Announcements
Building the AI-Native Platform for Subsurface Modeling
Abstract
Announcing our $20M Series A, led by Khosla Ventures, with strategic investment from BHP Ventures
Critical minerals, geothermal power, and gigaton-scale carbon storage are core to the net-zero transition, and all depend on understanding the earth’s subsurface. However, seeing it with clarity remains deeply difficult. That is why we founded Terra AI. Today, we’re announcing our $20 million Series A led by Khosla Ventures, with strategic investment from BHP Ventures.
The round follows Khosla's lead investment in our $3.4M seed round in 2023, a signal of conviction in our vision for artificial intelligence’s role in the subsurface industry, which also included Rio Tinto Founders Factory, Storyhouse Ventures, Plug and Play, The TomKat Center for Sustainability, and Climate Capital.
The Critical Resources Bottleneck
The world is heading into a structural shortage of the underground resources it needs most.
The IEA projects a 30% copper supply deficit by 2035 as ore grades decline and mines age out, while demand is accelerated by electrification, EV fleet expansion, and AI infrastructure. The world isn’t developing new resources fast enough to replace growing consumption: discovery rates have fallen sharply, grassroots exploration budgets sit at record lows, and the average timeline from discovery to production now exceeds 17 years. Yet, conventional methods trade more drilling, more time, and more capital for each incremental insight, with no certainty on potential return.
Simultaneously, enhanced geothermal systems are drawing billions in new investment as data centers search for reliable clean power, while demand for carbon storage grows, both remaining constrained by the cost and risk of characterizing underground reservoirs.
Across all three domains, the bottleneck is geological: you cannot see what is underground. The industry needs a fundamentally different way to navigate subsurface uncertainty — one that begins with modeling the uncertainty, rather than trying to eliminate it.
A New Approach to the Subsurface
Terra AI is building the industry's first AI-native subsurface modeling platform. The platform is built around quantifying uncertainty to accelerate exploration. From that foundation, our models produce continuous 3D mapping of subsurface resources alongside a map of uncertainty in every estimate, so teams can be confident about resource models to enable data-optimized development across mining, geothermal, and carbon storage.
Any geological description and dataset corresponds to a wide range of possible underground realities. Terra AI's platform holds the full range of uncertainty open to reason across every possibility.
Our patented technology fuses nearly every type of data explorers collect, from drill core data to conventional geophysics to emerging methods like muon tomography, and generates geological models with quantified uncertainty. This allows teams to evaluate the full range of subsurface scenarios to identify where target resources are located, what they could be, and critically, where the uncertainty remains. From there, uncertainty becomes the map — teams can pinpoint precise exploration targets, reduce risk, and progressively sharpen the subsurface picture, maximizing value from every exploration dollar spent.
The analogy to draw for our technology is not to chatbots or agentic wrappers. It is to AlphaFold. Where DeepMind compressed a 15-year drug discovery process into 12 months by applying artificial intelligence at the scientific frontier, Terra AI is pursuing the same kind of step-change for subsurface modeling and prediction.
By going beyond geological visualization to deliver risk-quantified insights, our technology supports explorers on the decisions that matter most. At the asset level, teams can determine where to drill and what data to collect next. At the portfolio level, teams can decide which assets to acquire and how to accelerate and de-risk resource development. In all cases, teams can rigorously assess project economics earlier than ever, building resource pipelines more efficiently to help the world meet its growing needs.
Historically, subsurface data has lived and been used across disconnected datasets and tools. Terra AI brings them together — hypotheses, maps, geophysics, drill cores, and more — into continuous models on a single platform where teams can build better models, test ideas, and plan their next move.
Validation in the Field
Our results have been consistent across multiple engagements: we have repeatedly reduced required drilling meters by 50-60%, accelerated Tier-1 projects by 4-5 years, improved resource model prediction by 2-3x, and increased defined mineral resources by up to 2.5x. Our customer base now includes BHP and Rio Tinto — the world's two largest mining companies — alongside mid-cap miners, junior exploration companies, and reservoir companies including carbon storage work with OMV. After validating individual components across projects last year, we are now deploying the full methodology and modeling stack together, with active engagements with several top-tier companies underway.
Expanding Across the Subsurfaces Market
Our platform is system-agnostic by design because geological uncertainty underlies every subsurface application. Our modeling stack has been deployed across heavy rare earths, gold, and polymetallic systems, and extends across broader subsurface domains including geothermal and carbon storage reservoirs. Over the next year, we are expanding into the junior mining market. Junior companies are collectively responsible for roughly 70% of global ore discovery each year, making them an essential part of building the critical minerals supply chain.
A fluid particle simulation of multi-well fluid injections in an enhanced geothermal system over a 25.5 year life. Traditional simulations require hours to run, with compute being the key bottleneck. Terra AI’s technology can run fluid flow simulations in seconds.
We have been exploring net-negative reservoir applications since our founding, and a new wave of demand is pulling our technology deeper into that space. Both face the same core challenge: capturing the dynamic subsurface and predicting how fluids behave is slow, costly, and high-stakes. Drilling accounts for 50-70% of project costs in geothermal, while carbon storage projects can stall for four to seven years before injection begins. Our approach pairs uncertainty-aware subsurface characterization with simulations orders of magnitude faster, enabling teams to optimize development plans with confidence. We’re already deploying on CCS projects, including a carbon storage project with OMV, where our modeling has safely compressed characterization timelines and reduced risk.
What's Next
The Series A funds the next phase of Terra AI’s growth: scaling the generative modeling engine, deepening commercial deployments across mining and reservoirs, and advancing the research that makes the platform possible.
Subsurface resources will shape the next decade of global development. Understanding them rigorously is critical to enabling us to build faster, waste less, and develop responsibly. To make this happen, we are growing our team with people who drive Terra AI forward: geoscientists, AI researchers, engineers, and other mission-driven problem solvers who collaborate effectively across disciplines.
If that sounds like it could be you, please reach out. We’d love to hear from you.
Media Contact
media@terraai.earth
Announcing our $20M Series A, led by Khosla Ventures, with strategic investment from BHP Ventures
Critical minerals, geothermal power, and gigaton-scale carbon storage are core to the net-zero transition, and all depend on understanding the earth’s subsurface. However, seeing it with clarity remains deeply difficult. That is why we founded Terra AI. Today, we’re announcing our $20 million Series A led by Khosla Ventures, with strategic investment from BHP Ventures.
The round follows Khosla's lead investment in our $3.4M seed round in 2023, a signal of conviction in our vision for artificial intelligence’s role in the subsurface industry, which also included Rio Tinto Founders Factory, Storyhouse Ventures, Plug and Play, The TomKat Center for Sustainability, and Climate Capital.
The Critical Resources Bottleneck
The world is heading into a structural shortage of the underground resources it needs most.
The IEA projects a 30% copper supply deficit by 2035 as ore grades decline and mines age out, while demand is accelerated by electrification, EV fleet expansion, and AI infrastructure. The world isn’t developing new resources fast enough to replace growing consumption: discovery rates have fallen sharply, grassroots exploration budgets sit at record lows, and the average timeline from discovery to production now exceeds 17 years. Yet, conventional methods trade more drilling, more time, and more capital for each incremental insight, with no certainty on potential return.
Simultaneously, enhanced geothermal systems are drawing billions in new investment as data centers search for reliable clean power, while demand for carbon storage grows, both remaining constrained by the cost and risk of characterizing underground reservoirs.
Across all three domains, the bottleneck is geological: you cannot see what is underground. The industry needs a fundamentally different way to navigate subsurface uncertainty — one that begins with modeling the uncertainty, rather than trying to eliminate it.
A New Approach to the Subsurface
Terra AI is building the industry's first AI-native subsurface modeling platform. The platform is built around quantifying uncertainty to accelerate exploration. From that foundation, our models produce continuous 3D mapping of subsurface resources alongside a map of uncertainty in every estimate, so teams can be confident about resource models to enable data-optimized development across mining, geothermal, and carbon storage.
Any geological description and dataset corresponds to a wide range of possible underground realities. Terra AI's platform holds the full range of uncertainty open to reason across every possibility.
Our patented technology fuses nearly every type of data explorers collect, from drill core data to conventional geophysics to emerging methods like muon tomography, and generates geological models with quantified uncertainty. This allows teams to evaluate the full range of subsurface scenarios to identify where target resources are located, what they could be, and critically, where the uncertainty remains. From there, uncertainty becomes the map — teams can pinpoint precise exploration targets, reduce risk, and progressively sharpen the subsurface picture, maximizing value from every exploration dollar spent.
The analogy to draw for our technology is not to chatbots or agentic wrappers. It is to AlphaFold. Where DeepMind compressed a 15-year drug discovery process into 12 months by applying artificial intelligence at the scientific frontier, Terra AI is pursuing the same kind of step-change for subsurface modeling and prediction.
By going beyond geological visualization to deliver risk-quantified insights, our technology supports explorers on the decisions that matter most. At the asset level, teams can determine where to drill and what data to collect next. At the portfolio level, teams can decide which assets to acquire and how to accelerate and de-risk resource development. In all cases, teams can rigorously assess project economics earlier than ever, building resource pipelines more efficiently to help the world meet its growing needs.
Historically, subsurface data has lived and been used across disconnected datasets and tools. Terra AI brings them together — hypotheses, maps, geophysics, drill cores, and more — into continuous models on a single platform where teams can build better models, test ideas, and plan their next move.
Validation in the Field
Our results have been consistent across multiple engagements: we have repeatedly reduced required drilling meters by 50-60%, accelerated Tier-1 projects by 4-5 years, improved resource model prediction by 2-3x, and increased defined mineral resources by up to 2.5x. Our customer base now includes BHP and Rio Tinto — the world's two largest mining companies — alongside mid-cap miners, junior exploration companies, and reservoir companies including carbon storage work with OMV. After validating individual components across projects last year, we are now deploying the full methodology and modeling stack together, with active engagements with several top-tier companies underway.
Expanding Across the Subsurfaces Market
Our platform is system-agnostic by design because geological uncertainty underlies every subsurface application. Our modeling stack has been deployed across heavy rare earths, gold, and polymetallic systems, and extends across broader subsurface domains including geothermal and carbon storage reservoirs. Over the next year, we are expanding into the junior mining market. Junior companies are collectively responsible for roughly 70% of global ore discovery each year, making them an essential part of building the critical minerals supply chain.
A fluid particle simulation of multi-well fluid injections in an enhanced geothermal system over a 25.5 year life. Traditional simulations require hours to run, with compute being the key bottleneck. Terra AI’s technology can run fluid flow simulations in seconds.
We have been exploring net-negative reservoir applications since our founding, and a new wave of demand is pulling our technology deeper into that space. Both face the same core challenge: capturing the dynamic subsurface and predicting how fluids behave is slow, costly, and high-stakes. Drilling accounts for 50-70% of project costs in geothermal, while carbon storage projects can stall for four to seven years before injection begins. Our approach pairs uncertainty-aware subsurface characterization with simulations orders of magnitude faster, enabling teams to optimize development plans with confidence. We’re already deploying on CCS projects, including a carbon storage project with OMV, where our modeling has safely compressed characterization timelines and reduced risk.
What's Next
The Series A funds the next phase of Terra AI’s growth: scaling the generative modeling engine, deepening commercial deployments across mining and reservoirs, and advancing the research that makes the platform possible.
Subsurface resources will shape the next decade of global development. Understanding them rigorously is critical to enabling us to build faster, waste less, and develop responsibly. To make this happen, we are growing our team with people who drive Terra AI forward: geoscientists, AI researchers, engineers, and other mission-driven problem solvers who collaborate effectively across disciplines.
If that sounds like it could be you, please reach out. We’d love to hear from you.
Media Contact
media@terraai.earth
Keep exploring
Load More
Load More