Canada makes scandium oxide, but still mines no rare earths.

SRC remains the clearest test of Canada's processing-first thesis at an integrated facility. Its Saskatoon complex was built to move material through separation and into metal, with dedicated recovery for dysprosium or terbium alongside an NdPr stream. Integrated start-up testing of the full minerals-to-metals circuit continued through 2026. 10 REalloys signed five-year agreements covering the large majority of planned output. The funded upgrade is designed for 525 tonnes a year of neodymium-praseodymium metal, 30 tonnes of dysprosium oxide and 15 tonnes of terbium oxide. Those ratings are capacity, not a forecast of first-year output. The plant still has to prove steady production and customer acceptance. 11

Kingston provides a second route, with Ucore's commercial demonstration plant reporting more than 7,400 simulated-commercial operating hours by August 2026. It had produced 99.5% NdPr using raw material from Vietnamese ionic clay and North American bastnaesite, a useful stress test for a separation system that may have to accept changing sources. 12 Ottawa conditionally offered up to C$36.3 million for a proposed commercial Sm/Gd facility in Kingston. 13 Conditional support and sample tests remain several practical steps short of steady commercial output, with Ucore remaining in demonstration.

Stage labels hide just how sharply China's control changes along the rare earth chain. Natural Resources Canada put its share of mine production at 69% in 2024. Refining was far more concentrated, with China holding 90% of output. North America's 0.8% share at that stage makes separation Canada's first useful target rather than a side project. 2 The IEA places China's share of sintered permanent magnets at almost 95% for the same year. 3 A Canadian separation line can reduce one bottleneck without fixing the magnet layer. Finished oxides and metals will still need qualified allied customers, likely across the border or in Europe.

Imported raw material bridges the gap between an empty domestic mine column and a usable processing plant during this early phase. Sovereign branding becomes risky when it hides a vulnerable raw-material supply or imported specialist reagents. Even after both inputs are secured, final conversion may remain offshore. Canada should report origin and product form with every capacity claim. An origin-and-product-form label makes a 525-tonne-a-year metal plant more credible. A vague promise of thousands of tonnes from a mine still at an early study stage remains appropriately conditional.

Canada can become strategically useful before it becomes a large miner, provided the product passes customer tests and the raw-material route is disclosed.

What the numbers say

Wicheeda is Canada's clearest reserve-backed candidate, though final feasibility and permitting remain unfinished. The preliminary study carries initial capital around US$1.44 billion for the proposed operation. Planned dry concentrate output is 62,500 tonnes a year, making it more advanced than a project with only a resource estimate. 4 Ottawa offered up to C$1.88 million in 2026 for infrastructure studies rather than mine construction. One study covers a 60 km power connection from the existing grid. The road design is a separate 43 km assignment, a reminder that access is part of the mine plan. 5 Useful preparation does not yet amount to construction finance for the mine.

Tardiff in the Northwest Territories is larger on the resource page but one study stage earlier. Vital's 2025 scoping concept describes a starter operation producing 56,000 tonnes of concentrate a year. Initial capital was estimated at US$291 million, well below Wicheeda's study case but still untested by a preliminary study. 6 The earlier North T pilot stopped, and its processed stockpile does not make Tardiff an operating mine. Any table that carries the old producer label forward turns a discontinued trial into national output.

Strange Lake carries Canada's most visible infrastructure burden among the advanced northern proposals. The federal registry describes a mill designed to process 17,000 tonnes of ore each day. It would also need roughly 170 km of new road to the Labrador coast, making the transport corridor part of the project itself. 7 Torngat secured C$165 million across two loan facilities for preconstruction and infrastructure work. 8 Those facilities can fund serious preparation without amounting to a grant or full project finance. The impact statement and final build capital still separate the remote design from construction.

Ashram's June 2026 early economic study gives another strong case at 17,466 tonnes a year of rare-earth content in mixed carbonate, on an oxide-equivalent basis. NdPr accounts for about 4,035 tonnes of that annual product. The excluded infrastructure cost deserves equal attention. Initial capital is estimated at C$1.231 billion before the access road. 9 A long northern road is part of the mine cost, not an accessory added after the plant works. It also widens the assessment footprint for local or Indigenous communities.

Canada is putting more policy weight behind infrastructure and market-making rather than relying on exploration grants. Budget 2025 proposed a C$2 billion Critical Minerals Accelerator for strategic projects. The proposed five-year program starts in the 2026-27 fiscal year with loans or equity. Guarantees and purchase agreements remain available through separate instruments under the same proposal, a useful fit when lenders need confidence in both price and customer demand. 14 The harder public decision is which project receives enough support to reach a final build decision, since backing every study would scatter capital without creating supply.

Among the current candidates, Wicheeda has the most legible reserve and preliminary study route. Tardiff could offer a lower-capital starter concept if its preliminary study confirms the scoping work. Strange Lake carries a more integrated ambition with heavier infrastructure, while Ashram still needs to convert an early study into a tighter engineering case. Resource tonnes alone cannot rank projects at such different stages. Reserve quality and permitting set the order until committed finance turns a study into construction. A project only joins the race once construction actually begins.

Our most likely result by the end of 2030 is one Canadian mine beginning to test output, not four mines opening together. Processing plants should be further ahead. SRC could be operating its line, while Kingston plants test materials or handle recycling. Canadian raw material may enter that network late in the decade, but imported material is likely to remain important. Canada could still become a useful allied processing base even with mine output well below Australia or the United States.

A bear case remains plausible if rare earth prices weaken and northern capital costs rise. Canada would then finish the decade as a processing and project-development jurisdiction with no sustained primary mine. The optimistic case needs two mines financed and under construction early enough to begin trial production by 2030. Neither is that advanced today. Signed construction contracts and final approvals would change the picture. The final proof is steady output accepted by paying customers.

One mine in start-up testing by 2030 would validate the Canadian strategy. Four simultaneous starts would require a timetable the current files do not show.

EarthRarest forecast