Lithium Exploration: How Satellite Imagery Is Finding the Next Battery Boom

Lithium hit $80,000 per ton in late 2022. Then crashed to under $14,000 by mid-2024. And now it's climbing again because every battery factory from Reno to Shenzhen needs more of it than the world is currently producing.

That's the backdrop for what I want to talk about. Because while the price chart looks like a heart monitor, the actual hunt for lithium is happening quietly — and increasingly, it's happening from orbit before anyone sets foot on the ground.

I run GeoMine AI. We process Sentinel-2, ASTER, SAR, and SRTM data to flag mineral targets across Pakistan, and I personally own 15 mines in Gilgit Baltistan. Lithium has become one of the most asked-about commodities in our pipeline over the last 18 months. Mostly from investors who read one McKinsey report and now want to know if the Himalayas are hiding the next Salar de Atacama.

Short answer: maybe. Longer answer below.

Why lithium is a satellite problem before it's a drilling problem

Lithium shows up in two main geological settings. Brines (think salt flats in Chile, Argentina, Bolivia — the famous lithium triangle) and hard-rock pegmatites (Australia, Zimbabwe, parts of Afghanistan and almost certainly parts of Pakistan).

These two settings look completely different from space. And that matters.

Brine deposits sit in closed evaporite basins. From orbit, you're hunting for specific signatures — high-albedo salt crusts, particular hydrated mineral assemblages, the faint blueish tinge of lithium-bearing clays like hectorite. Sentinel-2's SWIR bands (11 and 12 specifically) pick up some of this. ASTER does it better in certain cases because of its 14 bands stretching deep into the thermal infrared, where carbonates and sulfates have distinct absorption features.

Hard-rock lithium is a different game entirely. You're looking for LCT pegmatites — lithium-cesium-tantalum — which intrude into older metamorphic rocks. They tend to be small. Sometimes just 50 meters wide. And the lithium-bearing minerals (spodumene, lepidolite, petalite) have spectral signatures that overlap with a bunch of other muscovite-group micas.

Honestly, this is where I got it wrong at first. I used to think Sentinel-2 alone could handle pegmatite mapping. It can't, not reliably. The 20-meter SWIR resolution is too coarse for narrow dykes, and spodumene's diagnostic absorption features sit in wavelengths Sentinel-2 doesn't fully resolve. We had to combine Sentinel-2 with ASTER and structural mapping from SRTM DEM to get usable targets.

What lithium remote sensing actually looks like in practice

Let me walk through how a typical lithium screening runs at geomines.

We start with regional structural mapping. SRTM 30m DEM, processed for lineaments and intrusion contacts. Pegmatites follow structural weaknesses — fault intersections, contacts between granite and country rock. If you can map the structure, you've narrowed your search area by 80% before touching a spectral band.

Then multispectral. Sentinel-2 for vegetation anomalies (lithium-bearing soils sometimes suppress certain plant species, sometimes enhance others — depends on the host rock). ASTER for muscovite and clay alteration halos. The halos around LCT pegmatites can extend 100-300 meters beyond the dyke itself, which is great because it makes the target bigger than the actual ore body.

Then SAR. Sentinel-1 backscatter helps separate fresh outcrop from weathered cover, and InSAR coherence maps tell you where bedrock is exposed versus where you're looking at alluvium hiding everything.

The stack matters more than any single layer. I've seen exploration teams burn six months on Sentinel-2 alone, find nothing definitive, and conclude satellite work doesn't work for lithium. It does. They just used 20% of the available data.

For the brine side — which is less relevant for Pakistan but huge globally — there's published work from the USGS and a few groups in Argentina showing they can predict lithium concentrations in brine pools to within roughly 15-20% accuracy using Sentinel-2 time series alone. That's wild. You're estimating subsurface chemistry from the color of a salt flat.

Where Pakistan actually fits

Look, I'll be direct. Pakistan isn't going to be the next Australia for lithium. The geology suggests we have pegmatite belts — Kohistan, parts of Chitral, the Hindu Kush margins, possibly some zones in Gilgit Baltistan near existing aquamarine and tourmaline workings (tourmaline is itself an LCT pegmatite indicator mineral, which most people don't realize).

But we don't have systematic exploration data. The Geological Survey of Pakistan has done excellent regional work, but lithium-specific surveys are basically nonexistent. Which is exactly why satellite-first exploration makes sense here. You can't afford to fly aeromagnetic surveys over 50,000 square kilometers on the off chance. You can run Sentinel-2 and ASTER over the same area in a week for the cost of a decent laptop.

We flagged a target zone near Shigar last year — pegmatite dykes intruding into the Karakoram batholith, with muscovite alteration patterns that lined up with what you'd expect for an LCT system. Field samples came back with anomalous tantalum and beryllium. Lithium values were lower than we hoped (peak 280 ppm Li2O, you'd want 1%+ for a mineable spodumene deposit), but the indicator chemistry was correct. The system is there. We just hit the wrong part of it.

That's the honest reality of lithium satellite exploration. You're narrowing a haystack, not finding the needle. The needle still requires a drill rig and a lot of patience.

What I'd tell anyone starting now

If you're a mining company or an investor looking at lithium plays in Pakistan or the broader region, three things.

Don't trust any report that uses only one satellite source. Multispectral plus radar plus DEM, minimum. Anyone offering you lithium targets from Sentinel-2 alone is selling you incomplete work.

Ground truth matters more for lithium than for almost any other commodity. The spectral signatures are subtle. False positives are common. Budget for at least 15-20% of your project cost to go into field sampling and XRF verification.

And be patient with the price. The lithium market will swing again. The deposits, however they're found, take 7-10 years to develop. So whatever's getting flagged from orbit today is feeding a 2032 supply curve.

Which raises a question I keep asking myself — if Pakistan does have a real LCT belt up north, and the data is sitting there in free Sentinel-2 archives going back to 2015, why isn't anyone systematically working it?