Gabon's $2 Billion Rare-Earth Restart: No Offtake, No Financing, No Separation Plant — and No Price Floor

The engineering is detailed: three million tonnes of ore a year, concentrated to 1.5 million tonnes before separation, spanning US$2 billion across six to seven plants and nine finished products. The financing is not. SikaFinance reported the figure as roughly 1,150 billion CFA francs and confirmed that construction, financing and infrastructure decisions are still ahead, with physical work following the contracts. Nothing in the record shows a bank, a sovereign fund or an offtaker signing.

The output figures themselves are the tell. Three official numbers for the same line — 14,300 tonnes a year on the sponsor's own website, 16,000 tonnes cited at the Nov 6 audience, and 18,000 tonnes in the 2023 project data — differ by about 29% across documents nobody has audited. Niobium, by contrast, is quoted at the same 14,500 tonnes everywhere, because niobium is the part with a bankable product in a market with a price.

The chain has a sponsor, a state and a contractor. New Energy Metals is an unlisted Dubai holding company; Dusk Gabon holds the permit; ENFI is a Chinese engineering group that has been preparing the project and presented a works schedule aimed at a 2027 start. Indonesia's Perminas and Danantara signed a memorandum of understanding in February covering both equity or debt participation and Indonesian downstream integration. None of the three has disclosed a committed cheque: the money remains a stated intention, not a signed facility.

The memorandum of understanding between Perminas, New Energy Metals and Danantara is the only signed commercial document in the chain, and it is a working-group charter. It creates a joint programme from mining through magnet manufacturing, opens fast-track talks on Indonesian equity or debt, and designates Indonesia as a downstream processing and manufacturing platform. It contains no volumes, no prices and no floor. A mine without a contract sells into whatever market exists on the day, which here leaves the ore without a buyer of any kind.

Washington's answer to China's rare-earth leverage was not geological and not commercial. The Department of Defense agreed to guarantee $110 per kilogram on MP Materials's NdPr for ten years from the fourth quarter of 2025, taking 30% of any sales price above that level. That arrangement transfers the whole price risk to the buyer of last resort and takes the arbitrage out of the market by requiring the producer to stop shipping mined output to China.

Read against that template, Gabon's project is missing the load-bearing parts. The benchmark exists because a government promised to pay it; the arbitrage bar exists because a government wanted the ore kept out of the subsidised market; the offtake exists because a buyer and a seller signed. Gabon has a permit, an engineer and a memorandum. The instrument that turns a mine into a price-setter — a written promise from someone richer than the project — is the one thing the plan does not contain.

What actually pays for every Western rare-earth project today is the gap between two prices, and that gap is made of policy. In October 2026 NdPr oxide traded at roughly $103-106/kg inside China and $125-140/kg outside it, creating a spread of about 20-33% that licensing creates and a decree can erase. China extended the suspension of its broader export-control wave until a review on Nov 10, 2026, so the price that decides every non-China project's economics is scheduled to be re-tested this month.

The scale of the dependence is not subtle. MP's own output had fed an estimated 7-9% of China's NdPr oxide production over the previous three years before the Pentagon deal stopped the shipments. That is the template in miniature: a Western mine priced off Chinese demand while selling into it. Gabon's version, with a Chinese engineer and an Indonesian downstream, repeats the pattern without the prohibition attached.

Strip the politics out of the ore and the asset is a niobium mine with a rare-earth credit. Grades are 1.2% niobium, 1.4% rare earths, 0.03% tantalum and 0.03% uranium. Eramet's hydrometallurgical route leaches the ore and precipitates niobium and tantalum as a concentrate carrying 12% niobium by weight, with roughly 90% of the rare earths reporting to a double-salt stage and a patented follow-on route for residual middle and heavy rare earths. Puts the economics in a metal nobody floors.

The listed chain gives the cleanest verdict available. The two US rare-earth names without floors of their own trade at 17-20 times trailing sales, while the Chinese refiner and the Chinese mining house that actually process and buy the material trade at one to two times. The market is pricing the policy premium at roughly ten times the refining multiple, and that gap is the real trade in this supply chain.

Two readings follow. First, the premium is a policy rent, and policy rents are repriced by decree — China's export-control review on Nov 10, 2026 is the near-term test. Second, the downstream economics sit with whoever converts ore into oxide and magnets: Shenghe Resources grew revenue 15% year on year in the third quarter of 2026 yet saw net profit fall 42% to RMB 239 million, and Zhongke Sanhuan earned about 1% net margin on RMB 6.64bn of FY2025 revenue. Thin-margin processors are where cheap ore and a wide export spread actually turn into earnings.

Nothing in the Mabounié file changes the price of anything in the next three months. One date does. China's review of the suspended export-control wave on Nov 10, 2026 decides whether the ex-China premium that funds every Western project widens, narrows or disappears. Puts the trade in the licence calendar rather than the mining calendar.

The base case is not a mine that fails. It is a niobium mine that works, a rare-earth line that trickles, and a chain that stays downstream of the party that owns the separation and magnet capacity. The milestones that would break that base case are specific and observable.

This is not a call that Gabon never works. It is a call that a working Gabon is not, by itself, a price floor for anyone outside the existing chain. Five disclosures would flip that judgement, and each is checkable.