Copper Arc: Mingomba, Zambia
A century of geologists, a century of data, and the machine that read it all at once
A note on this story, its versions, and the limits of AI-assisted research
This is the fourth and most complete version of the Mingomba story.
The first version was published on LinkedIn on 14 May 2026 as P40 in the #MiningIsHuman series. It opened with an atmospheric description of the Central African Plateau in the dry season and focused on KoBold Metals’ use of machine learning to prioritise a drilling target from a century of accumulated exploration records.
That version contained two material errors: a title that implied the deposit had been found by an algorithm, and an account of the ownership history that omitted the work done by the Lubambe geological team between 2019 and 2021. It also incorrectly referenced Equinox Minerals as a former holder of the ground.
The errors exposed a limitation of AI as a research tool in cases like this. Mingomba’s exploration history is dispersed across records held under different project and company names and in different collections, including state-owned archives. AI helped me surface a first, challengeable account, but it did not reconstruct the full history. That required the knowledge of people directly involved, followed by conventional research and fact-checking.
Geologists Jon Stacey and Tim Williams, and academic Simon Jowitt, commented on the original post with corrections and important additional context. Jon Stacey, who led the 2019 to 2021 drilling programme alongside Elijah Mwelwa, provided a direct account of what that work produced: four holes beyond 1,500 metres, approximately ten million tonnes added per hole, a JORC-compliant resource declaration of 247 million tonnes at 3.64% copper, and a preliminary economic assessment that brought KoBold to the table. Tim Williams supplied the correct ownership lineage from the BSAC mineral rights through Anglovaal, ARM, TEAL, the ARM-Vale joint venture, Lubambe Copper Mine, and EMR Capital to KoBold.
The second version was published on LinkedIn on 20 May 2026. LinkedIn’s 3,000-character limit required the atmospheric opening to give way to the corrected text. That version remains live on LinkedIn and is linked below.
The third version was published here on Substack in June 2026 and incorporated further information surfaced from comments on the LinkedIn post. It carries the corrected ownership history; names Jon Stacey and Elijah Mwelwa and records their contribution; and incorporates the findings of a full fact-check, including a further correction to the JORC resource grade from 3.2% to 3.64%, the addition of Jeff Bezos to the investor list, and a correction to the Lubambe operating depth. With no character constraint, it also restores the atmospheric opening from the first version.
This fourth version is the result of a status and fact check in July 2026.
My thanks to Jon Stacey, Tim Williams, and Simon Jowitt for their contributions to the public record on this story.
The LinkedIn version was updated on 20 May 2026.
This work is AI-assisted—read about how I work.
It is May on the Central African Plateau. The wet season ended three weeks ago and the grass is curing at the tips, still green at the root, straw-coloured above. The sky at 1,300 metres has a quality you don’t find lower down: hard blue, high, the shadows very dark and the light between them very bright. The miombo opens and closes along the road from Chingola. Brachystegia at ten to fifteen metres, canopy open, long grass underneath. Not forest. Parkland. The Ming’omba stream threads through a dambo to the north, dark and slow, lined with reeds. It gives nothing away about what lies below.
Mingomba is in Chililabombwe District, Copperbelt Province, Zambia. Close to the DRC border.
The Lamba have mined copper from this earth since the seventeenth century. They found it where it came to surface. Mingomba never did.
The ground has passed through a longer chain of hands than any single company’s story would suggest. The Konkola North property carried an old shaft, Konkola No. 2, sunk by Anglo American in the 1960s and flooded soon after when the ore body proved inconsistent — an early false start on ground that would take another sixty years to give up its metal.
Anglovaal took up the property next, through its Zambian subsidiary Konnoco, holding Konkola North from 1996 and running a significant drilling programme through the late 1990s. Those holes extended into what would eventually become Mingomba and continued downdip into what was then called Block A, drilled as far as the Lubengele stream, the property’s northern boundary with Konkola Copper Mines. African Rainbow Minerals absorbed the position when it acquired Anglovaal’s mining assets, and Konnoco’s interests passed into TEAL Exploration and Mining Corporation, which listed on the Toronto Stock Exchange in November 2005 with ARM holding approximately 66%. ARM and Vale then bought out the minority shareholders and took TEAL private, delisting it in 2009 as the two companies established a 50:50 joint venture to develop the Lubambe mine, working the East and South limbs of the Konkola Dome from surface down to approximately 550 metres. EMR Capital acquired the ARM and Vale stakes in 2017 for $97 million. Each owner left data behind. None sank a shaft into the deep extension.
Between 2019 and 2021, geologists Jon Stacey and Elijah Mwelwa took up that unfinished work, drilling the Lubambe Extension ground on a constrained budget. Four holes beyond 1,500 metres, each adding roughly ten million tonnes to what was already known — fifty million tonnes in total. Their work produced a JORC-compliant resource and a preliminary economic assessment substantial enough, in Stacey’s own words, to bring KoBold and other bidders to the table. They honoured everything the previous generations had left in the archive, and drilled deeper than any of them had gone.
In 2022, KoBold Metals made a $150 million investment and renamed the project Mingomba. KoBold is a Berkeley, California-based machine-learning company backed by investors including Jeff Bezos, Bill Gates, Sam Altman, Andreessen Horowitz, and BHP Ventures. Their TerraShed platform normalised a century of exploration records, spatially aligned them to common coordinates, and built a three-dimensional model of the subsurface from more data than any human brain can hold at once. Not new data. All of the historical data, read simultaneously for the first time. They drilled against model uncertainty rather than on a grid.
KoBold’s internal estimate: 247 million tonnes at 3.64% copper, on par with Kamoa-Kakula in the DRC in grade and class. This figure is not yet JORC or NI 43-101 compliant. No independent Qualified Person or Competent Person report has been published. The New York Times reviewed KoBold’s projections and a separate independent assessment, both confirming the scale.
Copper above $13,000 a tonne. Powered by the energy transition. KoBold broke ground on 29 April 2026. Shaft sinking: early 2027. Capex: $2.3 to $2.5 billion. First production: early 2030s. Three hundred thousand tonnes per annum. ZCCM-IH holds a 20% stake in Mingomba Mining Limited alongside KoBold, with Zambia’s state investment vehicle actively working to raise that position to 25%.
The shaft is not being built alone. The Zambia spur of the Lobito Corridor, roughly 500 kilometres of greenfield railway terminating at Chingola, is currently in contractor evaluation, with nine international EPC contractors having submitted bids in May 2026 and construction expected to begin by late 2026 or early 2027, completion targeted for the early 2030s. Mingomba sits nineteen kilometres from that terminus, near Chililabombwe. The timing is not coincidental. In December 2024, at a summit co-hosted by President Biden and Angola’s President Lourenço, the Africa Finance Corporation announced an MOU naming KoBold as anchor client for the entire Zambia-Lobito railway, committing to a minimum of 300,000 tonnes of copper and related freight a year — Mingomba’s full planned output, formally pledged as the commercial foundation of the rail line. The shaft and the railway are being built because the other one exists. Both timed to open in the early 2030s.
Sinking a 1,700-metre shaft and holding a $2.5 billion project together is another discipline entirely. The Copperbelt has a long memory for companies that arrived confident and left quietly. The work still lies ahead.
Set against the wider basin, the number is smaller than it sounds. In 1979, Zambia and Chile produced roughly the same volume of copper: around 680,000 tonnes each. By 2025, Chile produced 5.3 million tonnes to Zambia’s 890,000. The gap isn’t geological — Chilean porphyry ore runs at 0.5 to 1.0% copper, well below Mingomba’s grade. The more instructive comparator is the DRC, sharing the same basin and the same geology: around a million tonnes of production in 2010, overtaking Peru as the world’s second-largest copper producer by 2023, and more than 3 million tonnes in 2024. That is what a changed investment climate does to a Copperbelt deposit. Zambia’s own target of 3 million tonnes by 2031 is less an ambition than a return to a scale the country once held. The USGS estimates 152 million tonnes of known copper resource across more than eighty deposits in the Central African Copperbelt, with a further 168 million tonnes estimated undiscovered. Most of it is still in the ground.
In September 2025, Zambia launched the GSD Digital Store, a searchable repository of exploration records from the 1920s to the present — the same century of records Anglo, Konnoco, ARM, TEAL, Vale, and EMR each added to and passed on without ever reading it whole. The mineral rights over this ground trace back further still, to the BSAC Royal Charter of the 1890s. The BSAC gave up political control of Northern Rhodesia in 1923 but retained those mineral rights until Zambian independence in 1964. Every owner in the chain left records. KoBold built a tool that could hold and read all of them at once.
The Lamba found copper where it broke surface. Every generation that followed drilled deeper. Jon Stacey and Elijah Mwelwa drilled the deepest, on the tightest budget, and built the foundation the deal was built on.
The Ming’omba stream will still be there when the shaft goes down. Dark and slow and entirely indifferent. The ground gave nothing away for a century. What changed was not the ground and not the ore. What changed was the instrument.