Terra AI raises $20 million to scale subsurface AI for faster mineral discovery
Terra AI raises $20 million to scale its subsurface AI, promising to cut drilling, accelerate mineral discoveries and reduce exploration risk across sectors.
Terra AI announced a $20 million financing round led by Khosla Ventures to expand its subsurface modelling platform, a move the company says will accelerate mineral discovery and reduce exploration costs. The funding is earmarked to scale the firm’s generative modelling engine and push enterprise deployments across mining, carbon storage and geothermal projects. The company’s claim that the technology can lower drilling needs and speed development has attracted interest from major miners and energy companies as demand for critical minerals intensifies.
Funding aimed at a strategic bottleneck
The new capital, Terra AI says, will be used to grow engineering teams, deepen integrations with industry data systems and deploy larger models that infer geological scenarios. Management positions the round as a stepping stone from pilot projects to wider industry adoption, enabling multi-site implementations and compliance-ready workflows. Investors framed the bet as backing a tool that addresses a structural constraint in global supply chains for copper, lithium, nickel, cobalt, graphite and rare earths.
Platform merges diverse subsurface datasets
Terra AI’s software ingests drilling logs, geochemical assays, magnetic and seismic surveys, and emerging techniques such as muon tomography, and then combines them into three-dimensional probabilistic models. Rather than producing a single static map, the platform generates ensembles of geological scenarios with associated likelihoods, aiming to make uncertainty explicit for technical and commercial decision-makers. Company founders argue this integrated approach converts fragmented observations into actionable intelligence about where to target further data collection or drilling.
Company-reported gains in drill efficiency and timelines
According to Terra AI, pilots have reduced required drilling depth by 50–60 percent and shortened discovery-to-development timelines for major deposits by four to five years in some cases. The firm also reports improvements in resource-model forecast quality by two to three times and, in isolated examples, an increase in defined mineral resources of up to 2.5 times. Those performance figures come from the company’s internal analyses and early deployments; industry observers say broader validation will be key before those metrics are accepted as standard.
Major miners and energy firms among early users
Several large mining companies and mid-tier explorers are testing or using the platform, the company says, including BHP and Rio Tinto, alongside junior explorers and energy-sector partners such as OMV. That mix reflects Terra AI’s positioning as a decision-infrastructure supplier rather than a niche mapping tool, with use cases ranging from grade estimation to reservoir assessment. For smaller explorers, where capital is tight and the number of new finds drives survival, the promise of producing better models with fewer holes is particularly appealing.
Carbon storage and geothermal pilots show cross-sector potential
A collaboration with OMV demonstrates the platform’s application beyond metals: Terra AI has been used to evaluate carbon capture and storage (CCS) options and geothermal sites, where subsurface uncertainty drives cost and regulatory risk. Project partners in an early North Sea CCS case reported that probabilistic modelling increased theoretical storage volumes, cut data-collection costs and raised expected project value by more than 25 percent while lowering quantified subsurface risk by roughly half. Those outcomes underscore why energy companies see the same modelling challenges in low-carbon projects as miners see in exploration.
Adoption barriers and data dependency
Industry executives caution that the mining sector is conservative, capital-intensive and highly regulated, and a software platform must win trust from geoscientists, financiers and permitting authorities to influence major decisions. Terra AI’s effectiveness depends on the quality and completeness of underlying data: legacy archives in mature mining regions can be rich but inconsistent, while frontier territories often lack baseline information. Moreover, probabilistic models do not replace drilling; they are designed to prioritise where and when to drill, reduce unnecessary holes and sharpen cost-benefit judgements.
Datadriven exploration proponents say there are broader economic and environmental upside potentials if these tools scale. Fewer unsuccessful drill campaigns would lower direct costs and reduce land disturbance, while more reliable forecasts could shorten project timelines that now average many years from discovery to production. For western governments and companies seeking more resilient supply chains for critical minerals, faster, cheaper and less invasive exploration could become a strategic enabler—provided the technology proves repeatable across diverse geology and regulatory regimes.
The Terra AI investment highlights growing appetite for technology that quantifies subsurface uncertainty, but the company still faces the challenge of turning pilot successes into routine industry practice. If its models consistently deliver lower risk and faster timelines, they could reshape how the mining and energy sectors prioritise projects and allocate capital, with implications for supply chains central to electrification and climate goals.
