Our society, and the extraordinary technological progress we’ve seen over the past century, is underpinned by materials that we dig out of the ground.
Without copper, we wouldn’t have electronics; without silicon, no computing; without iron, no transportation or machinery. Lithium—which underpins electric vehicles, battery storage, renewable energy, and numerous other technologies inextricably linked to decarbonisation—is predicted to quadruple in demand by 2030.
This demand underlines the pivotal role that mining plays in global industry—but also its inherent inefficiencies, and the knock-on impact on cost, labour, and carbon emissions.
Investing in early stage companies for Founders Factory alongside mining multinational Rio Tinto, I’m seeing attention shifting to alternatives that will not only allow the industry to cater to this growing demand through upgrades in efficiency and reduced-impact methods. This has sharply come into focus in recent weeks, as countries will start to look to unshackle themselves from global supply chains and on-shore their own mineral supply, all while complying with local regulation.
Across industries, AI enables resourcefulness and efficiency—and this is no different in mining. Embedding it into existing mining infrastructure, alongside a slew of other new breakthrough technologies, could be critical to drastically reducing mining’s environmental footprint and securing the industry’s long term future.
Mining: An industry riddled with inefficiency
Excessive focus on short-term profitability has left mining encumbered with decades-old processes. There are considerable opportunities for efficiency upgrades—in terms of cost, labour, and emissions—across the supply chain.
This starts with exploration. Between 2011 and 2020, around 15,000 exploration projects for new materials resulted in just 45 viable discoveries, an astonishing 0.3% success rate.
What does this ‘exploration’ actually look like? Modelling is often the start point, to try to map out and visualise deposits: but these old techniques struggle to predict accuracy, leading to costly delays. Drill holes are made (usually over 100 per project) to obtain samples, assess quality, and ultimately, to start the excavation process.
The mining lifecycle leaves an immense physical impact, ranging from core drilling into rock samples for testing, to blast hole drilling to place explosives for rock fragmentation and excavation. Certain types of mining are more damaging: lithium, for instance, is largely found in hard-rock, which requires more water and energy intensive extraction techniques.
That’s not to mention the waste produced as a result. Around 72 billion tonnes of mined material each year produces just over 10 billion tonnes of ore. The remains, known as tailings, are left in storage—despite containing an estimated $3.4 trillion in value globally.
An existential threat looms over the industry in the coming decades—nearly 50% of the workforce is set to retire within the next ten years, while recruiting new engineers is getting harder. This leaves a huge gap between demand for materials and the mining workforce’s ability to meet it.
The big picture isn’t pretty. Outdated processes are discovering materials at an incredibly low rate, leaving a vast environmental footprint and unthinkable amounts of waste in their wake, with a fast diminishing workforce. Transformation requires huge upgrades in efficiencies and rapid advances in data insights across the mining chain. A promise AI can fulfill?
AI can transform mining
AI’s capabilities start at the very top of the funnel—exploration—and hits each point of the mining process, from analysis through to extraction.
Modelling, a key part of exploration, is seeing a facelift, as a band of new companies approach how we are predicting deposits and accurately assessing their quality. KoBold Metals combines a wealth of geoscience data with powerful data aggregation tools and proprietary AI to great success. Last year, they captured attention when their technology uncovered the largest copper deposit in over a century.
We’re excited to be working with Terra AI, founded by KoBold’s former AI Lead John Mern. Their model’s USP is constantly evaluating the ‘next best’ prediction based on each new data point: an approach that promises 40% reduction in drilling.
Durin, led by mining and geology expert Ted Feldmann, are building automated drill rigs that consolidate drilling, lab testing, drill-hole planning, and data analysis into a single system. This aims to halve exploration timelines (from 10 years to 4.5 years), vastly minimising unnecessary drilling.
Extraction is another area where AI promises considerable efficiency gains. One startup I’ve met is modelling blast fragmentation during the excavation process, aiming to reduce the amount of energy required as well as the physical impact on the land.
These technologies hone in on reduction. So can AI totally remove mining’s physical impact?
Zero-impact mines are currently a flight of imagination. But certain companies are looking at decoupling mining from the material supply chain altogether by asking—can we use AI to replicate properties of rare minerals and produce them synthetically?
Matnex is pioneering this, combining powerful AI and quantum to rapidly accelerate the discovery of new materials. Last year, they developed the world’s first rare earth-free magnet (for EVs, wind turbines, drones, etc) produced at 20% of the material cost and a 70% reduction in material carbon emissions.
Looking for mining’s breakthrough
Our existing mining processes are clearly not fit for purpose—not least if we’re ambitious about meeting demand for critical minerals that will power society’s transformation and decarbonisation over the coming decades. Just as with other industries, we look to ‘breakthroughs’ for a solution—ideas that turn from impossible to implausible to promising as technological capabilities develop.
AI clearly presents this opportunity for the mining industry. It’s a clear path to helping find resources faster, make more with less, and reduce our reliance (perhaps altogether) on physical supply chains.
