Seismic Surveys of the Deep Continental Crust

Published June 14, 2026

People say tight reserve tables mean we are running out of critical minerals. They are wrong. A reserve table is not a census of the crust. It is the narrow slice of rock someone drilled, modeled, financed, and can mine under current rules and prices.

The mistake persists because reserve tables look authoritative. They have columns, categories, and the smell of accounting. People treat them like a planetary fuel gauge. They are actually a working inventory of projects that cleared enough technical and financial hurdles to count as mineable. When the list is short, the proof is short, not the crust. Treating the edge of the reserve book as the edge of the map means panicking because the map ends where the budget stopped.

The paying customer for a better map is not an abstraction. It is the junior explorer trying to turn acreage into something financeable. It is the major miner trying to replace what its mill burns through each year. It is the refiner that wants steady feedstock instead of a geopolitical headache. It is the state trying not to discover it outsourced the supply chain because nobody paid to drill deeper.

Scarcity panics routinely confuse what has been mapped with what actually exists.

Once you see that, the drama around running out looks less like geology and more like bookkeeping with bad PR.

The Deep Harvest, Chapter 1: The Resource Horizon Fallacy

In Germany, a scientific borehole called KTB was drilled 9.1 kilometers into the continental crust. Down in that hole, the drill met free fluids, and the water flowing in carried methane. Those fluids became known because a drill went down into the rock and the evidence came back up the hole.

Most of the deep crust has never had that kind of look. Less than 1 percent of the deep continental crust has been systematically surveyed. The deepest routine picture of the rock usually comes from oil and gas seismic surveys, which send sound into the ground and record the echoes, and those surveys often stop at about 3 to 5 kilometers. KTB went nearly twice as far down as the deepest of those surveys usually look.

That gap matters every time someone reads a reserve table as a fuel gauge for the planet. Mining uses three words for metal in the ground. The endowment is everything the crust holds, found or not. A resource is a deposit that has been found and estimated. A reserve is the part of a resource that has been drilled, sampled, and shown to be mineable at a profit with the methods that exist today. Underneath all three sits a question about chemistry: which mineral the metal is locked in. The same metal can sit in a mineral a processing plant can break open or in one it cannot, and a large amount of metal in the wrong mineral never becomes a reserve.

So when a headline divides a reserve figure by one year of mining and announces how many years are left, it is counting the holes companies have already drilled and the samples they have already tested. The rest of the deep continental crust, the more than 99 percent of it that has never been systematically surveyed, does not enter that count at all. The Deep Harvest starts from this point: what runs short first is good information about the crust, and the money it takes to collect it.

The tools for collecting it exist, and their costs are known. Seismic reflection surveys routinely image rock 6 to 8 kilometers down, and an optimized survey can see past 10 kilometers. A 3D seismic survey covering about 10,000 square kilometers costs roughly $50 million to $100 million. The research behind the book puts a full deep-earth survey program at $5 billion to $15 billion over ten years.

Machine learning has moved into the same work. Kobold Metals, which raised a $537 million Series C round at a valuation of $2.96 billion after the money came in, uses AI pattern recognition to choose where to look. In Zambia that approach is credited with the Mingomba copper deposit, which the research puts at about 300,000 tonnes of copper. Passive seismic work, which listens to vibrations already moving through the ground instead of setting off its own sound source, is part of the same set of tools. None of it replaces the drill. A ranked target tells a company where to spend its drilling money, and core still has to come up out of a hole before anything moves from resource to reserve.

The pressure to get this right is rising. Copper demand is projected to reach 40 million tonnes a year by 2050, up from 26 million tonnes in 2023. In one IEA outlook scenario, the mining projects already announced cover only 70 percent of the copper and 50 percent of the lithium needed through 2035. Global electricity demand is on track to grow by about half between 2024 and 2040, and the minerals needed to build that grid grow with it.

A 30 percent gap in announced copper projects is a gap in projects someone has already found and planned. It is a reason to pay for surveys and drilling. It is not a measurement of what the deep crust holds, because almost none of that crust has been looked at. If information is what runs short, the companies and countries that pay to image the deep crust and drill the best targets will decide where the next mines open.

The same research does not stop at the continents. On the deep ocean floor, in the region known as the CCZ, more than 21 billion tonnes of nodules, lumps of metal-bearing rock, lie on the seabed at average grades of 1.3 percent nickel, 0.2 percent cobalt, 1.1 percent copper, and 28.8 percent manganese. A tonnage like that is an endowment count. Turning any of it into a reserve means proving it can be brought up and processed, and that is a different job from finding it. The Deep Harvest goes down to that seabed in Chapter 2, The Ocean Floor.