Critical Mineral Exploration and AI

Through our work in industrial manufacturing and AI infrastructure, we’ve become captivated by the beast that is energy. We’ve watched the Department of Energy roll out ambitious domestic energy goals, while AI companies have quietly evolved into energy companies right before our eyes. The macro trends driving this shift are hard to miss and undoubtedly familiar to many of you: massive increases in domestic energy needs driven by both manufacturing capacity and compute, new industries springing up around the plummeting costs of clean energy production and storage, and, of course, geopolitical forces compelling a reconfiguration of global energy supply chains.

At Primary, our industrial focus has historically centered on supply chain and manufacturing. In 2025, we’re expanding our focus to include energy and developing a thesis for how to approach the sector as seed-stage investors. We’re excited to share this journey with you, as we believe the future of energy will be shaped in part by startups capable of creating tens, if not hundreds, of billions of dollars in value.

Throughout 2025 we’ll be expanding our networks with experts across mining, batteries, utilities, power, chemical engineering, and more, digging deep and learning as much as we can. And each month, we’ll share our learnings and insights along the way. We’ll never become true experts across these sectors, but we’ll strive to be their most curious students. If you’re receiving this email, it’s because we deeply value your expertise and perspective. We’re seeking your feedback, criticism, and insights to refine our understanding and thesis.

Starting at the Core: Mining Exploration

This first issue focuses on the literal foundation of the energy transition: mining exploration. Over the past six months, we’ve spent time with leaders at mining majors, startups innovating in mineral exploration, and other stakeholders. We’ve developed a clear perspective on where startups can win in this space and have even made our first investment in the sector (more on that soon).

Mining exploration is becoming harder, riskier, and more capital-intensive, yet its criticality to our energy transition has never been more apparent.  Demand for base metals such as copper, lithium, cobalt, and nickel is projected to skyrocket. For example, the DoE and EU anticipate a 40x increase in lithium demand and the need to double all historic copper mining by 2040. These materials are essential for grid buildouts, battery manufacturing, semiconductors, EVs, and more. While the debate continues about whether to meet this demand through increased mining or innovative recycling, one thing is clear: there is no viable future without more mining.

Despite these projections, base metal exploration is grossly underfunded relative to demand forecasts, and especially compared to more established resource markets like gold, coal and oil. Exploration remains a fragmented, net capital-destructive industry dominated by small, undercapitalized companies. On average, only 0.2% of exploration projects succeed, with hit rates declining as deposits become deeper and harder to find. Most exploration firms raise capital in piecemeal fashion while juggling a patchwork of service providers that survey land, test samples, and provide analysis. Unfortunately, many of these service providers operate on a cost-plus model, incentivizing additional capital deployment regardless of a project’s likelihood of success.

Exploration teams typically drill 10–50 holes in grids over 2–4 years, often wasting tens of millions of dollars on unsuccessful projects. However, the last five years have seen a surge of attention and capital focused on leveraging advanced computational capabilities to improve the odds of discovery, offering the promise of fundamentally altering the risk equation for mining exploration. Companies like KoBold Metals have raised capital on this premise, drawing inspiration from the historical application of computational methods in oil and gas exploration. For decades, oil and gas companies sat on vast troves of geophysical data, much of it undigitized. Today, AI-driven models can analyze this data, helping companies determine where to drill and justify the massive cap-ex required for exploration.

We believe the mining industry is at a breaking point—demand is rapidly outpacing supply, and existing technologies are no longer sufficient to close the gap. Each dollar invested in traditional exploration is yielding diminishing returns, making it clear that the status quo is unsustainable. Unless we accelerate timelines from site identification to production, we risk bottlenecking the energy transition. This urgency creates an opening for startups that can leverage AI and other breakthrough technologies to revolutionize the industry.

Market Breakdown: Emerging Models in Mining Exploration

We’ve observed several models emerging in the mining exploration space, each attempting to address inefficiencies and risks in distinct ways:

1. Geophysics Consultants: These service providers comprise teams of geologists and field technicians conducting surveys and analysis. Operating on cost-plus models, they often encourage continued investment in projects, even when the odds of success are low.
2. Vertically Integrated AI Exploration Companies: These companies develop proprietary technology to identify mineral deposits and oversee the entire exploration process. While capital-intensive, they hold the promise of capturing more value by owning the discovery and its associated assets. Examples include KoBold Metals and Earth AI.
3. AI Survey-Exploration Hybrids: These companies leverage advanced technologies for geophysical surveys and align themselves with customers by taking equity or co-investing in projects. Their focus on deploying capital efficiently incentivizes the development of better technologies. Examples of these include ExploreTech and VerAI. 
4. Hardware-as-a-Service (HaaS): Companies in this category aim to significantly reduce exploration costs through innovations like automated drilling rigs (e.g., Durin) or surgical extraction methods (e.g., Novamera). Their close ties to capital deployment create opportunities to invest alongside their customers.
5. Software Players: Focused on reducing costs and accelerating exploration, these companies digitize processes and offer tools for project management and optimization. Startups like MinersAI and platforms like Seequent and Vrify fall into this category. However, their growth potential is limited by the total addressable market (TAM), the challenge of selling into resource-constrained exploration players, and their lack of direct participation in mining asset upside.

Our Perspective

As seed investors, we’re most excited by AI survey-exploration hybrids and HaaS companies. These models balance innovation with capital efficiency and align with exploration projects to reduce risk and maximize upside exposure. Conversely, software players face TAM constraints, while vertically integrated companies like KoBold Metals are too capital-intensive for our investment focus. Additionally, many experts have questioned whether AI is the true differentiator for these vertically integrated companies, attributing much of their success to their unique access to de-risked projects.

Nonetheless, we believe AI-enabled discovery services will transform exploration by drastically reducing costs and increasing the odds of success. These companies will also be uniquely positioned to invest alongside exploration projects, leveraging their insights to cherry-pick the most promising opportunities. In public markets, asset-light models like royalties and streaming companies (e.g., Franco Nevada and Wheaton Precious Metals) have proven highly lucrative, trading at 30x P/E ratios, compared to mining companies’ 8.5x. However, these firms focus on lower-risk, construction-stage projects. We see an important opportunity for similar models tailored to high-risk, early-stage base metal exploration, and we believe AI will be the key enabler of this new category.

This is just the beginning of our journey into the energy sector. We’re eager to learn from and collaborate with you. What are we missing here? Where are we off base? Where should we be pushing our thinking further? We hope this becomes an ongoing conversation with you all. If you have thoughts, critiques, or ideas, please don’t hesitate to share them.