The majority of the world’s cobalt – a crucial metal for the lithium-ion batteries powering everything from cellular telephones to cars – is mined in exploitative conditions in the Democratic Republic of Congo, often by children. FreightWaves reported on the difficulty of sourcing ethical cobalt a year ago. Last month, FreightWaves covered a blockchain-based project led by Ford Motor Company, Huayou Cobalt, IBM, RCS Global and LG Chem to bring transparency to the cobalt supply chain and assure consumers that the metals in their electronics was ethically sourced.
Now another solution to the trouble with cobalt may be emerging. This morning, Quartz reported that venture capital firm Andreessen Horowitz (a16z) and Bill Gates’ Breakthrough Energy Ventures have invested in KoBold Metals, a startup that will use machine learning algorithms and artificial intelligence on large sets of geological and chemical data to hunt for new sources of cobalt.
The financial terms of the deal were not disclosed, but KoBold officials told Bloomberg that the raise provided the company with “well over four years of capital.”
KoBold’s co-founder and CEO, Kurt House, has a PhD in Earth & Planetary Science from Harvard University and serves as an adjunct professor at Stanford. Previously, House founded C12 Energy, a private equity-backed oil and gas company that redeveloped mature oil fields by applying new, high-tech recovery techniques involving CO2 gas injection. House launched KoBold in 2018 with physicist Josh Goldman, formerly of McKinsey & Company, and Jeff Jurinak, ConocoPhillips’ former Chief Reservoir Engineer.
Currently machine learning algorithms and big data techniques are mostly used in mining and drilling in the pursuit of operational efficiencies, not exploration.
“Exploration is a largely manual practice in which geologists visually scan spatial data looking for key features, guided chiefly by analogies, judgement and intuition,” House wrote in a company blog post. “The historic practice of manually searching disparate and unstructured datasets means that contemporary exploration efforts are not systematic, not comprehensive and not fast.”
Connie Chan, a general partner at a16z, characterized KoBold’s product as “Google Maps for the earth’s crust and below.”
“By building a digital prospecting engine – full stack, from scratch – using computer vision, machine learning and sophisticated data analysis not currently available to the industry, KoBold’s software combines previously unavailable, dark data with conventional geochemical, geophysical and geological data to identify prospects in models that can only get better over time, as with other data network effects,” Chan wrote.
Volatile demand for consumer electronics and uncertainty about the pace of electric vehicle adoption caused a price shock in cobalt prices in 2017 and 2018 that has only just now unwound. The chart below shows how cobalt futures contracts on the London Metal Exchange tripled in value and then collapsed over the course of two years.
Cobalt isn’t exactly a rare metal, it’s just that in years past, copper mining produced enough cobalt as a byproduct to satisfy global demand, and cobalt ore deposits were never explored and developed on their own.
“Over 98 percent of the 110 kilotons of globally produced cobalt a couple years ago was a byproduct of copper and nickel mining,” Chan wrote. “That byproduct supply was sufficient to satisfy historical global demand. But as the production of electric vehicle batteries rapidly accelerates, refining existing sources will not be enough.”
Bloomberg expects demand for cobalt to outstrip supply beginning in 2023, with the shortage increasing in severity through at least 2025.
KoBold’s goal is to find plentiful, low-cost supplies of cobalt, take on additional investment on a deposit-by-deposit basis, secure rights to the supply, and then turn it over to miners for development.
“We don’t intend to be a mine operator ever,” House told Bloomberg.