Episode Summary
Executive Summary: Josh Goldman explains Kobold Metals as an “intelligent mining” company that uses AI, sensors, and massive geoscience datasets to find high-value mineral deposits—especially copper and lithium—more efficiently than traditional explorers. The discussion covers exploration as an information problem, how Kobold combines old and new data to guide drilling, the economics of high-grade deposits, regulatory/social constraints, and why philosophy and uncertainty management are central to the company’s method.
Main Topics: Kobold’s business model: AI-driven mineral exploration (Priority: 5/5): Kobold focuses on exploration rather than mining, using AI and human judgment to identify the best places to search for copper, lithium, nickel, and cobalt. Goldman argues exploration offers outsized returns and is where technology can be most differentiating. Geoscience data aggregation and multi-scale analysis (Priority: 5/5): The company builds a unified system from satellite imagery, airborne geophysics, rock/soil samples, legacy maps, reports, and other public/private data across multiple spatial scales to improve predictions about ore deposits. Ore deposits as an information problem (Priority: 5/5): Goldman frames exploration as a search for scarce information rather than scarce minerals. The challenge is locating where geological processes concentrated metals into economically recoverable deposits. Zambia and the Mingomba copper deposit (Priority: 5/5): Kobold’s flagship success is in Zambia, where it has identified a world-class copper deposit with unusually high grade. Goldman uses it to illustrate how higher grade lowers costs, capital intensity, and environmental footprint. Technology stack: sensors, data system, and models (Priority: 4/5): Kobold’s stack includes proprietary sensors, a system that unifies structured and unstructured data, and multiple predictive models that are continuously retrained with new field observations and uncertainty estimates. Regulation, community relations, and jurisdictional risk (Priority: 4/5): Goldman stresses that exploration success depends not only on finding deposits but on property rights, tax/royalty stability, and earning local social license to operate through community and indigenous engagement. Philosophy, uncertainty, and epistemology in exploration (Priority: 4/5): The company has an internal epistemology framework and even a chief philosopher to help ensure hypotheses are falsifiable, uncertainty is quantified, and alternative explanations are evaluated rather than forcing a single model.
Key Arguments: Exploration is more valuable than mining itself because a small amount of capital can generate 100x–1000x returns if the right deposit is found. The industry’s main constraint is information, not ore in the ground; the real challenge is knowing where the deposits are. Traditional surface prospecting is largely exhausted; future discoveries require concealed, underground targets and better technology. Combining heterogeneous data—especially legacy public-domain records with modern imagery and sensors—improves predictive power more than any single dataset. AI is most effective when paired with geoscientists in the field who generate new ground-truth data and retrain models daily. High-grade deposits are economically superior because they reduce waste, hauling, plant size, capital intensity, and environmental footprint. Great projects still attract capital; the scarcity is quality deposits, not money. Exploration success requires social license, property-rights stability, and hyper-local stakeholder management, not just technical excellence. The company’s philosophical approach helps avoid confirmation bias and forces falsifiable predictions before collecting data. Copper and lithium demand will be enormous over the next decades due to batteries, AI data centers, and electrification, so finding new supplies is strategically important.
Data Points: Annual investment: Over $100 million - Kobold’s stated annual investment across its exploration portfolio Project count: 70 projects - Scale of Kobold’s portfolio mentioned in the introduction Continents of operation: 4 continents - Kobold operates across North America, Europe, Australia, and Africa Portfolio size: More than 60 projects - Goldman later describes the exploration portfolio Copper grade at Mingomba: Over 5% copper - Core of the Zambia deposit identified by Kobold Average grade of operating copper mines: About 0.6% copper - Used as a benchmark for why Mingomba is economically exceptional Relative grade advantage: About 10x higher grade - Comparison of 5% copper deposit versus a 0.5% deposit Target discovery cost: $50 million to $100 million per discovery - Kobold’s stated performance goal for exploration efficiency Industry success rate trend: 10x worse in the last 30 years - Goldman says exploration success has declined as targets have become harder Industry-wide discovery outcome: Less than one high-quality economic deposit per $1 billion of exploration spend today - Goldman contrasts current outcomes with the past Historical discovery outcome: Eight discoveries per $1 billion of exploration spend 30 years ago - Baseline comparison for industry decline Lithium production requirement: 10x increase relative to today - Needed to support a high penetration of battery-powered devices Copper requirement: More copper in the next 25 years than has been mined in all human history - Goldman’s estimate of future copper demand for batteries and AI infrastructure Spectral data resolution: 600 colors - Kobold’s proprietary hyperspectral imaging system Deposit lifecycle: 20+ years underwriting; often 50–70 years of operation - Goldman explains mine valuation and long asset lives
Pivotal Quotes: "Our business is focused on exploration, and it's focused on exploration for a couple of reasons. One is because there's way more value to be created there. And the second is that's where technology can be really differentiating." — Josh Goldman: Explaining why Kobold focuses on exploration rather than mining "The scarce resource is not lithium or copper metal in the ground. It's actually information." — Josh Goldman: Central framing of exploration as an information problem "There is no way to isolate the AI from the HI, there's no way to isolate one piece of data that's uniquely powerful." — Josh Goldman: Describing Kobold’s approach to combining AI with human intelligence and multiple data sources
Implications: The conversation suggests future mineral supply will depend on data integration, AI, and disciplined uncertainty management more than brute-force drilling. For investors and industry, the biggest opportunity is finding rare, high-grade deposits in stable, socially viable jurisdictions.