- Drifft Newsletter
- Posts
- Who's Making Money pt1 – Startups and Business Models in the Mining Value Chain
Who's Making Money pt1 – Startups and Business Models in the Mining Value Chain
Part 1 of a series exploring resource industry startups, business models, and why they matter.
David here.
💡What’s New: Physical Resource Tech - A Review of Business Models
🤔 Opinion: Why AI disrupts professional services and how mining benefits.
🛠️ Tools and Data: Stunning new opensource tools for imagery analytics.
Want to feature your service or product in DRIFFT? Grab an ad spot here.
💡What’s new
Until the 1970s most entrepreneurs and investors would have been familiar with mining. But as fiat currency replaced gold and silver, knowledge of mining began to decline along with employment in the sector. Today the trend is reversing–not fiats, just the interest in mineral resources.
I’ve been meeting resource startups and reviewing business models since the start of the year. As AI displaces professional services, I expect those costs to fall and to see a rise in value of physical resources–especially critical
minerals. Therefore, a review of the resource value chain may help us in the coming decades.
To simplify, I divide the resource industry into 6 phases: Exploration, Extraction, Processing, Services, Logistics, and Licensing. In this edition, we review Exploration.
— — — — — — — — — — — — — — — — — — — — — — — — — — — — — —
Exploration
Finds resources.
Find = locate + quantify (amount and grade). Activities include locating, drilling, assaying.
Value: 100s of $Ms. Exploration is one of the greatest value-additive links in the chain as it removes the bulk of uncertainty and risk.
Example Startups & Business Models:
verAI - uses proprietary AI modeling to locate resources underground. Their primary business model is using AI software to locate undiscovered resources and then buy+hold land / mineral rights and develop these projects together with their partners. In a recent deal in Chile, they partnered with a mining company and negotiated royalties on mined resources. Recall that large mines are assets that can pay out over 20+ years! Scale of tech + immediate cashflow = sweet deal.
kobold - Fully integrated exploration and extraction. Like verAI, they employ AI to locate resources. But Kobold accepts the additional risk of extraction. Currently operating in US, Canada, South America, Africa, and Greenland.
Why they matter:
Historically, exploration accounts for a large chunk of mining costs. It’s possible with today’s compute, databases, and algorithms to characterize underground resources using a combination of historical data (most of it proprietary) and new survey techniques. This means that companies can reduce the failure rate of resource discovery (currently at around ~98%), and drastically reduce the cost of the initial core sampling.
As these services become more widely available, we can expect to see costs for initial discovery to fall and for quality of assays to rise. Some likely outcomes would include:
Smaller funding checks would be more meaningful. Typically large banks, governments, and private equity are risking hundreds of millions. If capital needs can fall, more VCs might be interested in participating.
Hence, we would see additional mining operations encouraged.
From a macro view, greater resource supply will benefit industries that consume the resources–automotive, etc–but could result in an overall reduction in mineral prices.
— — — — — — — — — — — — — — — — — — — — — — — — — — — — — —
Mining AI startups are getting acquired.
Recoup your exploration costs by selling your data like Alaska Energy Metals corp did.
The White House announces efforts to assess critical minerals using AI. Projects include a DARPA funded hackathon in February, a new supply tracking website to launch in January, and partnerships with other countries to build out mineral supply chains.
More validation of the talent shortage in mining.
Kobold metals drills in Zambia.
A mining veteran interviewed about using AI to explore (see below).
🤔 Opinion
The context for this newsletter is the AI revolution and the greater shift in our economy towards physical resources. Simply put, AI is making professional services faster and cheaper at scale.
Software development. Legal advice. Handling emails. Even complex cybersecurity. It’s all up for grabs. If you don’t believe me, google some articles about how people have utilized this technology for each one of those uses. I know first-hand.
Something similar happened in textiles in the late 1800s with the advent of the sewing machine. It allowed incredible productivity gains so that the cost of goods could fall and more people could afford products. It’s hard to imagine a time when clothes cost several months worth of wages, but they did.
Today we are shocked at the high cost of labor and professional services like AI development and lawyers. But I believe these will be the same as textiles. When everyone can afford a top-notch lawyer, software designer, or business coach, living standards as a whole will rise and industry will accelerate.
The same applies to mining.
Today we’re faced with the challenges of mineral, labor, and talent shortages–all of which are easier to address with AI than without it. Those companies that engage the AI developer community to create their own unique solutions–from exploration to excavation–will achieve exponential gains compared to others who wait and scale linearly with new hires.
🛠️ Tools and Data
Satlas SuperResolution: https://github.com/allenai/satlas-super-resolution AI-generated geospatial data that is highly accurate, available globally, and updated on a frequent (monthly) basis. Includes globally generated Super-Resolution imagery for 2023. | ![]() |
Segment Lidar: https://github.com/Yarroudh/segment-lidar Segment LiDAR data using Segment-Anything Model (SAM) from Meta AI. | ![]() |
Python bindings for Apache Arrow for geospatial data. | ![]() |
Large-scale geospatial data visual analysis engine based on WebGL launched by the AntV data visualization team of Ant Group. | ![]() |
Thanks for reading! Want me to look into a particular topic? Email your suggestions and I will dig.