Mason Dykstra is co-founder and Chief Science and Technology Officer of MinersAI, a company building the foundational data layer for mineral exploration and mining to accelerate the discovery of critical minerals needed for the energy transition and technological advancement.
Can you tell me about your background and what led you to found MinersAI?
I have a mixed academic and industry background. I was a researcher at University of California Santa Barbara after my PhD and a research professor at Colorado School of Mines for five years. I’m a geologist by background, and was doing deep ocean research on sediment movement and continental margin formation. After academia, I went back to the oil and gas industry where I’d worked before my PhD, doing everything from operations geology to production.
I ran a research center for Statoil, the Norwegian state oil company that became Equinor, in Texas. My last oil and gas role was with Anadarko Petroleum before they were acquired by Occidental in 2019. I then joined Enthought, a scientific software company I’d hired to prototype machine learning applications at Anadarko. I ran their energy business for a few years, building machine learning-forward technical analysis software for supermajors.
Looking at the mining industry in early 2023, I realized they were much further behind technologically than oil and gas. There was huge potential to bring technology to the space, and help improve the vertical. I met my co-founder Tomi at PDAC, the big Canadian mineral exploration and mining conference, and we decided early on to build what’s fundamentally missing from mining: a foundational data layer for the entire vertical.
A weekly dispatch featuring exclusive interviews with deep tech founders & a roundup of the most important deep tech news.
What specific problem is MinersAI solving in the mining industry?
The mining sector is divided into junior exploration companies that find new resources, larger mining companies that develop and operate mines, and financial investors. Each group plays a key role but there are massive inefficiencies in how data moves between them.
Juniors aggregate their own data and do their best to standardize it. When they sell a project to a mining company, that company has to convert the junior’s data into their framework. A lot gets lost in these handovers. Then when attracting capital, they need to present data to financial investors in yet another format. Data rooms vary enormously in quality and standardization, making it difficult for investors to analyze projects easily.
We’re building unified data products across that entire value chain. We make it easy for juniors to start projects and find discoveries. We make handovers to mining companies seamless. We make it simple to attract capital. We’re reducing friction at every step by solving the data bottlenecks that slow down the entire industry.
How does your technology actually work to structure and standardize all this geological data?
The variation in data types is actually huge, which is why no one’s done this at scale before. We take public data from sources like the US Geological Survey, but it comes in tons of different formats because much of it is historic. Data from the 1960s through 1980s used different analytical methods and instruments than today. The error bars on 1970s data aren’t the same as data acquired last year. How do you use all this data together in large-scale models without accounting for that?
We structure and standardize the data enough that it can be used together while maintaining chain of custody information about the data source. Anyone can understand the quality of each dataset relative to others. That’s with public data. Companies also bring us their proprietary data, and we apply the same process to combine it with public geoscience data. Now they can use all the data together computationally. We can help them look for trends, assess prospectivity, run all kinds of analyses, and leverage rapidly evolving AI techniques for data analysis.
Why is this technology particularly important right now?
We’ve got eight billion people heading to 10 billion, with more people moving into the middle class. To deliver the quality of life people want, we need more natural resources: copper, nickel, cobalt, rare earth elements, all kinds of critical minerals that play key roles in everyone’s lives. Most people just don’t think about them.
The rise in AI is straining the value chain for critical minerals. Think about all these data centers being built. They need copper wiring, rare earth elements, specific magnets, all kinds of minerals. Then there’s the climate transition from fossil fuels to renewables. To facilitate that transition, we’ll need a lot more minerals: copper, rare earth elements, even high-purity quartz is in short supply.
There are huge bottlenecks in the value chain that will be very challenging to alleviate. We need to unlock resources we can’t find or develop right now. The easy deposits have been found. Future discoveries are hidden underground and much harder to find. By aggregating all this data in the right framework, we can query it in natural language ways to understand the geological evolution that leads to big deposits of critical minerals.
Where is MinersAI today and what’s on the horizon?
We’re right in the middle of a seed round. We’ve raised a bunch of Simple Agreement for Future Equity (SAFE) and are looking to convert that with a priced round. We’ve just signed a big contract with one of the largest mining companies in the world, and another contract with one of the hottest countries in the world to help them analyze the prospectivity of their mineral systems.
My vision is that we’ll have the ability to ask questions of geological data in ways that enable people to efficiently target big deposits of key minerals. That’s where it gets really exciting. I hope to use this for fundamental geoscience research as well, not just the commercial side.
Critical minerals security is a hot topic now. When we started, the only driver VCs cared about was the energy transition. Now they all see the geopolitical issues associated with critical mineral supply chains. It’s definitely a tailwind for what we’re doing. We’re sailing in the right direction.
How do you define deep tech?
Deep tech is technology that requires significant scientific or engineering breakthroughs to work, often involving long development cycles and substantial capital before you can validate whether it functions. It’s not just applying existing technology in new ways but pushing the boundaries of what’s technically possible. In our case, we’re not just building another data platform. We’re solving fundamental problems in how heterogeneous geological data can be structured, standardized and made computationally accessible at scale, which requires deep domain expertise in geology, data science, and machine learning working together. The problems are hard enough that they’ve resisted simple solutions for decades.