Rare earth elements
The AI buildout
runs on rocks
Collide Capital's phrase, and it holds. Data centres, the energy transition and defence modernisation are pulling on the same handful of elements at the same time. The shortage is not in the ground. It is in the chemistry that turns rock into a magnet, and China has spent thirty years owning that step. This map follows the material through six segments and names who is working on each one.
Capital raised sums disclosed totals only, so read it as a floor, not a market size. 41 companies still have not been checked against the CRM, so the met count is a floor too.
The chain, in six segments
Follow the material. A deposit gets found and mined. Concentrate gets processed and split into individual high-purity oxides. Oxides become metals and alloys, alloys become magnets. Recycling rejoins in the middle rather than at the start, because scrap arrives already part way there. And a sixth group is trying to delete the problem by building motors that need no rare earths at all.
The second step is the one that matters. Benchmark put China at 71% of light rare earth processing, 99% of heavy processing and 94% of magnet production. Read those in order: the share climbs as you move down the chain, and it is effectively total at heavy separation, which is where dysprosium and terbium come from. A mine on its own fixes nothing.
| Segment | What happens | Companies |
|---|---|---|
| 1. Exploration & Mining | Deposit discovery and primary ore extraction | 5 |
| 2. Processing & Separation | Concentrate to individual high-purity REE oxides | 8 |
| 3. Metallization & Metal Production | REE oxides to finished metals and alloys | 5 |
| 4. Recycling & Secondary Recovery | EoL magnets, e-waste, tailings to recovered REEs | 13 |
| 5. Magnet Manufacturing | NdFeB/SmCo sintered and bonded magnets | 3 |
| 6. RE-Free Alternatives | Magnets and motors without rare earths | 7 |
Open the value chain map to see who sits in each segment.
Where the weight sits
Recycling has the most companies by a wide margin. That makes sense: the feedstock is already above ground and already concentrated, capex is a fraction of a mine, and the supply grows on its own as EVs and turbines age out. Less than 5% of rare earths get recycled globally today, so there is a lot of room.
Magnet manufacturing has three. It needs a plant, an anchor customer and a qualification cycle that runs for years. Venture money struggles with all three.
So the busiest segment and the emptiest one are not the same, and neither is the segment with the widest gap. Worth holding onto when you look at the map.
Sitting across all of it are the integrators, the companies building two or more consecutive links rather than picking one. They get the middle of the map, with a strip showing which links each one owns outright, which it has secured through an offtake or a partner plant, and where the gaps are. The gaps are the useful part. They are where a company still has to go and find someone.