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

The Earth has abundant lithium and copper; the real scarcity is reliable information about where economically viable, highly concentrated deposits are located. KoBold’s business centers on improving prediction quality to reduce uncertainty and exploration failures.

With a few million dollars, a successful discovery can create 100–1000x returns, whereas operational mining is more commoditized. KoBold targets $50–$100 million of exploration spend per major discovery, versus an industry that now often spends $1 billion for less than one high-quality find.

KoBold integrates satellite imagery, airborne surveys, rock and soil samples, historical maps, regulatory filings, and text reports into a unified data system. Their advantage comes from high-dimensional data fusion and uncertainty-aware models, not from any one special dataset or sensor.

Geologists are directed to locations where models are least certain, collect new ground-truth data, and feed it back daily. This continuous loop between human experts and machine learning systematically sharpens predictions about surface and subsurface geology.

Mingomba’s core is over 5% copper versus the global average of ~0.6%. For the same copper output, a 10x higher grade means roughly 10x less rock moved, lower capital and operating costs, and a smaller environmental impact—key to being a low-cost producer in a commodity market.