Critical Mineral Midstream

The midstream of a mineral supply chain is everything between the ore body and the finished component: smelting and refining, chemical conversion, separation of chemically similar elements, alloying, and the qualification step by which a buyer certifies that a new supplier's material actually works in their product. It is the least visible segment and, for nearly every mineral the energy transition and the defence industrial base depend on, it is where the concentration actually sits. Mines are scattered across dozens of countries; the furnaces, solvent-extraction trains and conversion plants that turn their output into usable chemicals are not.

Established The concentration story in critical minerals is a midstream story, and it is getting worse at the aggregate level even as one segment improves. The IEA's 2026 outlook puts the average market share of the single top refined-material supplier across key energy minerals at 70% in 2025, up from 68% in 2020; excluding rare earths, where diversification has begun, the figure is 72% in 2025 against 70% in 2023. The 2025 edition measured the top three refining nations at 86% of the market in 2024, up from 82% in 2020, with China holding roughly a 70% average share across 19 of the 20 minerals analysed. Over three-quarters of all growth in refined supply between 2023 and 2025 came from just two countries: Indonesia for nickel and China for almost everything else.

This brief is the material-by-material atlas of that midstream: who refines what, where the separation capacity outside China actually stands, what the export-control weaponisation of 2023–2025 measurably did to prices, how long qualification takes, what substitution has and has not achieved, and what recycling can and cannot contribute on stated timelines. Its boundaries are load-bearing. What happens inside a battery cell — chemistry, cost per kilowatt-hour, the recycling-feedstock arithmetic — belongs to advanced battery technologies. Superconducting tape and its rare-earth deposition targets belong to high-temperature superconductors. What resource rents do to the exporting state belongs to resource economies. The accounting of material flows through economies belongs to industrial ecology. This brief owns the processing layer those four each touch and none owns: the tonnes, the plants, the percentages and the prices between mine and product.

The measured record, material by material, using the IEA's 2025 and 2026 outlooks as the spine because they are the only consistent public accounting across minerals.

Established Copper: the largest volume, the quietest concentration. China's share of global copper smelting rose from roughly 15% in 2005 to 50% in 2025 and accounts for about 90% of all smelting growth since 2005. Demand adds roughly 7 million tonnes by 2040 in the IEA's stated-policies case — the largest absolute addition of any mineral — against a projected supply shortfall of 25% by 2035, narrowed from the 30% projected a year earlier but still the biggest tonnage gap on the map. Smelting margins have been negative for many independent smelters as Chinese capacity overbuilds; treatment charges are the visible symptom.

Established Lithium: the price weapon is oversupply, not scarcity. Prices surged roughly eightfold during 2021–22, then fell more than 80% from 2023. The demand side kept growing throughout — about 25% a year over 2024–25 — but the price collapse cut lithium-specialist investment by roughly 40% in 2025 and lithium exploration spending by a similar fraction. China dominates the conversion of spodumene and brine into battery-grade chemicals; the mines are in Australia, Chile and Argentina, the conversion margin is not.

Established Nickel: the one case where the dominant refiner is not China — because Chinese capital moved the midstream to the ore. Indonesia, which banned nickel ore exports in January 2020, now leads refined nickel supply, with virtually all recent supply growth coming from Indonesian plants, the bulk of them Chinese-financed high-pressure acid-leach and rotary-kiln operations. The ore ban is the single most successful act of forced midstream relocation on record, and it relocated the midstream to Indonesia without diversifying its ownership.

Established Cobalt: quota-managed at the mine, concentrated at the refinery. The Democratic Republic of the Congo, the dominant miner, suspended cobalt exports in early 2025 and replaced the suspension with an export quota of 96,600 tonnes with a 10% strategic reservation; China holds roughly three-quarters of refining. The IEA's 2026 outlook widened its projected 2035 cobalt supply gap from about 15% to 25% specifically because of the DRC quota — the clearest case in the dataset of upstream resource nationalism, covered as a general phenomenon in resource economies, propagating directly into a midstream balance.

Established Graphite, manganese, gallium, rare earths: the over-90% club. For all four, the IEA projects the top refiner — China in each case — still above 90% of global supply in 2035 on current announcements. Graphite anode material was placed under Chinese export licensing from December 2023 and appeared again in the October 2025 battery-chain controls; high-purity manganese sulphate has announced projects covering only about 55% of expected 2035 demand.

Established Rare earths: the one measurable improvement, bought at policy expense. The top supplier's share of refined rare-earth supply fell from 90% in 2023 to 85% in 2025, and the IEA projects roughly 70% by 2035 — the only major mineral where concentration is moving down. The physical basis is thin: outside the top producer there is roughly 50 kt of mining capacity, under 40 kt of refining and separation capacity (mainly Malaysia and the United States), and only about 18 kt of downstream magnet-making capacity in rare-earth content terms. Lynas produced the first separated dysprosium and terbium outside China at its Malaysian plant in May and June 2025 (company's own reporting); MP Materials, under a July 2025 agreement with the US Department of Defense, took roughly US$400 million in preferred-equity investment, a guaranteed NdPr price floor of US$110/kg — about twice the then-prevailing Chinese price — and a commitment to a 10,000-tonne-a-year magnet facility around 2028, followed within days by a reported US$500 million magnet-purchase commitment from Apple (companies' own figures). The IEA prices full diversification of rare-earth supply at about US$60 billion over the next decade.

Established Gallium and germanium: the byproduct chokepoints where export controls were rehearsed. Gallium is recovered almost entirely from Bayer-process alumina refining liquor and germanium from zinc residues and coal by-products; neither supports a standalone mine. China introduced export licensing on both in July 2023, banned exports to the United States outright in December 2024, and the measured result is a split market: by 2025–26 European prices ran roughly five times Chinese domestic prices for gallium and heavy rare earths and roughly three times for germanium. USGS commodity data puts China near 98% of primary low-purity gallium output. The April 2025 controls on seven heavy and medium rare earths forced automakers outside China to curtail operations within weeks; the October 2025 extension to foreign-made products containing Chinese-sourced rare earths was suspended for one year, to November 2026, and the IEA estimates full implementation would place at risk on the order of US$6.5 trillion a year of downstream production outside China.

Established The investment record contradicts the urgency narrative. Critical-minerals investment fell 9% in 2025 — the first substantial decline since 2020 — with battery-materials companies down 20% and copper the exception at +8%. Exploration spending fell about 10%. What rose was public money: advanced-economy public finance commitments quadrupled from 2023 to reach US$65 billion in 2025. The private market, watching prices, is exiting the very segment governments declare strategic; the IEA's cost data explain why — projects outside established producers face capital costs 20% to over 150% higher and refining operating costs about 50% higher than in the incumbent countries.

The genuinely open questions are economic and institutional rather than geological.

Frontier Is midstream concentration an equilibrium or an artifact? The incumbent-cost view says China's 50–90% shares reflect real advantages — scale, integrated chemical parks, cheap capital, tolerant permitting — and will re-form after any subsidy wave, as the 20–150% capital-cost gap suggests. The policy-artifact view notes that several of those advantages were themselves built by two decades of state direction, and that the rare-earth share is already measurably falling under counter-policy. Both readings fit the data through 2026; the next price collapse discriminates between them.

Frontier Does the price-floor model generalise? The MP Materials arrangement is one plant, one product, one buyer of last resort. The IEA's own arithmetic finds price-support mechanisms roughly twice as effective per point of project return as upfront grants (2.4 versus 1.3 percentage points of IRR in its rare-earth refining example), but at a recurring fiscal cost, and nobody has yet run a floor through a full Chinese price war.

Frontier Do export controls accelerate or deter diversification? The 3–5× price divergence is simultaneously a subsidy to every non-Chinese entrant and a demonstration of what happens to demand when the incumbent chooses to squeeze. Which effect dominates is unresolved; the one-year suspension to November 2026 reads as the controller itself testing the answer.

Frontier Can byproduct metals be diversified at all? Gallium and germanium supply is capped by the alumina and zinc circuits they ride on. Western recovery retrofits — announced at alumina refineries in Quebec and Greece and zinc smelters in Australia and Tennessee — are demonstrations, not markets, and none had reached commercial scale by early 2026.

Frontier Where does qualification bind? The least-measured constraint in the whole subject: how long it actually takes a new refiner's output to be accepted into automotive, semiconductor and defence bills of materials, and whether buyers will pay the premium once the crisis headline fades.

Established The cost gap is the master bottleneck. Capital costs 20% to over 150% higher and operating costs around 50% higher outside the incumbent producers, per the IEA's project data, mean a non-Chinese refinery is structurally the marginal plant: first to close in a glut, last to attract private capital. Every other constraint operates inside that envelope.

Established Price volatility functions as an entry deterrent whether or not it is wielded deliberately. Lithium's post-2022 collapse of more than 80% erased the investment case for Western conversion plants faster than any policy could rebuild it; battery-metals investment fell 20% in 2025 while demand grew 25–35%. A market where the incumbent can absorb three loss-making years and the entrant cannot is concentrated by finance, not geology.

Established Byproduct entanglement caps several strategic metals. Gallium rides alumina, germanium rides zinc and coal, tellurium rides copper anode slimes, indium rides zinc. Supply cannot respond to price without moving the host industry, and the host industries have themselves concentrated in China — which is how a country with no unusual gallium geology came to hold roughly 98% of primary output.

Frontier Qualification is the hidden clock. Industry-reported ranges run one to three years for an automotive supplier to qualify a new magnet or cathode source and longer for defence and semiconductor applications; a refinery that mechanically completes in 2027 sells qualified volume years later. The number is poorly documented in public sources, which is itself a finding.

Frontier Chicken-and-egg feedstock loops. A magnet plant needs separated oxide; a separation plant needs offtake from a magnet plant; neither closes without an anchor customer, which is why the working Western examples — MP with the Pentagon and Apple, Lynas with Japanese offtake since 2011 — are all anchor-built.

Established Permitting and residue management. Rare-earth separation produces thorium-bearing residues; Lynas's Malaysian licence has been contested on exactly this for over a decade, and its US separation project has been slowed by wastewater permitting. Solvent-extraction trains are chemically intensive in ways that rich-country regulators price and incumbent jurisdictions historically did not.

Frontier Process workforce. Hydrometallurgical and separation-plant expertise concentrated where the plants are; the West's operating cadre is one generation thin. Widely asserted, weakly quantified.

Frontier Price discovery outside Chinese benchmarks. Most strategic-minor-metal prices are assessed, not traded; the 5× European–Chinese gallium divergence of 2025–26 shows a single global reference price no longer exists. Western contracting, stockpile release rules and floor mechanisms all need indices that do not yet reliably exist, and building them is a data problem before it is a market problem.

Established Capacity data are an estimate wearing a number's clothes. Chinese refining capacity, utilisation and stocks are opaque; the IEA's shares are best-available reconstructions, and every 2035 projection in this brief inherits that uncertainty. Traceability work — proving where a refined tonne actually came from once it is blended — is a live research programme, not a solved problem.

Frontier Separation chemistry beyond solvent extraction. Hundred-stage mixer-settler trains are the incumbent for rare earths; ionic-liquid, chromatographic, membrane and bioleaching alternatives promise smaller footprints and faster permitting but none has displaced solvent extraction at commercial scale. Progress here would shorten the single longest build item in any diversification plan.

Frontier Substitution science at the materials level. The iron-nitride and ferrite magnet programmes, silicon-anode blending, and sodium-ion chemistry are the research inputs that would let demand walk away from chokepoints; their cell-level and device-level status is tracked in advanced battery technologies. What this brief needs from them is one number each: performance per unit of avoided chokepoint material.

This subject runs on natural experiments; the laboratory is trade data.

Frontier The decisive test is already running: whether the rare-earth top-supplier refining share, 90% in 2023 and 85% in 2025, continues down to the roughly 70% the IEA projects for 2035, or reverts the first time Chinese prices collapse and the subsidised plants must sell into them. Rare earths are the only mineral where policy-built midstream capacity — the MP price floor, Lynas heavy-rare-earth separation, US$65 billion of public finance — is large enough relative to the market to show up in the concentration statistics. The scoreboard dates are fixed: the November 2026 expiry of the suspended October 2025 controls and the 2028 magnet-facility target. If the share is still falling in 2030 with the floor untested or survived, the policy-artifact reading of concentration wins; if it has reverted, the incumbent-cost reading wins and every other mineral's diversification plan needs repricing.

Frontier The gallium–germanium split market is a controlled experiment in byproduct economics. A sustained 3–5× Western price premium is the strongest entry subsidy these metals have ever seen. Whether any of the announced Western recovery retrofits reaches commercial sale by 2028 measures whether byproduct supply responds to price at all — a parameter no model currently has.

Frontier The DRC cobalt quota tests whether upstream cartelisation can discipline a concentrated midstream. The IEA already moved its 2035 cobalt gap from 15% to 25% on the announcement; the measurable outcome is whether refined-cobalt pricing power actually shifts from refiner to quota-setter by 2030.

Established The recycling test has a date. Announced battery-recycling capacity for 2030 is roughly seven times the feedstock that will exist to feed it, a collision documented in advanced battery technologies; the 2030 utilisation figures will separate the recyclers with feedstock contracts from the ones with press releases, as the Li-Cycle insolvency already previewed.

Speculative The unrun experiment is a coordinated buyer. Demand aggregation across allied economies — a joint purchasing floor rather than national ones — is proposed in the IEA's policy work and has never been tried at scale for minerals.

Established Separation is staged, not single-step, and the stages are the timeline. Rare-earth separation by solvent extraction runs mixtures through on the order of hundreds of mixer-settler stages because adjacent lanthanides differ minutely in chemistry; heavy rare earths need more stages than light, which is why dysprosium and terbium capacity outside China arrived years after NdPr and only in 2025. A greenfield separation plant is a three-to-seven-year build before qualification starts.

Established Purity specifications define the product. Battery-grade lithium carbonate, 4N-to-6N gallium for semiconductors, magnet-grade oxides: each step up in purity is a separate plant section and a separate qualification. Producing a 99% material is mining; producing 99.999% reproducibly is the midstream, and the premium lives entirely in the last decimals.

Established Magnet-making is its own chain, not an appendix. Strip casting, hydrogen decrepitation, jet milling, pressing, sintering and grain-boundary diffusion sit between separated oxide and a motor magnet; the roughly 18 kt of non-Chinese downstream capacity is the thinnest link in the rare-earth chain, thinner than separation.

Established Byproduct recovery is a retrofit discipline. Gallium circuits bolt onto Bayer liquor loops; germanium recovery bolts onto zinc residue treatment. The engineering is known — the incumbent did it at scale — but each retrofit is bespoke to its host plant, which is why announcements outrun commissioning.

Frontier Energy intensity anchors geography. Smelting and conversion follow cheap power and reagents; Indonesian nickel and Chinese graphitisation sit where coal-fired energy and acid are cheapest. Any Western buildout inherits a structurally higher energy and compliance cost per tonne, which engineering can narrow but not erase.

Established Four briefs border this one, and the borders are unit-of-account rules. Advanced battery technologies owns the cell: chemistry shares, cost per kilowatt-hour, and the recycling-feedstock arithmetic this brief cites but does not re-derive. High-temperature superconductors owns REBCO tape and already documents rare-earth deposition-target supply as a lead-time constraint — a live example of a tiny midstream tonnage gating a strategic industry. Resource economies owns what extraction rents do to states, including the DRC and Indonesian instruments whose midstream consequences are recorded here. Industrial ecology owns the flow accounting — where the tonnes go once embedded in the economy — on which every recycling projection in this brief ultimately rests.

Established Substitution evidence lives at the borders. The measured substitution wins are chemistry-level facts from the battery world: lithium-iron-phosphate, free of nickel and cobalt, passed 55% of EV batteries deployed globally in 2025, and sodium-ion exists at about 1% of lithium-ion manufacturing capacity. The magnet-free and reduced-dysprosium motor work sits with the automakers; the speculative iron-nitride path is a materials-science bet still far from sintered-NdFeB energy products.

Frontier The circular-economy claim is adjudicated next door. Circular economies and the IEA's recycling work agree on the ceiling: secondary supply already covers 33% of copper and 31% of nickel demand, but recovery reaches only about 20% for lithium, collection rates run 40–50% in Europe and North America against roughly 1% in Africa, and average recycling rates for key energy minerals are projected to rise only from about 10% today to 20% by 2040. Recycling relocates midstream leverage late, and only where collection works.

Established The EU wrote targets without money. The Critical Raw Materials Act (in force 2024) sets 2030 benchmarks — 10% of consumption mined domestically, 40% processed, 25% recycled, and no more than 65% of any strategic raw material at any processing stage from a single third country — but created no dedicated fund; it is a permitting accelerator and a monitoring regime attached to member-state budgets.

Established The US bought equity and a floor. The Department of Defense's 2025 MP Materials arrangement — preferred equity, a US$110/kg NdPr floor, guaranteed offtake — is the most interventionist mineral-market instrument any Western government has deployed in decades, and the IEA's arithmetic (a price floor delivering roughly 2.4 percentage points of project IRR against 1.3 for grants of comparable annualised cost) suggests it is also the efficient one. Its fiscal exposure is open-ended by design: the state is now short the Chinese price.

Established Japan built the reference institution two crises ago. JOGMEC's stockpiles and equity stakes date from the 2010 rare-earth embargo, financed Lynas when no private lender would, and are the reason Japan's magnet supply chain weathered 2025 better than Europe's — the standing counterexample to the claim that diversification is impossible, and the proof that it takes a decade.

Frontier Stockpiles are the untested layer. Advanced-economy strategic stocks for gallium, germanium and rare earths are being assembled with US$65 billion of public finance behind the broader effort, but release rules, sizing and coordination are unpublished or undecided; a stockpile without a release doctrine is a price subsidy to the next panic.

Frontier The exporter's institution matters as much. China's licensing bureaucracy — approvals case by case, end-use declarations, the one-year suspension lever — is a calibrated-pressure instrument, not an embargo machine; treating it as either pure signal or pure blockade misreads the 2025 record.

Established The externalities are concentrated where the midstream is. Artisanal cobalt mining in the DRC, with documented child labour; Indonesian nickel's deforestation, tailings disposal and coal-fired power; thorium-bearing residues at rare-earth plants from Baotou to the contested Lynas site in Malaysia. The 50% operating-cost gap between incumbent and entrant jurisdictions is partly a regulatory-stringency gap — which means some fraction of every cheap refined tonne is an unpriced harm borne elsewhere.

Frontier Diversification can be externality relocation in either direction. Moving separation to OECD jurisdictions internalises residue management at OECD cost; moving it to lower-income partner countries under friend-shoring labels risks reproducing the original bargain with new flags. Which pattern dominates the US$65 billion public-finance wave is not yet measurable.

Established Producer countries have their own ethics case. The DRC quota and the Indonesian ore ban are attempts to capture midstream value where the ore and the harms are, and Indonesia's succeeded on value-capture terms; the governance consequences of such rents are the subject of resource economies, and the honest statement here is that consumer-country security policy and producer-country development policy are pulling the same tonnage in opposite directions.

Frontier Traceability is the ethical instrument that doubles as a security one. Proving a refined tonne's origin serves both the child-labour audit and the export-control regime; the same opacity defeats both.

Established The leverage is out of all proportion to the money. The IEA's paradox numbers deserve stating together: tripling rare-earth prices adds about 0.1% to the cost of a car, rare earths are under 1% of a car's value — and the same materials, withheld, place on the order of US$6.5 trillion a year of downstream production outside China at risk. Chokepoint power scales with irreplaceability, not market size; the whole strategic-minor-metals trade is smaller than a mid-sized commodity market and moves states.

Frontier Midstream concentration is this century's refining question. Twentieth-century energy security was won not at oil wells but at refineries and sea lanes; the analogue holds. Whoever owns conversion sets prices, quality standards and, in crisis, allocation — and the 2023–25 export-control sequence was the first open exercise of that power at scale.

Speculative The long game is whether interdependence re-forms or fragments. A durable 3–5× two-price world for strategic materials would be historically novel: a standing tax on the diversifying bloc, a shrinking export market for the incumbent, and a permanent incentive on both sides to defect back to trade. Most historical dual-price regimes eroded; none involved inputs this substitutable-in-principle and this concentrated-in-practice.

Frontier Demand substitution remains civilisation's quiet exit. The LFP shift removed more cobalt exposure than every Western cobalt policy combined. Chemistry that walks away from a chokepoint is the only response the chokepoint owner cannot license.

Horizons track measured concentration and dated policy tests, not announcements.

• 10 yr: Frontier The rare-earth scoreboard resolves: top-supplier refining share at or near the IEA's projected 70% by 2035, or reverted after the first price war; the MP floor and the 2028 magnet-facility target tested against real Chinese pricing; November 2026 decides whether the suspended extraterritorial controls return. Copper's 25% projected 2035 gap either narrows through smelter builds outside China or becomes the binding constraint on electrification. At least one Western gallium or germanium byproduct retrofit either sells commercially or the announcements lapse.
• 25 yr: Speculative Average recycling rates approach the IEA's projected 20% by 2040 and end-of-life vehicle batteries finally dominate recycling feedstock around 2050 on current fleet arithmetic; secondary supply becomes a genuine geographic diversifier for copper and nickel where collection works, and not elsewhere. Substitution verdicts land: sodium-ion and iron-nitride either take real share or join the long tail.
• 50 yr: Speculative If electrification stock saturates, OECD copper and battery-metal demand is met substantially from urban mines, and midstream leverage migrates to whoever owns collection and separation of end-of-life material — a different geography, not necessarily a less concentrated one.
• 100 / 250+ yr: Handwave Claims about permanent mineral hegemony or post-scarcity material abundance at these ranges are storytelling; every input — chemistry, geology of demand, the state system itself — is unknowable at this distance.

• Depends on Nothing on this map blocks it. No pending scientific result gates midstream capacity; every binding constraint in this brief is a price, a plant, a permit or a purchase agreement. The demand-side facts it consumes — cell chemistry shares from advanced battery technologies, tape-industry rare-earth exposure from high-temperature superconductors — are inputs, not blockers.
• Requires (not on this map) Heavy-rare-earth separation capacity outside China beyond the first 2025 Malaysian tonnes, since dysprosium and terbium are where the April 2025 controls actually bit. A price-support mechanism proven through a Chinese price collapse rather than around one, because the US$110/kg floor has been paid but never stress-tested. Gallium recovery circuits actually commissioned at Western alumina refineries, converting a 5× price divergence into tonnes. Qualification throughput — the one-to-three-year buyer-acceptance clock — compressed enough that new plants sell within their financing window. And one scientific result nobody has produced: an iron-nitride or comparable rare-earth-free magnet at sintered-NdFeB energy product, which would delete the magnet chokepoint rather than diversify it.
• Enables Diversified cell manufacturing, non-Chinese magnet and motor supply, superconducting tape production at scale, and the credibility of every Western electrification and rearmament schedule that currently assumes refined material will be purchasable.
• Adjacent Resource economies for the upstream states, industrial ecology for the flow accounting, circular economies for the loop-closure claim this brief prices, and advanced battery technologies for the cell-level substitution evidence.

The claims in circulation, stated and answered.

Handwave “The bottleneck is in the ground.” The reserves conversation is mostly beside the point. Mining is the diversified segment; refining is where 70–90% single-country shares live, and the IEA's own framing — mining capacity expanding across diverse regions while refining and downstream capacity remains limited — is the one-sentence correction.

Handwave “China has all the rare earths.” China holds a large but minority share of global reserves; what it holds almost entirely is separation and magnet-making. The West's problem was never geology — Mountain Pass and Mt Weld are excellent deposits — it was thirty years of exiting the chemistry.

Frontier “Recycling will free us from this within the decade.” The arithmetic refuses: announced 2030 battery-recycling capacity is roughly seven times available feedstock, end-of-life batteries do not dominate feedstock until around 2050, and the flagship Western recycler was sold out of insolvency for about US$40 million after spending US$485 million on a hub that never ran. Recycling matters — a third of copper and nickel demand already — but on a 2040s clock, and today three-quarters of battery pre-treatment capacity is itself in China.

Frontier “The 2025 export controls prove China can shut the West down at will.” They proved calibrated leverage, which is different. The controls were licensed, not absolute; the extraterritorial layer was suspended within weeks of announcement; and every month of a 5× price divergence finances the competitors China's planners least want built. Weaponisation demonstrated the chokepoint and simultaneously started the clock on it.

Speculative “Substitution will save us” / “substitution is impossible.” Both circulate; both overreach. The systematic finding across dozens of metals is that no major metal has adequate substitutes for all its uses and some have none at all — yet LFP's rise past 55% of EV batteries deployed removed more nickel and cobalt exposure than any mine ever added, at chemistry-development speed, roughly a decade. Substitution is real, slow, partial and demand-side; it is a portfolio hedge, not a rescue.

Handwave “High prices prove scarcity; low prices prove the problem is over.” Lithium did both within four years while demand grew 25% annually throughout. Midstream security risk is a structure variable, not a price variable, and the 9% investment decline of 2025 happened at low prices precisely because markets price the spot tonne, not the chokepoint.