Lynas Rare Earths Ltd made 17% less rare earth oxide in FY26 than it did in FY21. And got 171% more per kilogram for it.
The gap is not all market. Most of it is that Lynas is no longer selling the same thing.
Trace the basket. NdPr was 34.6% of total oxide production in FY21 and 55.5% in FY26. Over the same six years, non-NdPr output fell from 10,300 tonnes to 5,829 tonnes, down 43%. NdPr output rose 33%.
A$1.97bn of capital spend between FY22 and FY26 bought less oxide and more NdPr. That was the trade. It worked.
But it breaks the number everyone is quoting.
Average selling price per kilogram of REO is a blended figure across whatever mix went out the door that year. Compare FY26 to the last cycle peak in FY22 and it reads A$80.7 against A$60.3, a record, 34% clear of the previous high.
Measure revenue against tonnes of NdPr produced instead. FY26 gives A$134.7k per tonne. FY22 gave A$156.5k. On that basis FY26 sits 14% below the last peak, not 34% above it.
The FY26 basket is roughly one and a half times as NdPr-rich as the FY22 basket, and a blended price will always flatter the richer basket. One caveat that cuts the other way: FY26 also carries separated dysprosium, terbium and samarium, small volumes at high value, so the NdPr-denominated figure understates FY26 somewhat.
The point holds regardless. Every ex-China comp built on dollars per kilogram of REO carries a mix assumption nobody wrote down. Contained TREO in a developer deck is the same error one step further upstream.
Assembling this manually would take a day: finding and sifting through six annual reports and two figures from each. Lynas has never printed the ratio. Pulse Intelligence Partner does it in minutes.
Most of what separates good analysis from consensus is sitting in filings nobody has time to line up.
Less searching. More strategising.™
Where does your team's data infrastructure sit today?
Answer 10 questions. Get a private diagnostic on your AI readiness, in minutes.
Less Searching. More Strategising.™
See the platform running on real mining data, and what it would look like for your team.