Mining the minerals to power AI

JOSH GOLDMAN: In an economy powered by AI and batteries by midcentury, we will need to mine more copper over the next 25 years than has been mined in all of human history.

RANA EL KALIOUBY: Why can’t other exploration companies take a similar approach to what KoBold is doing?

GOLDMAN: The amount of capital the industry is deploying has changed dramatically. In base metals, about $80 million used to produce a winning project. Today, it’s well over a billion dollars of failures.

EL KALIOUBY: That’s Josh Goldman, co-founder and president of KoBold Metals. They’re using AI to find the critical minerals that will shape our future, and they’re backed by some of the biggest names in tech like Jeff Bezos and Bill Gates.

These minerals are essential to electric vehicles, robotics, semiconductors and the infrastructure behind the AI economy. They’re also at the center of geopolitical competition as countries race to secure the resources that will define the next era of power and progress.

In this episode, Josh and I talk about how AI is transforming one of the oldest industries in the world, why critical minerals have become so strategically important, and whether mining can be done in a more ethical, more sustainable way that is better for people and for the planet. On Earth Day, that last question feels especially important, so let’s dig in.

I’m Rana El Kaliouby, and this is Pioneers of AI, a podcast taking you behind the scenes of the AI revolution.

[THEME MUSIC]

Thanks for joining us on Pioneers of AI. A few months ago, a common friend of ours, Chris Schroeder, texted me and said, “You gotta meet this guy.” And he was right. So shoutout to Chris. Full disclosure: I’m an investor in KoBold. I love what you guys are doing and am very proud to be an investor and support you on this journey.

GOLDMAN: Very glad to have you as one.

Copy LinkHow a physicist ended up reinventing mineral exploration

EL KALIOUBY: I want to start with your origin story. You have a PhD in physics, and you spent some time in private equity in oil and gas, then decided to start an AI and data company in the mining and exploration space. How did that come about?

GOLDMAN: I’ll start with my background. I studied physics because I thought it was the most interesting thing I could work on. It had the hardest problems, and I got to think about the very big, like the origin of the universe, and the very small, like atoms, the things you can’t hang your hat on as a physicist just by doing an undergraduate degree.

You don’t even reach the frontier of science. You mostly learn what happened up through about 100 years ago in an undergraduate degree. So I went on to graduate school. I actually went to the UK for a couple of years. First, I did a master’s program in applied math and theoretical physics.

EL KALIOUBY: Where was that?

GOLDMAN: Cambridge University and then Imperial College.

EL KALIOUBY: Did we overlap at Cambridge?

GOLDMAN: 2002.

EL KALIOUBY: Yes, I was there from 2000 to 2005 for my PhD. Go figure.

GOLDMAN: We were inhabiting the same place.

EL KALIOUBY: Yeah.

GOLDMAN: I had a Marshall Scholarship, so I had two years of funding there. Then I came back and did a PhD in computing, just because the science is really interesting. That’s still the case. Nobody does a PhD in the sciences for some kind of practical reason. You do it because you’re interested.

EL KALIOUBY: Right.

GOLDMAN: Throughout my career, one of the things that’s been most important to me is learning, and learning lots of things. Learning science is one of the most satisfying things. I knew I wanted to focus on science problems that were really relevant, and that led me to energy and sustainable energy.

I met my co-founder, Kurt House, as a graduate student. He had started a reading group on energy at Harvard. We did things like organize trips to power plants and coal mines.

EL KALIOUBY: Fun.

GOLDMAN: Exactly. This is what nerds do in graduate school and in adult life.

I think a deep grounding in the sciences makes you a really good thinker and a really good problem solver. So much of what we’re doing at KoBold is applying first-principles thinking to hard problems.

Copy LinkWhy copper and lithium are becoming the backbone of the AI economy

EL KALIOUBY: So you went on to have a career in oil and gas, and then at some point you were like, enough of that.

GOLDMAN: I wanted to get to know the incumbent energy industry. I went to McKinsey in Houston and worked with power companies, oil and gas exploration and production companies, and equipment manufacturers who make power equipment, everything from wind turbines to subsea oil field equipment and things in between.

I worked on strategy questions and then started working on private equity in oil and gas. After doing that for a short while, we decided we were going to stop working on this. We wanted to work on the energy transition. There were much better business opportunities there, and it was more compelling personally.

We thought from first principles about what we should be working on and asked, what are the raw materials the future economy needs?

EL KALIOUBY: OK. Those are?

GOLDMAN: Materials like copper and lithium. Why do we need lithium? You need lithium for batteries. If you want to make a device you can pick up and move around, whether that’s a car, a truck, a drone, a robot or an airplane, and you want a vehicle that is durable and has long range, lithium wins. It is the lightest and most electropositive metal.

We have the periodic table we have. We’re not going to invent a new element. So if we’re going to have an economy powered by battery-powered devices by midcentury, we need lithium production rates to be 10 times what they are today.

There’s a very different material profile. If you recycle an internal combustion engine car, you can’t just turn it into an EV because it’s not made out of the same materials. You need something like five kilograms of lithium in an EV. There’s no lithium at all in an internal combustion engine vehicle.

So we need new materials. Another big trend is materials like copper. If you want to move electrons around, copper is the metal of choice. Again, we have the periodic table. We’re not likely to invent a metal that is cheaper and more effective than copper.

We need this for broader electrification, but also for AI. All of these data centers need power, and we need the power transmitted to them and the data centers themselves built. All of this drives demand for copper. In an economy powered by AI and batteries by midcentury, we will need to mine more copper over the next 25 years than has been mined in all of human history.

EL KALIOUBY: Wow. That’s great.

GOLDMAN: These are incredible demand signals for things we need.

EL KALIOUBY: Right?

GOLDMAN: And so the question is, where are they going to come from? We started to work on that problem and started pulling on the threads.

Copy LinkUnderstanding ore deposits and why most minerals are still waiting to be found

EL KALIOUBY: For people who are not geologists or in the mining industry, like me, I want to paint a picture for our audience of where these resources are found. Let’s dig into that a little bit. I actually have a visual prop I want to bring up for my next question. You’ve compared mineral deposits to a bakery item.

GOLDMAN: Yes.

EL KALIOUBY: Right?

GOLDMAN: I love it.

EL KALIOUBY: All right, so tell us: What are we looking at here? We’re looking at raisin bread.

GOLDMAN: What Rana is referring to here is one of the myths about scarcity. Is there enough copper? Is there enough lithium? The vast majority of the ore deposits that contain these metals are still out there to be found.

EL KALIOUBY: How do we know this?

GOLDMAN: Because we know how these ore deposits get made, or we know enough about them to know. We know the pressures and temperatures they form at. You can determine those things by looking at the minerals themselves.

EL KALIOUBY: OK.

GOLDMAN: You need some microscopes, but these are pressures and temperatures that form deep under the Earth, thousands of meters below the surface.

EL KALIOUBY: But the?

GOLDMAN: Yet the majority of the ore deposits we’ve found are right at the surface.

EL KALIOUBY: Really close here. Literally.

GOLDMAN: Sticking out of the ground. The way to think about this is that an ore deposit is like a raisin, and the eroded surface of the Earth is like slicing off the end of a loaf of raisin bread.

EL KALIOUBY: Yep. OK.

GOLDMAN: The ore deposits are at different depths, but the surface of the Earth is an irregular cut.

EL KALIOUBY: It’s like this, exactly. If you sliced off the top, you…

GOLDMAN: You only see the raisins on the surface where you just cut.

EL KALIOUBY: OK.

GOLDMAN: But if you know the recipe, you know there are many, many more.

EL KALIOUBY: Raisins all over here. Right. Love it.

GOLDMAN: The rest of them are out there to be discovered.

Another misconception: copper and lithium are not rare earths. Rare earths are a suite of elements, mostly the lanthanides on the periodic table. People commonly think, well, rare must mean valuable. But rare earths is just a chemical name for this set of important elements.

Ore deposits themselves are all very rare, but the Earth’s crust is very, very big. So there are many out there, even though they occupy a very small fraction of the Earth’s crust.

Copy LinkWhy finding minerals is harder than mining them

EL KALIOUBY: Talk us through, historically, how you find and mine these ore deposits.

GOLDMAN: Before you mine it, you have to find it.

EL KALIOUBY: Yep.

GOLDMAN: Finding is the hardest part.

It’s not the case that we just know where the minerals are and it’s just a matter of digging them up. Mostly, we just don’t know where they are.

EL KALIOUBY: Yep.

GOLDMAN: Historically, the single most important method for finding ore deposits has been humans walking around looking for strange-looking rocks. Actually prospecting, walking across the surface of the Earth.

You can think of copper minerals this way: wind and water expose the rocks, and water and oxygen turn them blue and green. Think of the patina on the Statue of Liberty. In places where there’s a lot of copper, like the African copper belt, you can see the green rocks.

But then you start to deplete this inventory of deposits sticking out of the ground. Certainly for copper deposits in places already known to host copper, if you can find it sticking out of the ground, somebody probably already found it.

EL KALIOUBY: Already. Yeah.

GOLDMAN: There could be deposits that people have never looked for before that are still sticking out. Lithium, for example, people haven’t been looking for until very recently because we didn’t need big chunks of it.

But for copper deposits, what’s left to find is harder. It’s concealed somehow. It may have several layers of rock above it, potentially hundreds of meters of rock. In the case of our Mingomba project, more than 1,000 meters.

The key thing is that they’re not easy to detect with one data type. You can’t just look at a satellite image and see a bull’s-eye pattern of how the rocks have been altered by the flow of fluids. You can’t just take a bunch of soil samples over a grid and make a plot of the copper in the soil and drill the hot spots. You can’t just fly an airborne survey or gravity survey and say, here’s a red spot, I’m going to drill that.

The data is very high-dimensional, and you need a combination of signals across different data types in order to predict where an ore deposit is. That’s a hard science problem.

EL KALIOUBY: That’s one of the things I loved when I first learned about KoBold, and it’s super aligned with our investment thesis at Blue Tulip Ventures. It’s this trifecta of sensors, multimodal data, and both generative and predictive AI.

GOLDMAN: I love the way you framed that. The technology we use isn’t one black box product. There’s really no one product at all.

EL KALIOUBY: Yeah.

GOLDMAN: What we’re trying to do is make better predictions. To do that performantly, we’re collecting many terabytes a day of data, so you need to be able to interact with it.

EL KALIOUBY: Index it.

GOLDMAN: Yes.

EL KALIOUBY: Query it.

GOLDMAN: Right. Then you’ve got to be able to put it into models and combine it with other data. We use many models trained on that data.

We’ll take data like that and use a combination of historic archive data, which we can talk more about, and models trained on that data to make predictions on a given project: Where should we go next on this project? But we also use it to make predictions on other continents. Data we’ve used in Canada has helped us find the right rocks for hosting lithium deposits in Australia.

The features of ore deposits are in common. It’s the same geological process. It’s the same recipe to cook the raisin bread. When we see those processes at work in one place, the model learns how to recognize them and can identify them in new data in totally unrelated domains.

This is the whole point of the company. Anything we do doesn’t just help us execute one project. We get better models and new technologies that allow us to execute the next one. Each thing we do makes the next thing better.

EL KALIOUBY: Let’s take a short break, and we’ll be back to dive more into the sources of KoBold’s data.

Copy LinkTurning fragmented geological archives into a modern AI advantage

EL KALIOUBY: Let’s talk about this historic data and the archives. Where do you get that data from? What does it look like? How do you digitize it?

GOLDMAN: The vast majority of data ever collected about the Earth is in the public domain, but it’s incredibly fragmentary. There are tens or hundreds of thousands of different sources. In many jurisdictions, the data of prior explorers is all public.

EL KALIOUBY: Wow. Cool.

GOLDMAN: In Quebec, in Finland and in many other places, you have to write a technical report on what you did every year. These are 100-page documents describing what you were looking for, what people had done before you, what activities you undertook and what the results were. You also have to deposit copies of the raw data with the regulator, and then that goes into the public domain.

There are hundreds of thousands of these. Sometimes an individual data set might be a folder with a dozen or so files, and it might represent one survey. These things are differently formatted, not all in the same language, and they contain all kinds of different mistakes. A lot of the data is in the text of these reports, in figures, hand-drawn maps and tables.

Since we started the company, we have been aggregating this, extracting structured data, providing tools to our teams and putting it all in a system so they can interact with it. That’s very powerful, and the latest developments in frontier models help us build on that.

EL KALIOUBY: All because it allows you to bring all this unstructured data together.

GOLDMAN: Yeah, that’s right. It allows us to work with much more of this data, much faster.

Some of the data is still on paper.

EL KALIOUBY: Wow.

GOLDMAN: In Zambia, for example, where our first major discovery is, there are historic archives that include maps that are almost 100 years old.

EL KALIOUBY: How cool is that?

GOLDMAN: It’s very cool.

EL KALIOUBY: And they’re just in some governmental office in Zambia?

GOLDMAN: The Geological Survey of Zambia has been the caretaker, so they are accessible.

EL KALIOUBY: You can go in and pull the map and look at it.

GOLDMAN: Yeah, that’s right. The staff who work there care enormously about the data. They’re the conservators of this legacy, and it is data. It’s extraordinary.

EL KALIOUBY: And it’s still relevant.

GOLDMAN: Absolutely. The rocks haven’t moved.

EL KALIOUBY: Right.

GOLDMAN: They’re still there. It’s ground truth for models trained on modern airborne geophysics and satellite imagery. But you need the ground truth. The ground truth for these maps was collected laboriously over thousands of hours of work by skilled geologists, and you could never go collect this again. It would be too hard.

So we are digitizing it. In the case of the Zambian Geological Survey, we do it publicly. You can go to gsd.gov.zm. This is part of our partnership with the government of Zambia.

We’re not doing it just for ourselves. We benefit, obviously, because we want to use the data, just as we want to use every data source in the world. But really the benefit of the data is for the people of Zambia, to stimulate investment.

EL KALIOUBY: Why can’t other exploration companies take a similar approach to what KoBold is doing? What’s so hard about it?

GOLDMAN: The amount of capital you have to deploy is enormous. On average, in base metals, about $80 million used to produce a winner, a project that worked. Today, it’s well over a billion dollars of failures.

EL KALIOUBY: Wow.

GOLDMAN: That’s for two reasons. One is that the problem is getting harder. The easy-to-find things have been found, the raisins on top.

EL KALIOUBY: The ones at the top, right.

GOLDMAN: But the second reason is that the pace of learning has been too slow.

Copy LinkTreating exploration as a science of knowledge and inference

EL KALIOUBY: You talk about mineral exploration as an information problem, and in order to tackle that, you have a chief philosopher on the team. What does a chief philosopher do, and why that role?

GOLDMAN: We do have a chief philosopher. It’s funny, I had a conversation with a journalist who reached out because he saw that a mineral company had a chief philosopher, and he was a little disappointed to learn that our chief philosopher is an epistemologist and not an ethicist.

KoBold’s ethics are also incredibly important to our business. It’s right there on the front page of the company, and we start every all-hands meeting with an ethics vignette.

EL KALIOUBY: We’re going to come back to that. But define epistemology for anybody who hasn’t heard that term before.

GOLDMAN: It’s all about thinking about knowledge. How do we know what we know? How can we reason about things we can’t see? How can we make inferences about things we can’t observe? We’ve written an epistemology of exploration.

As we think about it, the core technologies are epistemic in nature. They help us know better.

The idea at KoBold is, number one, we need hypotheses. What we’re doing when we go out in the world is deploying some capital to get some information. That information is useful if we’re testing an idea. It has to be a well-defined idea, and we need to make some predictions about what we’re going to observe.

Then we need to reconcile the results of our work with our prior understanding and use that to guide the next step. If you’re an explorer at KoBold and you’re writing a technical memo to justify an expenditure — “Josh and Kurt, I want $2.5 million” — you’ve got to get that on the record before you get an allocation.

First, you need a good hypothesis and a clear characterization of it. Second, you’ve got to have alternatives.

So you can’t just have one view of what the world is, and then the alternative is, “Well, not that.” There are many possibilities.

We enlisted the help of our chief philosopher, who has written about the nature of science: What is science? Why is it powerful? How does inquiry work? He helps us think through those kinds of problems.

Copy LinkBuilding an ethics framework for mining that protects people and the planet

EL KALIOUBY: Very cool. Let’s talk about your ethics framework. What does that look like? What are some of the ethical principles underlying the company? Give us some examples. And you open your team meetings with an ethics vignette? That’s super cool.

GOLDMAN: Great question. This is important because we want to build a great company. Part of being a great company is leading in science. Part of it is doing something good for the world, and obviously providing the raw materials we need for the future economy is really valuable.

But mining has local impact.

EL KALIOUBY: Yeah.

GOLDMAN: It’s not sufficient to say there are global benefits, therefore we accept all of these local impacts. Actually, we need to work to minimize those and to do good locally too.

KoBold’s ethics include protecting the environment, protecting the safety of our people, anti-discrimination and anti-corruption.

EL KALIOUBY: Right. In some of the geographies where you operate, that’s real.

GOLDMAN: And we’re committed to working at the highest international standards everywhere we operate. We’ve built some extraordinary operations in Zambia. Our CEO there is Mfikeyi Makayi. She’s an amazing engineer.

EL KALIOUBY: She was on the TIME AI 100 list last year, which is super awesome.

GOLDMAN: Yeah, and she has helped us build an incredible team in Zambia.

Our Zambian colleagues are fully integrated with our global company.

EL KALIOUBY: Yep.

GOLDMAN: That’s creating opportunities in the community. It’s not sufficient to hire people who live far away. We’ve got to create opportunity in close proximity as well.

There are lots of other things we can do. Managing the waste from a mine is very important, and there are ways to design a better facility to protect against harm.

You can spend a little bit more on the design of your tailings facility to make it more robust, to have something that’s going to stand up over time and have a much lower risk of failure. Those are good decisions.

One of the questions we get from prospective employees is: Great, I understand you and Kurt are really committed, but are your shareholders willing to forego a basis point of returns in order to protect the environment?

And we have a shareholder base that allows us to say yes. We’re in this together for the long term, and we’re making long-term investments in exploration and in ensuring that the benefits are shared.

EL KALIOUBY: Coming up, more on the international perspective and the implications of mineral discovery around the globe. Stay with us.

[AD BREAK]

Copy LinkWhy critical minerals have become a geopolitical race

EL KALIOUBY: The work KoBold is doing has major strategic implications, especially in the geopolitical climate we’re in today. China has built a dominant position in the critical mineral supply chain. The U.S. is arguably significantly behind. How serious is that gap, and what is KoBold’s role in all of this?

GOLDMAN: You’re right. China has been focusing on critical minerals for a long time. This isn’t an overnight thing.

Where the Chinese position is really strong is downstream. Lots of minerals, wherever they are mined, will go to China for processing: from concentrate into metal, from those metals into battery precursors, and from that into batteries and cells. That’s a huge strategic advantage, and it has been built with many decades of investment and industrial policy.

The U.S. is playing catch-up. I think, commendably, the U.S. is playing catch-up quite aggressively. There are lots of different arms of the U.S. government providing financing, both in the U.S. and in support of American-led projects abroad. We’ve gotten a lot of government support all over the world.

This is very important. One place where KoBold, and where American companies especially, can lead is in discovering new deposits.

It’s not very interesting to just go acquire a handful of assets that exist today, which are depletion businesses. That’s valuable. But where are the next-generation assets? Where are these huge growth wedges? Chinese companies are not leading in discovery.

They’re great at building low-cost processing, particularly in China. They’re great at operating around the world, and they’ve acquired a number of existing assets. But that’s much less interesting than asking where the new tier-one, low-cost assets of the future are going to be.

Even though it is causing short-term pain, for the long game, the real question is: Where are the new deposits, and what are the assets that are going to be the future low-cost tier-one assets that sit on the commodity cost curve? We’re looking for those, and we at KoBold, with the support of the U.S., are looking for them.

EL KALIOUBY: There are a number of parallel explorations happening at any one point in time, correct?

GOLDMAN: We have a portfolio of 80 or so projects around the world, and that’s constantly growing and churning. We are falsifying hypotheses and dropping projects from the portfolio. We’re getting encouraging results and then accelerating our work.

We’re exploring in the U.S., in Canada and in Australia. We have teams in all of these places undertaking active exploration and deploying technology, and then we’re working on new projects as well.

EL KALIOUBY: What is the timeline for some of these exploration projects? Are we talking years, months?

GOLDMAN: Some ideas can be tested quickly. We can generate an idea, stake a claim or form a partnership, stand up a team, deploy operations wherever we’re going, maybe fly airborne surveys, collect rock or soil samples, and drill a few holes. Sometimes we can rule out an idea in one campaign.

But if we get encouragement, we might want to keep going.

Other times we’re starting with the concept that there’s a whole district there, and we need to collect a foundational layer of information to generate more specific hypotheses and figure out where in that district the specific geological target is.

There we might be flying hyperspectral surveys, doing other airborne work, aggregating all of the historic data, doing wide-area soil or glacial till sampling, and building a detailed geological picture that will help us refine our hypotheses. A program like that might take years, and if it’s encouraging, potentially longer.

Copy LinkWhy recycling alone cannot meet surging mineral demand

EL KALIOUBY: Some critics don’t think the U.S. should be investing in exploration and mining new sites, and instead should be investing in recycling materials or recovering them at mining waste sites.

In fact, I want to reference a study from the Colorado School of Mines that found the U.S. could meet most of its critical mineral needs by using the byproducts at domestic mines. How do you respond to that perspective?

GOLDMAN: You can’t recycle something until you have it. The problem is that if you want to make things with lithium-ion batteries, and you don’t already have enough lithium in circulation, there isn’t enough to recycle.

If we’re talking about electrification of transportation, for example, and you want to make an EV, you cannot recycle an internal combustion vehicle into an electric vehicle. There’s no lithium in that vehicle.

If you have steady-state demand, and not growing demand, then once you get enough material in circulation, recycling can supply most of it. But you’ve got to build up the circulation first.

For copper, again, the story is that demand is rising dramatically. There just isn’t enough scrap out there to supply the needs for new data centers, new power lines, new power generation and so on. So we’re going to need more of those materials.

Where there are economic incentives for recycling, much of that is already happening. It is generally cheaper to recycle copper scrap than to mine new copper, and so those things are done preferentially. It just won’t actually satisfy demand, because the metals intensity of the economy is changing. That’s one of the reasons.

EL KALIOUBY: You talked about human intelligence married with artificial intelligence. What is the secret to getting that marriage right?

GOLDMAN: A lot of experimentation. We have many, many ideas. Many of our ideas pan out, and many of them don’t. We invest more heavily in the ones that do.

Those experiments are both about which technologies are going to be really valuable and which exploration projects are the right places to deploy them. They’re also experiments in how we work together to get these systems really effective and to deliver results.

EL KALIOUBY: Josh, thank you so much for joining us on the show. This was a fascinating conversation.

GOLDMAN: Thank you so much, Rana. Really a pleasure.

EL KALIOUBY: On Pioneers, you’ve heard me talk a lot about the trifecta of sensors, data and AI, and how that can unlock new insights, especially in antiquated industries. This is exactly what KoBold is doing in the mineral exploration space, transitioning the industry from a heuristic craft into an AI-driven science.

This is more critical than ever. The global economy is going through a transition driven by electrification, AI and massive compute infrastructure.

The reality is that mining these minerals is disruptive to communities and to the planet. But there are methods to extract these resources more ethically. I invest in human-centric AI, which is why I believe in what KoBold is doing.

Thanks so much for listening. We’ll be back with a new episode next week.