Actionable insights — Energy Fuels' Rare Earth Magnet Plan

3:11 1. Find the permit, not the deposit — licence as the real barrier to entry

1. For any critical-minerals story, ask what is actually scarce. Ore bodies rarely are: "they're not rare… what's very rare is the ability to mine them and process them economically" (20:54).
2. Look one step downstream from the mine for a facility whose barrier is regulatory, not capital: a licence, a tailings facility, a radioactive-materials permit, a grandfathered environmental approval.
3. Quantify the barrier in two units management can be held to: replacement cost and years to permit. A big number in dollars is only capital; a big number in years is a moat.
4. Verify that the licence actually covers the new use. A uranium licence that can lawfully accept and process thorium/radium-bearing rare-earth feed is a different asset from one that cannot.
5. Then invert the test: count how many competitors hold the same permit. If the answer is "one," you have found the choke point; if it is "everybody," the moat is a story.

• Licence amendments and renewals (the moat's expiry date), state/NRC actions on the tailings cells, and any competitor filing for the same permit class — that filing, not their drill results, is the first real threat.

21:35 2. The byproduct-economics test — who has to carry the full cost?

1. For every producer in a commodity, ask: is this the primary product or a byproduct? A byproduct producer allocates only incremental cost to the metal; a single-commodity mine "has to apply all of their costs" to it.
2. Rank the cost curve by that structure, not by headline all-in sustaining cost — the byproduct names sit at the bottom of the curve by construction and survive the price troughs that kill pure plays.
3. Check the diversification claim is real: does the company actually sell the co-products into separate markets with separate cycles, or is it one revenue line dressed up as several?
4. Apply the same test to the supply side of a bull thesis: if a metal is mostly a byproduct, supply responds to the primary metal's price, not its own — so a price spike does not necessarily bring supply.
5. Finally, invert it as a short/avoid screen: single-commodity juniors in a byproduct-supplied metal are structurally disadvantaged before management quality is even considered.

• Monazite/tailings off-take announcements by sand miners (each one is supply diverted from or to China); and, in any commodity, the ratio of byproduct to primary supply before believing a supply-response argument.

9:25 3. Split the basket — never accept an aggregate commodity claim

1. When a company says it produces "rare earths" (or "battery metals," or "PGMs"), refuse the aggregate. Ask which specific elements, in what split, and which of them carry the margin.
2. Learn the one distinction that matters in that basket. In rare earths it is light vs heavy by atomic weight: NdPr is the key light; terbium and dysprosium are the key heavies — and "the heavy rare earth is really where the game is at… there's no other source for these heavy rare earth oxides besides China."
3. Rate each project by the distribution of the valuable elements in its ore, not by tonnage. Tonnage of the wrong element is inventory, not revenue.
4. Convert the claim into a milestone: heavy-oxide production is a physical, dated, verifiable event — separated Tb/Dy oxide shipped, not "capability."
5. Re-check the demand side per element too: only about four of the 17 elements go into commercial magnets (24:04) — the rest need a defense buyer to matter.

• First separated terbium/dysprosium oxide production and its assay/volume; any quarter where "rare earth revenue" is reported without a light/heavy split is a step backwards, not progress.

19:16 4. Buy the bottleneck vs. build it — count the world's facilities

1. Map the supply chain end to end and count the operating facilities at each step outside the dominant jurisdiction. Not projects, not announcements — plants running today.
2. The step with the smallest count is the bottleneck, and it is usually a manufacturing/know-how step rather than a mine.
3. Decide build vs. buy on the nature of the barrier: capital and permits can be bought with money and time; process know-how and customer qualification cannot. Where the barrier is craft, acquisition is the only fast route — "trying to develop that internally or organically would be very very difficult."
4. Price the acquisition against replacement cost of the physical plant and the years of qualification embedded in it, then ask what currency is being used to pay (cash, stock, debt).
5. Stress the count: if "only four exist" but three are captive to another market, the effective count for a Western buyer is one — which is a scarcity claim worth verifying independently, because it also sets the price the seller can demand.

• Deal closings as the actual milestone (Korea ~end-Aug 2026; VAC early 2027, "government approvals" pending) — a chain that only exists on closing is not yet a chain; and new entrants announcing the same step, which resets the count.

13:40 5. Mine the volunteered weakness — the most valuable sentences in an IR interview

1. Read a promotional interview for the concessions, not the claims. A promoter who volunteers a specific negative is usually conceding something already unarguable — and it is the most reliable data in the transcript.
2. Log each concession with its number and its date. Vague admissions are worthless; dated, quantified ones are a model input.
3. Build the downside case out of the concessions alone, ignoring the projections entirely, and see whether the business still works.
4. Then check whether the concession undercuts the headline identity of the company. If the "uranium company" admits it cannot compete on uranium cost, the equity is really being sold on something else — price it that way.

• Reserve/mine-life updates that would extend Pinyon Plain past 2030; any new low-cost US pound; and whether later interviews quietly stop repeating the $60-80/lb admission.

17:01 6. Read the share structure before the story — insider buying vs. convertible overhang

1. Ask for shares outstanding and fully diluted first. An executive who cannot answer it on camera is a flag on its own when the growth plan is acquisition-led — go straight to the 10-K/10-Q rather than accepting the anecdote.
2. Locate the convertible: maturity, conversion price, and whether a capped call is attached. The capped call tells you management already paid to limit dilution — useful, but it also marks the price at which the paper lands on the register.
3. Treat the conversion price as a soft ceiling reference, not a target: it is where dilution becomes real, so model share count at that price, not today's.
4. Weigh the insider signal by size relative to the buyer's own stake and by mechanism — an open-market purchase by the CEO is a real signal; option exercises and grants are not.
5. Cross-check the ownership mix. High institutional ownership (here "60-70%") means the story is already sponsored; a retail-heavy register means the promotion is doing the work.
6. Finally, price the acquisitions: two pending deals of this size will be paid for somehow. Whatever is not cash is either new stock or new debt.

• Form 4 / SEDI filings around deal announcements; any equity raise, ATM usage or debenture amendment; and the diluted share count in the next 10-Q against the count on the day the deals were announced.

26:17 7. Cross-check every "not priced in" claim against the peers it implies

1. When management says the market gives it too little credit, extract the comparison being made and name the unnamed peers yourself. A relative-value claim with anonymous comparables is not testable.
2. Build the peer table the executive did not: market cap, production, processing capacity, and — critically — position in the chain. "As much or more capacity as some of our peers" is a claim about a specific metric; go find it.
3. Separate the two halves of the argument. Replacement cost ($600M plant, $0.5B mill) is an asset-value floor; "billions of dollars per year of cash flow" is a projection. Only the first is checkable today.
4. Date the projection and discount it honestly: cash flow "by 2030, 2031" that requires four countries, three industries and two unclosed acquisitions is an option, not an earnings stream. Ask what has to be true each year between now and then.
5. Ask the mirror question: if the assets are worth so much more than the market cap, what does the market believe that management is not addressing? Usually dilution, execution timelines or jurisdiction risk — all three are present here.

• Whether the discount closes on milestones (deal closings, heavy-oxide production, Donald FID) or only on sector sentiment — the former validates the argument, the latter means you were paid for beta, not analysis.

10:05 8. Rank the asset ladder by jurisdiction before quality

1. List every project by country and score political risk before geology — a 100-year mine life in a fragile jurisdiction is not comparable to a 10-year mine in Utah.
2. Note which projects the strategy actually depends on versus which are optionality. Feedstock the processing plant needs is load-bearing; a distant giant is not.
3. Watch how management itself hedges. A volunteered "a little bit more political risk" is the floor of the real risk, not the ceiling.
4. Sequence the catalysts by jurisdiction: the nearest-term, lowest-risk milestone (an Australian FID) is the one the market can underwrite; the frontier asset should be valued near zero until permitted.

• The Donald FID actually being taken (and its capex/funding), Madagascar permitting and fiscal-terms news, and whether Toliara ever moves from "wonderful project" language into a dated development schedule.