No Priors• Apr 17, 2025• 40mInterview

No Priors Ep. 111 | With KoBold Metals Co-Founder and President Josh Goldman

From No Priors

Josh Goldman•Co-Founder and President, KoBold Metals

Executive Summary

Cobalt Metals is a mineral exploration company using proprietary AI, machine learning, and a massive collection of geoscience data to improve the success rate of finding critical minerals like copper and lithium.
The company has discovered the Mingomba deposit in Zambia, described as the highest-grade large copper deposit in the world not yet in development, with a core concentration over 5% copper.
The demand for copper and lithium is projected to surge dramatically due to the energy transition (batteries, EVs) and the build-out of AI infrastructure, requiring more copper to be mined in the next 25 years than in all of human history.
The mineral exploration industry's success rate has declined tenfold over the past 30 years, creating a significant opportunity for a technology-driven approach to de-risk and accelerate discovery.

8 quotes

Concerns Raised

The historically low and declining success rate of the mineral exploration industry.
Geopolitical and regulatory risks associated with operating in various global jurisdictions.
China's dominant position in the downstream processing of critical minerals.

Opportunities Identified

Massive structural demand for copper and lithium driven by the energy transition and AI.
Applying modern AI and data science to a traditionally slow-to-innovate industry.
The potential for 100-1000x returns on successful mineral discoveries.
Leveraging vast amounts of underutilized public and historical geoscience data.

Key Themes

AI-Driven Mineral Exploration

Cobalt Metals is applying modern data science to the traditional field of mineral exploration. By aggregating vast, disparate datasets (both structured and unstructured) and using proprietary AI/ML models, they aim to increase the probability of discovering high-quality ore deposits.

This technology-first approach addresses the industry's declining success rate and positions the company to more efficiently find the resources needed for the global energy and technology transition.

The Material Bottleneck of the Future Economy

The global shift towards electrification, batteries, and AI is creating unprecedented demand for specific minerals. The speaker highlights that the next 25 years will require more copper than has ever been mined, and lithium production needs to increase tenfold to meet demand.

This massive supply-demand imbalance creates a significant economic opportunity for successful exploration companies and highlights a critical potential bottleneck for major technological trends.

Exploration as an Information Problem

The speaker reframes mineral exploration not as a problem of resource scarcity, but of information scarcity. The core challenge is not a lack of minerals in the Earth's crust, but a lack of precise information about where high-concentration deposits are located, which is what their data-centric approach is designed to solve.

This perspective justifies the heavy investment in data aggregation and AI, suggesting that the highest returns will go to those who can most effectively process information to reduce uncertainty.

The Economics of Discovery

While the success rate of exploration is historically low, the financial returns on a successful discovery are extraordinary, potentially 100 to 1,000 times the initial investment. Once a deposit is found, its value can be calculated with relative certainty based on ore grade, size, and commodity prices.

This high-risk, high-reward profile explains the business model. Improving the success rate, even marginally, through technology can lead to massive value creation.

Integrating Human and Artificial Intelligence

The company emphasizes a symbiotic relationship between geologists (human intelligence, HI) and AI. AI models provide predictions and quantify uncertainty, while geologists provide domain expertise, make hypotheses, and collect new data in the field, which in turn retrains the models daily.

This highlights that even in a data-driven field, domain expertise is irreplaceable. The most effective approach combines the predictive power of machines with the interpretive and hypothesis-generating skills of human experts.

Get started free

Topics

Mineral Exploration Artificial Intelligence Machine Learning Geoscience Copper Lithium Energy Transition Battery Supply Chain AI Infrastructure Data Science Mining Technology Zambia Resource Scarcity Hyperspectral Imaging Geopolitics of Minerals

Processed Mar 31, 2026 yt-dlp + mlx-whisper + Gemini