordering suggests only the later stages can convert price volatility into capturable advantage.
The logic is simple. A process captures cheap power only if margins are sensitive to the electricity price (high electricity intensity) and it can modulate without ruining yield (dispatch flexibility). Margin sensitivity is just the negative of electricity intensity, so the more electricity-intensive a stage, the harder it pulls on a falling price. Mining and oxide separation fail both tests; the metallurgical stages pass them — and pass them harder the further down the chain you go.
| Stage | Electricity in cost · dispatch | Reading |
|---|---|---|
| Mining & beneficiation | ~5–10% · diesel-led · rigid ORDERING | Electricity is a minor cost; a 100% price swing moves total cost by <5%. Cannot arbitrage curtailment — it provides atoms, not leverage. |
| NdPr oxide separation | ~3–6% · steady-state · rigid ORDERING | Solvent-extraction chemistry needs >70% capacity factor (CF — the share of maximum possible output actually delivered, on average); frequent stops degrade the process. Chemistry-constrained, not energy-leveraged — a platform, not a destination. |
| NdPr alloy reduction | ~10–20% · batch · schedulable INFLECTION | High-temperature batch reduction; electricity becomes a top-tier cost and furnaces can be timed to cheap windows. Viable at 40–60% CF — the first stage where surplus genuinely shifts viability. |
| NdFeB magnet preforms | ~15–25% · batch · dispatch-optimised MAX LEVERAGE | Strip-cast, jet-milled, aligned/pressed/sintered — extreme electricity intensity, batch-dispatchable, high value density, priced on magnetic performance not as a commodity. Viable at 30–50% CF. |
This is the same lesson as Niue, in a different key. Nothing here needs an invention — strip casting, jet milling, vacuum sintering are all known processes. What binds is capital coordination and the willingness to build assets that deliberately run below baseload utilisation (below the flat, round-the-clock output level a plant would normally run at). Conventional feasibility studies assume continuous, high-utilisation operation, and so systematically misprice exactly the modular, dispatch-responsive, electricity-intensive stages this regime rewards.
The mispricing has a physical basis. For chemistry-dominated assets (oxide), wear runs on calendar time, so idling buys nothing. For the downstream metallurgical stages, wear runs on operating hours and thermal cycles — so deliberate, curtailment-aligned under-utilisation can extend effective asset life and lower sustaining capital. A plant designed to pause is not a broken baseload plant; it is a correctly-designed flexible one. The studies that assume 85–95% utilisation never see it.
So the constraint is rarely the engineering. It is the coordination, the capital structure, and the rules — whether an investor will fund a modular plant that runs when power is cheap and rests when it is not, and whether the market rewards that flexibility as the grid asset it is.
Resolve MLF and effective delivered price per candidate region — Pilbara, SA, QLD read differently — so capacity lands where and when curtailment is real, not where a national average suggests.
Anchor on the energy-sensitive stages (alloy → magnet preforms) where curtailment converts to margin, rather than stopping at mining or oxide where it cannot. The long chain is the strategic asset.
Model low capacity factors as the design point, not a failure — extended asset life and reduced sustaining capital are features of curtailment-aligned operation, missed by continuous-operation assumptions.
Priority access to curtailed-energy zones and pricing that rewards flexibility turn an industrial load into part of the response to the surplus problem, not a subsidy line.
The window is path-dependent: as storage, demand electrification, and transmission mature, curtailment narrows — but physical over-generation precedes institutional adaptation, and industrial capital, once deployed, locks in advantage. Finite, but sufficient.
First-instance regional calibration is in progress against AEMO and ABS (Australian Bureau of Statistics) data; the deeper feasibility-grade read is available under engagement, not as a finished public artefact. Where a calibration is not yet validated, this page says so — the firm does not present in-flight work as completed work, and it does not commit to a publication date it cannot control.
The rare-earth analysis is offered as ordering and direction: which stages can absorb curtailment, in what sequence, and why the binding constraint is coordination rather than technology. The framework and the ordering are the high-confidence core. The magnitudes are illustrative. Because Australia is a nation, there is no absolute national "prize" on this page — and no bankable return at any stage.
Negative-priced intervals are unallocated industrial input. The fix is not only more storage — it is industry redesigned to consume power when it is abundant and pause when it is not.
Energy-leverage rises non-linearly downstream. Mining/oxide can't arbitrage; NdPr alloy inflects; NdFeB preforms peak. That sequence is the high-confidence claim — the dollar magnitudes are not.
Nothing here needs new technology. What binds is the willingness to fund modular plant that runs below baseload — exactly the case conventional feasibility studies misprice.
Under the F1 honesty rule, a continental jurisdiction gets ordering and direction with method and caveats — never a bald absolute prize, never a bankable return on a public surface.
Investors, sovereign clients, and DFIs can engage now for pre-publication access to the regional calibration and the full stage-resolved rare-earth analysis. The framework is the same; the node-level data is being assembled. The bankable instrument layer is shared under NDA at calibrated fidelity.
Pre-publication engagements run under confidentiality. Standard procurement applies.