South Africa could supply up to 2.2% of 2030 neodymium-praseodymium oxide demand.

Phalaborwa's feed sits in two phosphogypsum stacks left by historic phosphoric-acid production. That earlier chemistry already attacked the phosphate mineral, allowing Rainbow to avoid a new pit and most hard-rock comminution. 3 Historic processing also left acidic water beneath unlined stacks, and reprocessing can clean up the site only if the new lined storage and water circuit work for years.

Steenkampskraal has the inverse profile: its narrow monazite vein carries exceptionally high grades, while an old underground mine gives the restart physical infrastructure. 11 The feed also contains enough thorium to make worker dose and product transport part of the operating licence. Small tonnage reduces the footprint; it does not make radiation obligations small.

South Africa's 2025 critical-minerals strategy asks for more local ore upgrading alongside new exploration. 2 Phalaborwa fits that processing aim from its first commercial design. Steenkampskraal's first phase sells concentrate and only later adds cracking, so its policy value depends on whether Phase 2 gets funded. The distinction should sit in the headline rather than disappear in a project count.

South Africa is testing whether waste chemistry can beat a new mine, while its highest-grade ore starts with the least processed product.

What the numbers say

Rainbow reports 35 million tonnes of phosphogypsum at 0.44% total rare-earth oxide. NdPr makes up about 29% of that oxide basket, giving a low-grade feed more magnet value than the headline suggests. 3 Uranium sits near 2 parts per million, far below many monazite projects. The real environmental argument therefore turns on acidic water and residue storage rather than a high-radiation circuit.

The current design targets about 1,850 tonnes a year of separated NdPr oxide. A separate product containing medium and heavy rare earths carries roughly 80 tonnes of Dy/Tb at the company's planned rate. 3 Yttrium adds potential revenue that the interim economics did not count. Those products would place Phalaborwa farther into processing than Longonjo or Ngualla, provided the solvent-extraction circuit holds purity across variable stack feed.

Rainbow's December 2024 study put the initial construction cost near US$326 million. It estimated an after-tax value of about US$611 million in today's money, using the study's assumed prices and a 10% annual discount. 4 The detailed study remained unfinished in July 2026, with roughly three quarters of the processing route then in engineering. 5 Final separation design and waste restacking can still move costs materially, so the earlier value estimate remains provisional.

DFC classifies Phalaborwa as Category B and identifies water use plus energy demand as core issues. Its screening also flags potential contaminant releases and hazardous-material management. 7 The company still needed water-use permits and waste approvals at the cut-off. 6 DFC support through TechMet validates strategic interest, while full funding and construction approval remain separate tests.

The mine page's narrative reports roughly 665,000 tonnes at 14.5% total rare earth oxide, yet its table totals 605,000 tonnes at 14.4%. 10 The detailed figures yield about 87,100 tonnes of contained oxide, which fits the smaller tonnage. The gap is too large to dismiss as rounding. A current formal technical report should settle the publishable figure.

Ore chemistry is dominated by light rare earths, led by cerium and lanthanum. Neodymium accounts for about 18% of the reported rare-earth oxide basket, while dysprosium is close to 1%. 12 That smaller heavy rare earth share still matters economically because Dy/Tb prices can outweigh their mass. Extraction value arrives only after cracking and separation, which Phase 1 does not perform.

Steenkampskraal began building its new concentrate plant in April 2026. The design calls for 13,400 tonnes a year of material above 50% total rare-earth oxide, and the company said Industrial Development Corporation (IDC) funding covered Phase 1. 13 The company scheduled trial production for August, but no later public confirmation appeared by 27 August; without later confirmation, the project remains in construction rather than production. 14

A company radiation presentation reports about 2.14% thorium dioxide in measured and indicated material. 16 The mine holds a National Nuclear Regulator certificate for naturally occurring radioactive material, while later cracking needs change-of-scope approvals and a durable thorium route. Phase 1 also requires a buyer licensed to receive monazite concentrate. A signed destination matters as much as the first bag leaving the plant.

DFC invested US$50 million in TechMet to support a portfolio that includes Phalaborwa. 8 Associated Press reported that the US moved ahead despite a wider diplomatic clash with Pretoria, with strategic output aimed largely toward American supply needs. 9 Funding through a portfolio vehicle should not be described as a fully drawn Phalaborwa construction cheque.

South Africa and the EU signed a critical-minerals partnership in November 2025. The first investment roadshow followed in June 2026, carrying a broader partnership pledge of EUR 12 billion. 17 18 Since that envelope spans sectors beyond rare earths, project-level allocations and Transnet improvements will decide whether the policy reduces plant costs.

China remains a major commercial partner for South Africa, while neither lead project reviewed here was Chinese-controlled. The country can accept Western capital without choosing a single geopolitical camp. Its harder industrial choice is domestic: Phalaborwa reaches separated oxide in one design, while Steenkampskraal postpones that step. A large local magnet factory remained outside the most likely forecast for 2030.

Steenkampskraal should reach the first proof point because its concentrate plant is smaller and already funded. Phalaborwa offers greater value from later processing, but its unfinished detailed study and permits make the 2028 target vulnerable. Commercial supply begins only after customers approve the product. Concentrate and separated oxide remain different measures.