AI trained on ten million synthetic Earths outperforms conventional copper estimates at TSXV-listed ATEX's Valeriano project

Fleet Space Technologies describes Tau 3.8 as a system designed to combine its 3D seismic intelligence with drilling, geology, geophysics, and mineralogy data. The stated goal is to reconstruct hidden geological structures, rank potential targets for resource expansion, and identify further measurements that could reduce uncertainty, allowing mining companies to make more informed drilling decisions.

The company frames mineral discovery as a bottleneck for the AI industry, citing increasing demand for copper for data centers and transmission lines, particularly in North America. Traditional exploration methods are characterized as slow, expensive, and uncertain. Tau 3.8 aims to make exploration more capital-efficient by providing clearer insights into orebodies, potentially reducing the number of unproductive drillholes, which can cost hundreds of thousands of dollars each.

At ATEX Resources' Valeriano copper-gold project in Chile, Tau 3.8 was reportedly tested by combining Fleet's ambient noise tomography and magnetotelluric surveys with machine learning. The model, when evaluated against drillholes withheld from its training, predicted copper grades with 38% less error than a conventional geostatistical estimate. For deeper step-out holes, the reported error reduction was up to 67%, suggesting improved accuracy in predicting mineral concentrations.