How Satellite Intelligence Is Helping Governments Map Lithium and Rare Earths

A geologist in Lisbon told me last year that his team had spent eleven months on a survey block in the north of Portugal. Helicopter time, ground crews, the works. Then a small contractor pulled hyperspectral satellite data over the same block and flagged three lithium-bearing pegmatites in about a week. Two of them his team had missed.

That's the shift happening right now. Quietly. Across mining ministries from Lisbon to Islamabad to Santiago.

Governments are realizing they don't actually know what's under their own soil. And the cost of finding out the old way — sending crews, drilling holes, waiting for assays — is brutal. A single exploration program in a remote province can run $40-60 million before anyone confirms whether the deposit is real.

Satellite intelligence is rewriting that math.

Why lithium and rare earths broke the old playbook

The old exploration model worked fine when we were chasing copper or gold. Big deposits. Known signatures. Known geology. Companies had a hundred years of literature to lean on.

Lithium and rare earths are different animals.

Lithium hides in spodumene-bearing pegmatites, in salt brines, in clay sediments — three completely different geological settings that require three completely different exploration approaches. Rare earths (the seventeen elements everyone keeps lumping together) show up in carbonatites, ion-adsorption clays, and weird alkaline complexes that most geological surveys frankly weren't mapped to find.

And here's the thing — global demand for both is projected to climb 487% by 2040 if you believe the IEA's stated policies scenario. So governments that don't have an inventory of what they're sitting on are about to look very foolish.

This is where remote sensing critical minerals work has become unavoidable. Not optional. Unavoidable.

What satellites actually see

I'll be honest, I used to think satellite mineral mapping was mostly hype. Pretty maps. Limited real-world calibration. Then I spent time with the team at GeoMine AI and watched them walk through a spectral analysis workflow on a block in Balochistan, and I had to adjust my view.

Here's what's happening technically. Sensors like Sentinel-2, ASTER, WorldView-3, and the newer hyperspectral platforms (EnMAP, PRISMA, EMIT on the ISS) capture reflected light across dozens — sometimes hundreds — of narrow spectral bands. Different minerals reflect and absorb light at specific wavelengths. Lithium-bearing micas have a fingerprint. So does monazite, the main rare earth host mineral. So do the alteration halos around carbonatite intrusions.

Feed enough of that data into a trained model and you can produce a probability map. Not a certainty. A probability. But a probability over 80,000 square kilometers in two weeks instead of a decade.

Satellite lithium detection has gotten precise enough that the USGS, Geoscience Australia, and the EU's GSEU initiative are all running parallel programs. The Portuguese national geological survey published a lithium prospectivity map in 2023 that used Sentinel-2 spectral data as a primary input. Western Australia did the same for rare earths the year before.

What's wild is how cheap the input data is. Sentinel-2 is free. Landsat is free. ASTER archives are free. The cost sits in the modeling, the calibration with ground truth, and the geological interpretation. That's where companies actually compete.

The governments getting this right (and the ones aren't)

Look, not every country is moving at the same speed.

Australia, Canada, and Brazil have built well-resourced satellite-driven critical minerals programs with proper funding. The DRC and Zambia have started but lean heavily on Chinese partners for the modeling layer, which creates its own geopolitical wrinkles. Pakistan, Mongolia, and parts of Central Asia are sitting on what's likely enormous rare earth mineral mapping potential but have only recently begun systematic spectral surveys.

The smartest move I've seen is what Saudi Arabia did with its $182 million geological mapping program announced in 2022. They didn't try to build everything in-house. They contracted satellite intelligence specialists, ground-truthing crews, and AI modeling firms separately, then integrated results through their own ministry. That kept costs honest and avoided vendor lock-in.

The worst pattern I keep seeing? Countries that buy a single proprietary platform from one vendor, get a glossy dashboard, and have no idea how to verify the outputs. Three years later they've spent $20 million and produced maps no exploration company trusts.

What this means for the next decade

Satellite intelligence isn't replacing drilling. It never will. You still need to put steel in the ground to confirm a deposit. What it's replacing is the wasted years — the helicopter surveys over barren rock, the soil sampling campaigns in the wrong watersheds, the exploration capital burned on hunches.

A junior mining exec in Toronto put it to me bluntly: "We used to spend 70% of exploration budget figuring out where not to drill. Now we spend 30%."

That 40-point swing is the entire story.

For governments holding mineral rights, the implication is sharper still. If you can map your own prospectivity before licensing rounds, you negotiate from a position of knowledge rather than hope. Chile is already doing this with lithium concessions in the Atacama. Indonesia is moving toward it with nickel and cobalt. Pakistan's recent moves around Reko Diq suggest the same thinking is creeping in — slowly.

The countries that figure out spectral mapping in the next three years are going to be the ones writing the rules on critical minerals supply for the next thirty. The ones that don't will be selling concessions blind, again, like they did with copper and oil.

And honestly? I'm not sure most ministries have understood yet how short that window is.