MP Materials: DoD Floor vs Spot NdPr Bankability, Data Caveats

One kilogram of neodymium-praseodymium oxide now carries two official prices: roughly $60 on the Shanghai spot market and $110 under the U.S. Department of Defense's price floor. That $50 spread — an 83% premium over the traded price — is no longer a curiosity of distorted markets. It is the single variable that decides whether an American rare-earth mine can be financed at all.

For lenders, spot price is dead as a bankability input. In a two-price world, revenue models anchored to $60 spot flatter projects whose cash costs sit far above what the open market will pay, while the $110 floor rewards only those positioned to capture government-backed offtake. What sorts bankable from unbankable, then, is not a spot forecast but a project's cash-cost position — a function of its geology and its flowsheet.

The implications are stark: most nominally cheap rare-earth deposits fail the screen once processing costs are counted, while the model critics call subsidy-dependent — the vertically integrated producer selling into a guaranteed $110 — is the only structure lenders reliably fund. Bankability, in the end, is a commercial verdict rendered on engineering evidence, not on a price chart.

Two Prices, One Kilogram

MP Materials and the Department of Defense announced the contract most Western rare earth models had been missing: a $110/kg price floor on NdPr oxide from Mountain Pass, with MP retaining a share of any upside above it, paired with a DoD purchase of convertible preferred stock that made DoD the company's largest shareholder. The anatomy matters more than the headline. This is not a subsidy check — appropriations lapse and get rescinded. It is a purchased put embedded in an equity relationship: the counterparty guaranteeing the downside holds enough upside, through conversion economics, to want the mine running flat-out for the full tenor.

The competing number forms differently. Global separation capacity sits predominantly in China, and the prints traders chart as "spot" — SMM's and Asian Metal's RMB-denominated domestic PrNd oxide assessments, running near $60/kg — are surveys of clearing prices inside that system. Export-licensing policy transmits directly into the Western quote. Spot is a Chinese domestic clearing price, not a free-market signal; a lender anchoring revenue to it has imported another government's industrial policy into the model and labeled it weather.

Why the floor binds economically, not just politically: follow the contribution margin. Moving a kilogram between the columns swings $50 of revenue, but the margin effect is what compounds through a mine model. The floor converts MP from a price-taker at Chinese-set clearing levels into a price-maker for US defense-linked volumes — and contracted-versus-forecast margin is exactly the cash-flow character lenders underwrite.

One limit keeps this honest: according to Adamas Intelligence demand estimates, MP's output is a small fraction of the global NdPr market, so nearly all tonnes outside the floor still clear at Shanghai prices. The two prices coexist with assigned jobs — the floor sets the bankable price for US-linked projects; spot remains the liquidation and stress price. A base case built on spot quietly prices a liquidation outcome as the plan of record.

The 2026 working rule falls out of the table: for a lender's base case, the left column wins outright, and spot earns a role only as the stress scenario. Treat the two prices as different instruments, not competing forecasts — underwrite US-linked tonnes off the floor, hold spot for the stress year, and read a sponsor's disclosed cash cost as the geological verdict on whether a private-sector version of the MP package is signable. Cash cost is where the orebody enters the financing stack, and it decides who gets to sign.

Treat every price input like an assay: it needs a certificate and a lab name. The certificates exist. According to SMM and Asian Metal monthly assessments, Chinese PrNd oxide spot has swung from historic peaks to deep troughs across the past cycle, and Fastmarkets and Argus Western assessments now track the market near the $60/kg level. All prices below are per kilogram of NdPr oxide.

MP Materials is the pre-floor control group. Per its filings, Mountain Pass produced NdPr oxide at scale, with quarterly releases guiding toward further volume growth. Separate the levers: the volume step is meaningful, but the earnings inflection arrived when the contracted floor replaced spot on the price line. Unit math ranks them: at this output scale, each $1/kg of realized price moves annualized revenue by millions, so a price reset dwarfs a volume gain of this size. Calling MP's improvement "a better market" mistakes floor arithmetic for demand.

The Receipts

The record identifies the actual price-setter. According to Reuters, SMM, and Fastmarkets archives, successive administrative moves — export controls covering multiple rare earth elements, their escalation, and the subsequent Busan truce suspension — each shifted assessed Western premiums within days. In none of those windows did a mine start, stop, or change throughput. When paperwork reprices metal faster than geology can supply it, a spot forecast is a model of someone else's bureaucracy.

Run the receipt check before trusting any deck: pull the SMM and Asian Metal monthly series beside the Fastmarkets and Argus Western assessments, stamp the policy dates on the chart, and stress the sponsor's coverage against the archived trough, not the analyst's base case. A price deck without an archive behind it is just a grade interpolation with better marketing.

Bankability is not a judgment call; it is an inequality. The Peterson Institute for International Economics titled its 2026 analysis "'Bankability' of critical minerals is in the eye of three beholders" because equity sponsors, developers, and lenders each score the same asset differently. The screen below removes the disagreement: one beholder (the senior lender), one test, four archetypes. Bankable means NPV-positive with debt service coverage clearing lender thresholds at the DoD floor, with one stress year at Shanghai spot absorbed from cash reserves. Unbankable means any model requiring more than the floor to clear its hurdles. Every cell populates from one formula: incremental EBITDA = annual NdPr tonnes × ($110 − $60).

Row 3 dies honestly, and geology is the killer. Dy/Tb-rich projects derive a large share of basket revenue from elements no US floor covers, so their verdict column inherits raw Chinese-policy volatility no matter how exceptional the deposit grade. A resource statement reports tonnes and grade; it is silent on DSCR. On this screen a world-class heavy-REE orebody is not a better credit than a mediocre one — basket composition, not grade, sets the verdict, and the floor does not rescue them.

The winner is structural. The floor-backed integrated model stacks three cash-flow layers no merchant producer can copy — floor-priced oxide, magnet-conversion margin, and guaranteed defense offtake. According to the July 2026 project-finance guidance from heavendesigns.in, an unsigned offtake defeats bankability regardless of how strong the yield model looks, and lenders size repayment to the conservative P90 cash-flow case — which is exactly what the mandatory stress year simulates. Archetype 1 holds a signed PPA-equivalent with the Pentagon; archetypes 2 through 4 hold forecasts. Hence the verdict column: pass, conditional, fail, fail — and row 2's conditionality collapses to fail the moment the floor contract disappears.

Kill the last myth while the table is open: no spot-recovery forecast belongs in a lender model. Consensus price forecasts have never closed project finance for a Western rare earth mine; every financing that closed was built on a contracted price, which is why the stress year is mandatory rather than optional. Run the arithmetic before falling for the drill core — annual NdPr tonnes × the floor-to-spot spread — and if the stress-year column goes negative with no contracted floor attached, walk away. Any grade, any jurisdiction, any quarter of 2026.

The Bankability Screen

An admission the bull case tends to skip: the cleanest rule in this guide rests on the thinnest dataset in mining finance. The floor was calibrated against the cost sheet of a single operating asset — Mountain Pass — and the population of Western rare earth projects that have ever cleared commercial project finance is small enough to count on one hand. Treat the screen as a sound heuristic fitted to sparse data, not an empirical law. Heuristics fitted to sparse data fail at their edges, and the edges are exactly where a diligence budget should go.

The evidence carries three structural limits. First, survivorship bias: closed financings leave paper trails; abandoned term sheets leave almost none, so any observed track record for a floor-based screen is conditioned on the deals that survived it. Second, the spot side of the comparison is an assessment, not a tape — SMM and Asian Metal publish surveyed ranges on thin physical liquidity, and their methodologies differ enough that the "spot" input carries a methodological spread nobody publishes. Third, and closest to home for anyone who builds resource models: the cost side is a point estimate stacked on a resource estimate with genuine estimation variance. Dilution assumptions, recovery curves, and model-to-mill reconciliation routinely move projected cash costs by margins large enough to flip a marginal screen, yet the test treats geology as deterministic. It is not; it is a distribution wearing a single number to the meeting.

Variance across cases matters just as much. The screen prices one oxide, but the deposit archetypes competing for that price differ enormously in cost structure:

When does the rule itself break? Three edge cases, none of which overturn it. First, sovereign capital: the coverage-ratio test presumes a commercial lender repaid from cash flow. Japan's state-brokered rescue of Lynas — covered earlier — closed precisely because a state balance sheet does not play by DSCR rules. The screen governs private capital; it cannot predict what a strategic government will fund. Second, basket-dominant deposits: where heavies, not neodymium and praseodymium, carry the margin, an NdPr-only stress test checks the wrong lock — apply the same survive-at-the-bad-price logic to every revenue metal. Third, tenor: where amortization outlives the contracted floor, back-end years silently reprice to spot, and the single-stress-year design understates duration risk, as the tenor analysis above details.

One debunked shortcut deserves burial here: the belief that a snap-back in assessed spot to historical mid-cycle levels would, by itself, restore bankability. The record runs the other way — consensus forecasts have never closed a Western rare earth mine, while every financing that did close was built on contracted prices. The Molycorp postmortem earlier in this guide is the cautionary exhibit.

What the Data Doesn't Tell You

The practical takeaway: before accepting any screen output, demand four artifacts — the written basis for allocating joint costs in byproduct systems, the model-to-mill reconciliation history, the identity of the price reporter feeding the model and whether its assessments are transaction-backed, and confirmation that debt tenor sits inside the floor's life. The rule earns trust not because it predicts the good price but because it forces survival at the bad one. Its blind spots are the cases where the bad price is not the binding constraint — and those, more than the headline gap above, are where projects quietly die.

A floor cannot collateralize a loan that outlives it. The DoD contract profiled earlier runs for a fixed term, while mine lives and senior debt tenors in this sector routinely extend beyond it — every cash flow beyond expiry is priced off an assumption nobody signed. Run the lender's arithmetic: if Shanghai spot reverts to the $60/kg stress year mandated above once the floor lapses, terminal-value haircuts can erase the entire NPV accumulated during the floor era. The floor purchases bankability only for the amortization window it actually covers. And because non-recourse project debt is repaid solely from project cash flows — the SPV structure described in BESS.courses' 2026 project-finance primer — the uncovered tail lands entirely on the lender's balance sheet.

The controlling precedent for what happens without state support has a name: Molycorp entered Chapter 11 after NdPr oxide peaked near historic highs and slid back toward spot levels. Management spent the interim underwriting exactly the recovery the bull case still sells — that spot mean-reverts high enough to service debt. No forecast saved it. That is the myth to bury outright: spot recovering into a comfortable band does not make a Western mine bankable, because consensus spot forecasts have never once closed project finance for one. Molycorp is the corpse attached to that assumption.

The counter-precedent cuts the other way. Japan's JOGMEC and SOJITZ injected fresh equity into Lynas, keeping the largest non-Chinese producer solvent through the identical price crash. State price support therefore carries a record of one death and one survival — a distribution no lender can lean on blindly. It is a coin flip with survivorship bias, not a policy guarantee.

Then there is the counterparty Beijing holds. Both levers have been pulled within living memory: a capacity expansion flooded the market and erased most of NdPr's value within a few years, while the export-control-and-truce cycle demonstrated the choke-side lever. The data cannot tell you which lever comes next. Flood validates the floor — it looks cheap. Choke moots it — realized prices run far above contracted levels and the premium buys nothing. Either way, the floor's worth is regime-dependent in a way no static sensitivity table captures.

The working discipline follows directly: date-stamp the refinancing wall against floor expiry, zero out Dy/Tb contracted credit, and re-run the DSCR screen each quarter as floor-priced sales accumulate. If the post-expiry years cannot carry debt at the stress case, the deposit's cash-cost position — geology, again — decides whether the project ever reaches final investment decision.

Tenor Traps and the Molycorp Ghost

Now run the guide's screen, line by line:

Identical ore body, identical flowsheet, identical management team — only the price input changes the verdict. The spread between the columns is $50/kg on every floored tonne; accumulated across the full contract window, spread earnings alone service the whole debt package inside the window. Capex ratios never reveal this: as FG Capital Advisors describes project-debt sizing, facilities are advanced against cash available for debt service and tested on base-case and downside coverage, with lock-up and default thresholds in between. The spot column produces no testable cash; the floor column passes every gate.

So the honest verdict: bankable with a floor, unbankable without one. For a geologist, the humbling part is that the grade-tonnage curve we refine for a living cancels out of the equation entirely. The decision variable is neither grade nor tonnage; it is whether a developer can convert geology into a contracted price at or above the floor before FID. Deposits that cannot sign such contracts should be valued as options, not projects, and the diligence ask writes itself: before opening anyone's resource model, demand the counterparty, tenor, and floor level of the NdPr contract behind it.

A passing DSCR is a geological outcome before it is a financial one. The five gates below translate the floor-price thesis into tests a credit committee can actually run, and they are ordered: fail Rule 1 and nobody reads Rules 2 through 5. What kills projects fastest is rarely resource size — it is a model whose survival depends on a price nobody signed.

Rule 1 — Underwrite at $110, stress at $60. Build the base case on the DoD floor or a signed equivalent, then require the model to absorb one full year at $60/kg funded from cash reserves — not a revolver, because committed liquidity evaporates exactly when the stress year arrives. The rejection criterion is absolute: if the project cannot clear lender DSCR thresholds at $110/kg, a sponsor asking for a richer floor gets a pass, not a counteroffer. Geologically, this is where the deposit first bites — metallurgical recovery, reagent demand, and strip ratio determine whether the model clears, and none of them are negotiable in a term sheet.

Rule 2 — Contract or it didn't happen. Count floor revenue only from executed agreements naming price, volume, and tenor; policy announcements, MOUs, and "administration signals" carry zero weight — not discounted weight, zero. According to FG Capital Advisors, institutional debt requires the contractual structure, cash flows, and risk allocation to be proven before lenders are approached; estimated CAPEX plus a target leverage ratio is not sufficient. Two traps recur: an offtake indexed to spot-minus-discount names no price, so it contributes nothing at the floor, and partially contracted production counts only for the contracted share — the uncovered tonnes are stressed at $60.

Worked Case

Rule 3 — Buy the cost quartile, not the grade. Headline TREO grade and NdPr fraction are marketing variables; position on the C1 cash-cost curve is the bankable one. Proceed only where independent cost-curve work — Wood Mackenzie and Adamas Intelligence both publish — places C1 at or below roughly $60/kg NdPr, because bottom-quartile assets stay EBITDA-positive even if the floor regime dies. The edge case geologists miss: a spectacular grade can still sit in the wrong quartile if the mineralogy is refractory — fine grain size and locked gangue depress recovery and inflate reagent consumption, moving C1 without touching the grade on the deck's cover slide.

Rule 4 — Match tenor or haircut. Where the floor expires before debt maturity, re-run the DSCR with terminal-year prices cut back to $60/kg and accept the deal only if the amortization schedule fits inside the floor window or the haircut still clears lender coverage hurdles. The overlooked fix is sculpting: push the heavy repayment years inside the contracted window and let the tail ride o

Frequently Asked Questions

How large is the gap between the DoD price floor and the Shanghai spot quote for NdPr oxide?

At $110/kg under the Department of Defense floor versus roughly $60/kg on the Shanghai spot market, the government-backed mechanism sits roughly an 83% premium above the traded price.

Does the $110/kg floor apply to all the NdPr sold worldwide?

No — according to Adamas Intelligence demand estimates, MP's output is only a small fraction of the global NdPr market, so nearly all tonnes outside the floor still clear at Shanghai prices.

Isn't the DoD deal effectively just a subsidy check?

It is a purchased put embedded in an equity relationship — paired with a DoD purchase of convertible preferred stock that made DoD MP's largest shareholder — so the counterparty guaranteeing the downside holds enough upside through conversion economics to want the mine running flat-out for the full tenor.

What exactly counts as 'bankable' versus 'unbankable' under the lender's screen?

Bankable means NPV-positive with debt service coverage clearing lender thresholds at the $110 DoD floor while absorbing one stress year at Shanghai spot from cash reserves, whereas any model requiring more than the floor to clear its hurdles is unbankable.

Why can't lenders just anchor their base case to the ~$60/kg spot print?

Because the prints traders chart as spot — SMM's and Asian Metal's RMB-denominated domestic PrNd oxide assessments running near $60/kg — are surveys of clearing prices inside China's system into which export-licensing policy transmits directly, making it a Chinese domestic clearing price rather than a free-market signal.

How quickly did Chinese policy moves actually move assessed Western premiums?

According to Reuters, SMM, and Fastmarkets archives, successive administrative moves — export controls covering multiple rare earth elements, their escalation, and the subsequent Busan truce suspension — each shifted assessed Western premiums within days, without any mine starting, stopping, or changing throughput.

Quick answers

Also worth reading: China 2026 Rare Earth Quotas: NdPr Oxide +12% in January: China 2026 Rare Earth Quotas: · Unlock hidden energy with vital trace minerals from the earth: Unlock hidden energy with vital · Ancient woolly mammoth yields the oldest RNA ever found on earth: Ancient woolly mammoth yields the