Rare Earth Exploration in Gilgit-Baltistan: What We're Actually Seeing in the Field

Last month I spent four days on a ridge above the Indus near Chilas with a portable XRF, two geologists from Islamabad, and a cook who kept complaining about the wind. We pulled 23 grab samples. Three of them came back with combined light REE oxides north of 0.4% from a single weathered syenite outcrop. That's not a deposit. But it's the kind of number that makes you stop packing up and stay an extra night.

I'm writing this because I get the same question on calls with buyers in Brussels, Tokyo and increasingly Houston: does Pakistan actually have rare earths, or is this another story someone is selling? Honest answer — we don't know the size yet. But the geology says we should be looking hard, and we are.

Why the Himalayan collision zone is worth your attention

The Kohistan-Ladakh arc and the Karakoram block collided with the Indian plate roughly 50 million years ago. That collision did three things that matter for REE exploration Pakistan-side.

First, it generated alkaline and peralkaline intrusions along reactivated suture zones — the kind of magmatism that concentrates light rare earths (lanthanum, cerium, neodymium, praseodymium) and sometimes the heavies too. Second, it pushed deep crustal material toward the surface, exposing what would otherwise be 10 km down. Third, it created the kind of intense weathering profiles in the lower valleys that can host ion-adsorption clay deposits — the same style that made southern China the world's dominant heavy REE producer.

Look, I used to think Pakistan's rare earth story was mostly carbonatite-adjacent and probably modest. Then we started actually walking the ground in 2022 and the picture got more interesting. The carbonatites are real — there's documented work going back to the Geological Survey of Pakistan's reconnaissance in the 1990s around the Peshawar plain and the Koga complex in Swat. But what's under-explored, and what we hold ground over, is the alkaline syenite belt running through parts of Gilgit-Baltistan and the monazite-bearing placers in the river systems.

Monazite. That's the one that keeps coming up.

The Indus, Gilgit and Hunza rivers all drain catchments containing Himalayan leucogranites and metamorphic complexes that are known monazite hosts. We're already running alluvial gold operations on stretches of these systems. The same heavy mineral concentrates that give us gold also give us monazite, zircon and sometimes xenotime (which carries the heavy rare earths — dysprosium, terbium, yttrium). Right now most of that is going to tailings because nobody's set up to separate it. That's a real near-term opportunity and it doesn't require a 10-year mine build.

What we've actually sampled, and what we haven't

I want to be careful here because the worst thing I can do is oversell. So let me split this into what we have data on versus what's still hypothesis.

What we have:

• Stream sediment sampling across four concession areas showing anomalous Ce, La and Nd, with several spots over 800 ppm total LREE
• Grab samples from weathered syenite showing the 0.4%+ combined LREO I mentioned above (small sample count — 23 — so don't quote this as a resource)
• Heavy mineral concentrate analysis from Indus placer operations showing monazite content that on a per-tonne basis is economically meaningful if recovered as a co-product
• Confirmed presence of allanite and bastnäsite-group minerals in two intrusive bodies

What we don't have yet:

• Drill data. None. We're at surface and trench stage on the hard-rock targets
• A defined resource under JORC or NI 43-101. We're some years and a lot of capital away from that
• Confirmed heavy REE enrichment at scale. We see xenotime signatures in panning but haven't quantified the ratio properly
• Ion-adsorption clay confirmation, though the weathering profiles in some lower valleys look promising

This is what early-stage Himalayan REE exploration looks like. Anyone telling you they have a defined rare earth resource in Pakistan today is either ahead of where the public data sits, or they're stretching.

Where this fits in the supply picture

China processed roughly 240,000 tonnes of REO in 2023 and still controls something close to 90% of separation capacity globally. Every Western buyer I talk to — and the calls have picked up noticeably since the December 2024 export controls tightened — is looking for mine-to-magnet supply chains that don't route through one country.

Pakistan isn't going to fix that problem on its own. But a credible monazite stream out of Gilgit-Baltistan placers, even at 2,000-5,000 tonnes a year of mixed concentrate, would be a meaningful addition to a diversified feedstock strategy for a separation facility in Europe or Japan. And the hard-rock targets, if they pan out at drill stage, could be a different conversation entirely.

The licensing situation in GB is workable. Mineral rights sit with the Gilgit-Baltistan government under the 2017 mining concession rules, and exploration licences for REE-bearing ground are being granted where applicants can show technical capacity. We hold 16 concessions and have been through the process enough times to know where it slows down (environmental clearance and security coordination) and where it moves (title and renewal).

Logistics — the part nobody wants to talk about until they're three months in. Concentrate moves by road from Skardu and Gilgit down the Karakoram Highway to Islamabad, then either Karachi port (about 1,500 km from concession to ship) or overland through the Khunjerab Pass into Xinjiang when that's the better routing. Neither is cheap. Both work.

If you're a buyer or a JV partner thinking about REE exploration Pakistan as part of your 2026-2028 sourcing strategy, the right time to start the conversation is before drill data exists, not after. That's when you can shape what gets drilled and where the off-take terms land.

We're not there yet on the rare earths. But we're closer than the public picture suggests, and the geology is on our side. What would you want to see in the next round of sampling?