Pattern Counsel is written for General Counsel, Chief Legal Officers and their teams — and for the law firms, legal operations professionals and technology providers who work alongside them. This work is AI-assisted — read about how I work.

Pattern Counsel reflects my vision of how in-house legal functions could work. Many General Counsel will see it differently.

Introduction

Being inside is not the same as learning from inside.

The first instalment of Pattern Counsel argued that legal value compounds through embeddedness. Being inside the enterprise produces a different kind of knowledge, and each cycle of presence can sharpen the next judgment.

The second looked more closely at what “inside” means. Position matters: where legal sits, who it reports to, which rooms it can enter. But position does not guarantee presence across the enterprise’s work. Processual embeddedness is something different: legal being present in the phases where the enterprise’s work creates, manifests or exposes legal risk.

There is another consequence of that distinction. Processual embeddedness does not simply put legal in the work early enough to influence it. It puts the lawyer inside an experience from which she can learn — seeing choices being made, living with what happens to them and remaining present when the consequences become clear.

That changes the lawyer.

Changing the lawyer is not the purpose of embeddedness. The purpose is what the enterprise can do differently because legal is there.

What do you know?

A month before I became Head of Group Legal South Africa for a multinational mining company in 2010, I met one of the general counsel whose world I was about to enter. He had been in the system for years. I had not.

He looked at me across the table. “What do you know?”

It was a formidable question because he knew an enormous amount. Mining law. Regulatory history. The institution. The matters that had shaped it. I knew very little of that. For a while I tried to catch up.

It was miserable.

There was always another piece of legislation, another regulatory turn, another fact buried in years of correspondence or litigation. What I understood only later was that there were different ways of being inside. That veteran of Group Legal was profoundly positionally embedded. Years inside had given him knowledge I could not reproduce by reading harder.

But position is not the same thing as being embedded in the processes through which the enterprise does its work.

My mistake was thinking the answer to “What do you know?” was simply to know more law.

The more consequential opportunity was to get further inside the work.

Learning from the process

A lawyer can know almost everything about a matter without having experienced the process that produced it. An external lawyer can reconstruct a transaction from the papers. A litigator can reconstruct a dispute. A regulatory specialist can master the history of a permit. All may know more law than the lawyer inside.

Processual embeddedness offers something different.

At Plan, you experience why choices are made while alternatives are still open. At Do, you watch those choices encounter operations, economics, regulators and people. At Study, you see which assumptions held and which consequences nobody predicted. At Act, you see what the enterprise decides to carry into the next cycle. Then you arrive at the next Plan.

The cycle belongs to the enterprise, not legal. Legal rarely runs it. The value lies in being sufficiently present to read it from inside: to see how different disciplines interact, where risk actually emerges and what becomes significant when a choice encounters reality.

That last part matters. The record can tell you that an option was abandoned. Participation may tell you why an apparently minor operational constraint suddenly made it impossible. The papers can record what a regulator said. Being there may teach you why one comment changed the room while another was ignored. A risk register can tell you what people thought might happen. Staying through the work tells you what actually constrained the decision.

That is more than additional information. It is learning what acquired significance. And because you were accountable alongside others for some part of the outcome, the consequence is not merely something you observe. You have to live with what the judgment did.

The first instalment called the result “diagnostic precision”: the lawyer who had lived successive permitting cycles knew which questions to ask because she had been there when the earlier answers formed. She was accountable alongside engineers, hydrologists and the community team for whether the permit landed or failed.

The record tells you what happened. Participation begins to change what you notice next.

What you begin to notice

Over time I developed a habit of looking backwards before answering forwards. What happened before this? Why are we here? Who has seen another part of it? What apparently unrelated work touches this? What happens if we pull that loose strand?

Earlier experience began doing something more useful than supplying answers.

Sometimes it supplied a question.

That is a small distinction, but I think an important one. Processual experience creates a reference set. You have seen choices become consequences. You have seen legal, operational, financial, human and regulatory considerations interact rather than arrive as separate inputs. You have learned something about what matters here, rather than what ought to matter in theory. That begins to change how the next problem looks.

It can also mislead you.

I learned that on work in Latin America. A community challenge around a water-use licence looked familiar through my South African experience. Community litigation. Organised opposition. Mining. A colleague asked a simple question that made me look again.

The mechanism I had read as opposition was, in that setting, part of an ordinary way in which a licensing decision could be challenged or reconsidered. My experience had allowed me to recognise a pattern quickly. It had also allowed me to recognise the wrong one.

That is an important limit to the embeddedness claim.

Experience does not automatically become judgment. Sometimes it makes you wrong faster.

The reference set has to remain open to context.

Learning safety from pilots

I saw the more valuable possibility in a lawyer we deployed into the centre of our mine safety management system. That description matters. He remained a lawyer. He did not run the safety system. But he was accountable from within it, working alongside people whose enterprise objective was to prevent people being injured or killed.

He was no longer encountering safety principally through legislation, investigations and the consequences of incidents. And the system began teaching him.

One of the things he came to see was the strategic importance of learning itself. A serious incident at one operation contained knowledge that might prevent another somewhere else. A safety system therefore needed not merely to investigate incidents, but to allow learning from them to travel.

That created a tension with something legal was doing for perfectly defensible reasons: privilege. There are good reasons to protect investigations and related material when facts are uncertain, consequences serious and legal exposure real. But he was now looking at privilege from inside the safety system. If protection became so complete that one mine could not learn from another, what did that do to the enterprise’s ability to become safer?

The law had not changed. The lawyer had. And that changed the question.

It was no longer simply: what should we protect? It became: how do we preserve the protection genuinely needed while allowing enough learning to travel to help prevent the next injury or fatality?

Then he made another connection. Mining cannot be the only activity in which people make difficult decisions inside complex systems where mistakes can kill. What might mine managers learn from airline pilots? What might both learn from heart surgeons? What had those disciplines discovered about near misses, reporting failure, designing checks and learning without making every admission an allocation of blame?

That insight was his. We opened the door. I am not sure we walked far enough through it. But the sequence matters more than whether we completed the journey.

He went inside a process. The process taught him something. What he learned changed how he saw a legitimate legal practice. And that allowed a lawyer to identify a strategic possibility for the enterprise.

Before the legal question

That is where the change in the lawyer starts to matter to the business.

In-house lawyers often talk about getting involved earlier. But, as the last instalment put it, earlier is a direction, not a diagnosis. A lawyer can arrive earlier and still wait for somebody to ask a legal question.

Processual embeddedness offers something more. The lawyer begins to understand the enterprise problem before it has been reduced to its legal surface.

At its most reactive, legal asks: How do we manage the legal consequences of this decision?

Earlier legal involvement asks: What are the legal risks in the alternatives being considered?

But an embedded lawyer can sometimes ask: Does what we know change the alternatives the enterprise should be considering at all?

That is a different contribution.

Risk changes with it. Once a strategic choice has hardened, risk can be mitigated, allocated, insured, disclosed, litigated or accepted. While the choice can still move, risk can help shape the choice itself.

And this is where the fourth pattern from the first instalment becomes important. If legal is an enterprise resource, its contribution cannot ultimately be measured by the quality or quantity of legal work it produces. The first instalment put the test elsewhere: what the enterprise achieves where legal is embedded.

For the safety lawyer, the important output was not a better privilege memorandum. The enterprise outcome at stake was a safer system. For a lawyer participating in strategy, the measure is not how many legal risks she identifies. It is whether the enterprise makes a better choice because legal judgment was part of making it.

That is ultimately what embeddedness is for. Not changing the lawyer. Changing what the enterprise can do because the lawyer has been changed by being inside its work.

Legal judgment is no longer applied only to strategy after it has been formed. It becomes one of the disciplines from which strategy is formed.

A miner with a law degree

Just before I retired from corporate life, in our final session together, one of my country lead lawyers was preparing for the next stage of her development. She was an exceptional lawyer who had come into mining from an unrelated, highly regulated industry. I asked where she most needed to grow: as a lawyer, as the leader of her team, or as a member of the executive committee?

Mining law worried her. But underneath that was a different anxiety. Not knowing enough about mining made her less confident about what she could contribute when the executive committee was discussing the business rather than the law.

I knew that feeling. My advice was not to start with more mining law.

Study copper mining.

Know the important mines and how they compare. Understand enough geology to know why the orebody matters. Learn about water, communities, logistics, labour, environment and economics. Understand why one mine performs differently from another, what constrains it and what gives it choices. And know as much as you reasonably can about the legal and permitting position of those mines.

The geologist should know more geology. The finance director should know more finance. The specialist external lawyer may know more about a particular field of law. The point is not to beat any of them at their disciplines. It is to understand enough of the enterprise for your discipline to become useful inside the formation of its choices.

My shorthand became simple.

You are a miner with a law degree.

In another enterprise, change the noun. Keep the discipline.

That last part matters. Embeddedness is not an invitation to dissolve into the business. The last instalment argued for calibrated presence, not maximum presence. Legal still needs enough distance to exercise independent judgment and enough professional depth to bring something distinctive to the conversation.

A miner without the law degree is not the objective. Nor is a lawyer standing at the surface of mining. The value lies in the combination.

In another enterprise, change the noun. A banker with a law degree. An investment manager with a law degree. Keep the discipline.

I should add one thing. When I returned to Cape Town, I started my LinkedIn series #MiningIsHuman for much the same reason. If part of my post-corporate life was going to involve working in and around mining, I wanted to keep learning about the industry, not simply its law.

And then there is AI

Which brings us back to the pressure with which this series began.

The first instalment deliberately rejected legal expertise as the defensible asset. Expertise is available outside, and AI can already hold more legal information than an in-house team.

AI will increasingly read the contracts, permits, board papers and investigations. It will retrieve history, compare positions, draft, analyse and find connections across a record at a scale no individual lawyer can match.

Good. Use it.

I would not build an in-house career around a list of things we hope machines will never do. That list will keep moving. Nor would I claim that participation creates some mysterious category of knowledge that machines can never reach. Much of what lawyers learn through experience can be recorded, reconstructed or inferred, and technology will become better at doing all three.

The more useful distinction is what participation does to judgment. It exposes the lawyer to consequence, interaction and context. It teaches what acquired significance when choices met reality. Repeated often enough, it creates a reference set against which the next problem is read — imperfectly, as my Latin American mistake shows. And because the lawyer remains inside the enterprise’s work, that changed judgment can be brought back into the next choice.

AI offers no immunity from that challenge. It sharpens it.

The more technology can do at the legal surface, the harder it becomes to justify spending scarce embedded human capacity there.

Use the machine at the surface. Use presence where participation can change what you learn and what the enterprise does.

What being inside is for

So perhaps I finally have a better answer to the Veteran.

“What do you know?”

Knowledge matters. His certainly did. But positional knowledge was not the whole opportunity I had been given by being inside. The larger opportunity was to participate deeply enough in the enterprise’s work for the enterprise itself to become part of my education.

To see choices become consequences. To learn what became significant and why. To see disciplines collide around the same problem. To carry one experience into another without assuming they were the same. To allow what I learned to change the question before reaching for the answer. And then to bring the discipline of law back into the next enterprise choice.

That is the loop.

The enterprise teaches the lawyer. The lawyer brings what she has learned back into the enterprise. Sometimes that changes what the enterprise does next.

That is the point.

And if AI continues to make the legal surface easier and cheaper to service, the question for an in-house lawyer becomes less comfortable, but perhaps more useful:

Are you merely inside the organisation, or are you sufficiently inside its work for it to be changing what you can contribute?

Because legal is an enterprise resource only to the extent that being inside ultimately changes the enterprise outcome.

A note on sources

This instalment is more experiential than the first two. Its central argument comes principally from things I saw while working inside legal functions: how participation in enterprise processes changed what lawyers noticed, what they learned from consequence, and what they were subsequently able to contribute.

There is a substantial literature around experiential and situated learning, tacit knowledge and professional judgment. I have deliberately not used it to build the argument here. Pattern Counsel started with practice, and this instalment stays particularly close to it.

The examples are therefore offered for what they are: experience from which I have tried to identify a pattern, rather than evidence that the pattern is universally true.

Continue the discussion

If this reading matches your own experience, or if it points somewhere different, I would value the exchange.

Kevin Lester is an independent counsel and thinking partner. He spent fifteen years as a Country Head of Legal and subsequently as General Counsel at Anglo American. He advises General Counsel, senior executives, and boards at kevinlester.co.za.

Subscribe to Kevin Lester

Please subscribe. There’s no cost. When you subscribe, you get free access to essays, stories and advisory notes on in-house legal, mining and regulated sectors, and the decisions that shape organisations under pressure. Always free.

This is the Substack for Kevin Lester — independent advisor at kevinlester.co.za and the writer and host behind #PatternCounsel, #MiningIsHuman, #MiningOurHistory, and #BelowTheSurface.

It carries four publications.

#PatternCounsel is a monthly essay series on the management and delivery of in-house legal — the patterns that emerge across cycles, roles, and pressures. Field notes from fifteen years inside, for General Counsel, CLOs, their teams, and the firms and providers who work with them. Published first Wednesday of the month.

#MiningIsHuman is a series of stories about mining in southern and central Africa — its geology, capital, labour, infrastructure, decisions and consequences across time. The stories are published in arcs. Each arc follows a commodity, a region, or a theme, from the first concession to the current owner, mine by mine. Arcs can overlap: a mine that appears in the Northern Cape arc may also belong to the copper arc. The same ground, read differently each time.

#MiningOurHistory tells the institutional stories of Southern African business — the versions institutions have told about themselves, the versions the archive tells, and the versions that emerge when institutions are called to account. Same records. Different eyes.

#BelowTheSurface is a collection of materials from an independent advisory practice — notes, essays, presentations, and a podcast recorded with various collaborators and produced by Solid Gold Podcasts. Written and recorded for General Counsel, senior executives, and Chairs in mining and other regulated sectors. Listen on Apple Podcasts or iono.fm.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

Copper was recorded in the Lumwana East River in Zambia’s North-Western Province in the 1930s. It took another 70 years to build a mine.

Roan Selection Trust picked up the ground in the 1950s. American Metal Climax followed, forming a joint venture with Anglo American in 1970. After Zambia nationalised copper in 1975, the project passed to the state. Another JV drilled it through the 1980s. Phelps Dodge held the licence in the early 1990s.

None built the mine.

Long before them, copper was part of the region’s history. Kaonde and Lunda communities trace their roots to the Luba-Lunda Empire. At Kansanshi, 100km away, Kaonde ancestors worked copper from around 400 AD, carrying smelted metal another 100km to Kipushi. When a British prospecting expedition arrived in 1899, a local chief showed them the ancient workings. The workings had been abandoned, but the memory remained.

In 1999, Equinox Minerals earned into Lumwana, buying Phelps Dodge out for $5m.

It spent $760m building the mine.

Production began in December 2008, months late after a plant fire.

Barrick Mining Corporation acquired Equinox in 2011 for C$7.3bn. Then came another test.

In 2014, Zambia raised the open-pit royalty from 6% to 20%. Barrick prepared to suspend Lumwana, putting roughly 4,000 jobs at risk.

Around 2,000 workers struck for a day. President Lungu ordered the rate reviewed. Cabinet cut it to 9% from July 2015, and Lumwana kept running. The Chamber of Mines estimated the original hike could cost the industry 12,000 jobs and 150,000 tonnes of national output.

By 2019, Barrick was considering selling Lumwana. It was high-cost and underperforming. Standard Chartered ran the process; China Minmetals, Jiangxi Copper and Zijin Mining were reported as interested.

No deal happened.

Barrick’s CEO later said he wasn’t in a hurry to sell an asset he believed was undervalued at its $735m book value.

So Barrick did something else.

It doubled down.

The Super Pit Expansion, launched in 2024, is a nearly $2bn investment to double throughput and target 240,000 tonnes of copper annually. First ore is due in early 2028. In Q1 2026, Barrick said the project remained on schedule, while Lumwana helped lift group copper production 11%.

Around the mine, a village of fewer than 1,000 people became a town of roughly 5,000. Today, 98% of Lumwana’s workforce is Zambian. Barrick says that since 2019 the mine has contributed more than $4bn to Zambia through taxes, royalties, wages and procurement.

For decades, companies drilled Lumwana, owned it, transferred it and moved on.

Now Barrick is spending nearly $2bn to make it much bigger.

Sometimes the difference between a prospect and a world-class mine isn’t merely what lies underground.

It’s also who is prepared to stay — and keep investing.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

South African gold spent a century going down.

At Mponeng, the deepest mine on earth, miners work nearly 4km beneath the West Wits line. For generations, that was the story of South African gold: deeper, hotter, harder, because the richest ore was down there.

Then the new mines stopped coming.

Modder East opened in 2009. Burnstone followed in 2010. Then silence. It became easy to believe South Africa's great gold story was reaching its end.

And then came Qala Shallows.

South Africa's newest gold mine reaches only about 800m below surface. It sits on the Central Rand, working reef that a century of deep mining largely stepped over. The gold was always there. There was simply richer ground somewhere below it.

And this is what makes Qala so exciting: it is opening a new mine on the goldfield where the whole extraordinary story began.

The 1886 gold rush started here, on the ridge that built Johannesburg. Miners chased the richest reef down and away, while shallower ground was left behind. Durban Roodepoort Deep — from which DRDGOLD Limited later evolved — worked the broader Central Rand leases until 2000, before closing as the economics turned against them.

But the gold did not disappear. Through the closed years, artisanal miners continued working the reef the formal industry had left behind.

Now the economics have changed again.

On 17 March 2026,West Wits Mining poured Qala Shallows' first gold at Sibanye-Stillwater 's Ezulwini plant.

And what a proposition this mine is.

At 800m, there is no 4km pumping bill and none of the extreme heat of the deepest mines. Qala expects to produce around 70,000oz a year over a 17-year mine life, with West Wits' 2025 feasibility study putting all-in costs at US$1,289/oz.

The remarkable new opportunity in South African gold turned out to be ground everyone had passed over.

And Qala is bringing more than ore back to work. By June, around 450 people were working on site in a part of the Rand that had watched mining leave. West Wits owns 74% and empowerment partner Lilitha Resources 26%. Qala mines the ore; Sibanye-Stillwater processes it.

A new mine. New jobs. Local procurement. Gold flowing again from ground written off years ago.

And there is one final, wonderful detail.

Qala means "to begin."

Nobody agrees on the exact day George Harrison found gold on Langlaagte. But we know the year was 1886.

In the 140th year since the Witwatersrand gold rush began, a new mine on that goldfield poured its first gold.

The place where South African mining began has begun again.

After 140 years, the Witwatersrand is not finished. It is not even half empty.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

Zambia's Fifth Mining Law in Fifty Years (Mining Is Human, Story 62)

Zambia has rewritten its principal mining law roughly once a decade since 1972. Each rewrite reflects the mining challenge of its time.

In 1995, the priority was privatisation. In 2008, it was unwinding the Development Agreements privatisation had produced. In 2015, it was restoring stability after years of fiscal and licensing volatility.

Beneath the rewrites lies a consistent pattern: incremental institutional reform, with a persistent commitment to widening the societal benefit of mining.

The fifth rewrite advances that pattern while addressing a different challenge.

I spent my working life around mining law, but Zambia’s statute book was not my jurisdiction. Here is what I noticed from the public record.

Zambia wants 3Mt of copper annually by 2031, up from roughly 800,000t. Ambition on that scale requires something less glamorous, but no less important, than another mine: a regulatory system capable of administering it.

That, I think, is the opportunity the Minerals Regulation Commission Act, No. 14 of 2024 is designed to seize.

Much of the law is familiar. The Tribunal, royalty structure, environmental fund and citizen-empowerment orientation existed under the 2015 Act—continuities refined rather than invented.

The new element is institutional. The MRCA replaces fragmented ministerial directorates with an independent Commission empowered to grant, suspend and revoke rights, monitor operations and rehabilitation budgets, and advise on policy.

Three provisions reveal the settlement being pursued.

Section 65 extends liability for environmental harm beyond the licence holder to any person who directly contributed to the act or omission causing it. Corporate responsibility no longer necessarily ends with the corporation.

Section 87 caps mining rights at five per person or entity, subject to compliance and demonstrated financial capacity. While seeking to triple copper output, Zambia is also limiting concentration of the rights required to produce it.

SI No. 68 of 2025 reserves 20% of core-mining procurement for local companies in 2026, rising to 40% within five years. Non-core services, including catering, security and haulage, are reserved for Zambian companies.

On paper, the design is coherent. The opportunity now is to translate that design into institutional capacity.

The Tribunal existed under the 2015 Act but was never constituted. Under the new regime it was sworn in only in June 2026. The World Bank documented outstanding implementing regulations, a Commission not fully operational, a digital cadastre suspended after launch, and substantial licensing delays.

The promise of the MRCA is not only whether Zambia has designed a better mining law.

It lies in whether its institutions become durable enough to make the fifth rewrite the foundation for Zambia’s next era of mining, rather than leave a sixth to solve its unfinished business.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

Tanganyika Concessions sent an expedition into the Kansanshi ground in 1899. On 6 September, it reached the existing workings there and cabled back that the deposits were “scandalously rich.”

But the expedition had arrived at a place with a much older mining history. Kaonde oral tradition holds that ancestors worked this ground long before 1899, while later research dates mining at Kansanshi as far back as the fourth century AD.

Over the next hundred years, Kansanshi changed hands four times. ZCCM held approval to work the open pit from 1969 until 1998, when Cyprus Amax completed an 80% acquisition signed the year before. Phelps Dodge bought Cyprus Amax in 1999, later folding into Freeport-McMoRan.

In August 2001, First Quantum Minerals bought an 80% stake from Cyprus Amax for US$2.5m cash, 1.4m shares, and payments tied to a future construction decision. ZCCM Investments Holdings Plc retained its 20%; in 2022 it converted its dividend rights into a 3.1% life-of-mine revenue royalty, approved by shareholders at an Extraordinary General Meeting in Kitwe on 31 March 2023.

The original licence dates to 1997. Renewed in 2022 for another 25 years, it now carries Kansanshi into the 2040s. Under FQM, the mine became Zambia’s largest copper and gold producer by output, yielding 181,000t in 2025, its highest since 2021.

Kansanshi’s ore is unusually varied for the Copperbelt: sulphide, mixed, oxide and quartz-magnetite types, hosted in a carbonate sequence on the Solwezi Dome rather than the region’s typical single stratabound ore shale.

Proven and Probable Reserves stood at 1.07 billion tonnes at 0.52% copper in December 2024, among the belt’s largest single endowments.

In August 2025, FQM completed the US$1.25bn S3 expansion, opened by President Hakainde Hichilema. Its 50MW combined SAG and ball mill is among Metso’s largest installations at a single site. The expansion nearly doubled processing capacity from 28Mtpa to 53Mtpa and shifted Kansanshi from selective high-grade ore to bulk sulphide processing at depth. Its smelter, expanded alongside S3, will produce over 400,000t of copper anode annually; FQM reports more than 1,500 local suppliers across its Zambian operations.

In Q1 2026, Kansanshi produced 45,345t of copper, with the new circuit running above design capacity according to company investor materials. Full-year guidance is 175,000-205,000t. FQM reports more than 13,000 people employed at Kansanshi; FQM’s 2025 reporting puts the year’s wage bill at US$390m, up 65% on 2024.

The pit is now around 220m deep; ultimate design depth estimates range from 400-500m, depending on the technical source.

Sixteen centuries after this hill was first worked by hand, Kansanshi is still going deeper, toward a pit that will approach half a kilometre below where that story began.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

The deepest mine on earth runs nearly 4 km straight down, under a farming town west of Johannesburg. Getting to the working face takes nearly an hour. At that depth, almost everything about mining becomes harder.

Mponeng was also the last South African gold mine held by the lineage that defined the industry. Anglo American was founded on the Rand in 1917. Its gold became AngloGold, then AngloGold Ashanti. When Mponeng was sold to Harmony Gold Mining Company Limited in 2020, that lineage no longer mined an ounce of gold in the country where it began.

But the ground was lived in from below long before any shaft. In the Lepalong caves outside Carletonville, the BaKwena, led by kgosi Maselwane, built an underground village, sheltering from Mzilikazi's raids between 1827 and 1836, beneath ground later mined 4 km down.

The Witwatersrand reefs tilt down into the basin, and the richest have been chased down-dip for a century, further from surface each decade. Western Deep Levels began sinking in 1981 and poured first gold in 1986. The reef it chases is about 2.7 billion years old.

Mponeng's recovered grade in 2025 was 11.27 g/t, eight times the global average. The cost is depth. The rock at the bottom sits well above sixty degrees Celsius, and about 6,000t of ice a day is pumped down the shafts to keep the work survivable. Seismic events are recorded underground every day. The ice is the price of standing where the gold is.

Around 6,000 people work the mine. Carletonville exists because of mining. Khutsong next door depends on the shafts, and sits on dolomite that decades of dewatering have turned treacherous, with 3,187 households under sinkhole threat. The deepest question at Mponeng is not in the rock. It is whether the community built on the mine outlives it.

The ownership runs the length of South African gold. Anglo American folded its gold interests into AngloGold in 1998, which became AngloGold Ashanti in 2004. Through the 2010s, AngloGold Ashanti sold down its South African portfolio, including the Vaal River operations to Harmony in 2017 and 2018. Mponeng, the deepest and the last, followed in 2020, sold for about R4.4bn against a book value of about R9.9bn for the package.

Then Harmony did what the previous owner would not. In 2024 it committed R7.9bn to carry Mponeng to around 2044, nearly two more decades for a mine that had been given seven, adding 3.05 million ounces to reserves.

In Carletonville that is not counted in ounces. It is counted in wages, in school fees, and in another generation of work in a town that has watched the shafts around it close. Harmony did not simply buy the last mine of an old South African gold lineage. It gave the deepest mine on earth another life.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

South Africa has not replaced its principal mining statute since the Mineral and Petroleum Resources Development Act was enacted in 2002 and commenced on 1 May 2004 — 22 years under one Act.

The MPRDA was amended in 2005 and 2008, but never replaced. In May 2025, the state published a Draft Amendment Bill for public comment. The pattern is not serial rewriting. It is serial amendment of a statute the state has never been willing to abandon.

I lived with South African mining law for my working life. This Act defined 20 years of practice for me. But this piece is not a legal opinion; this is reading my professional domain as a story.

The MPRDA placed mineral resources under state custodianship, framed around equitable access and sustainable development. The 2025 Bill does not propose a new settlement. It sharpens the existing one — and its provisions suggest the original architecture did not deliver what it promised on its own terms.

The Bill defines “meaningful consultation” on the face of the Act, requires publicisation of applications within seven days, and routes community objections to the Regional Mining Development and Environmental Committee. The Act already required consultation. The Bill now defines what “meaningful” means.

That matters after the Baleni judgment. Baleni held that where communities hold informal or customary rights to land affected by mining, consultation is not enough: full and informed consent is required before a mining right may be granted. I doubt defining consultation more precisely will displace that higher threshold.

On beneficiation, the Bill seeks to amend section 26 in two moves. First, it turns the Minister’s promotion duty from permissive to mandatory — from “may” to “must.” Second, a new subsection requires every producer to make minerals available for local beneficiation.

But the obligation is insufficiently defined and leaves untouched the economic and structural constraints that have frustrated beneficiation ambitions for more than 2 decades.

That matters because South Africa has been here before. In 2021, the High Court held that Mining Charter III was policy, not law, and set aside key provisions. The Charter was designed to carry what the Act did not. And a policy instrument cannot carry more than its statutory foundation.

In the Bill, government appears to have accepted the lesson: empowerment obligations move into the Act, where they belong, still subject to constitutional safeguards.

On beneficiation it risks repeating the pattern. The vulnerability is different, but the difficult question — whether the obligation is defined with sufficient precision — is again left for the courts. My prediction: this provision has a difficult road ahead.

The MPRDA endures because it carries the post-apartheid mining aspiration. It keeps being amended because it avoids answering the difficult questions.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

The Nchanga Open Pit wraps in an 11-km arc around Chingola. Nearly 30 km². The second-largest open-cast mine in the world. Nineteen kilometres north, the Konkola shaft descends 1,505 m into an orebody grading 2.9% copper.

This is not one mine. It is a copper system: smelter, refinery, tailings leach plant, pyrites mine. In a hundred years, it has never been fully operated.

The Lamba people worked copper on what they called Ilamba long before any concession existed. Francisco de Lacerda recorded them trading copper and ivory in 1798. In 1895, a colonial scout encountered people wearing copper bracelets along the Kafue River. The concessions arrived in the 1920s.

Anglo American built the system through Nchanga Consolidated Copper Mines. Nchanga's open pit commenced in 1957, with ore grades around 6% copper. Konkola's shafts went down in the 1940s and 1950s. Chingola was founded in 1943 to house the people the mine required. In full production, KCM became Zambia's largest private employer.

Zambia nationalised the mines in 1969. ZCCM Investments Holdings Plc absorbed NCCM in 1982. During privatisation in the late 1990s, Anglo re-entered with 51% of the newly formed Konkola Copper Mines plc. The system it inherited was not the system it had left. Decades of deferred investment made the Konkola Deep project's economics unworkable. Anglo took a US$350m provision in 2002 and walked away. In November 2004, Zambia Copper Investments sold its 58% stake to Vedanta Group for US$25m.

A UK claimant law firm active in extractives claims across Africa represented 2,500 Zambian villagers in pollution claims against Vedanta. The UK Supreme Court ruled on jurisdiction in April 2019. Separately, the Zambian government placed KCM under provisional liquidation the following month. Vedanta settled the villagers’ claims in January 2021 without admitting liability.

The Hichilema government reversed course in September 2023. Vedanta paid US$245.75m under a July 2024 scheme of arrangement, and the mine was handed back in August. Monthly output rose from roughly 5,500 tonnes in December 2024 to about 8,200 tonnes a year later.

Vedanta's return has moved from settlement to execution. In February 2026, KCM engaged Australian contractors to advance the Konkola Deep Mining Project. In May it reopened the Chingola B shaft after eighteen years, targeting 200,000 t/month.

Three weeks later, the Nchanga smelter began a sixty-day planned maintenance shutdown, part of a US$2.7bn programme set out in an SEC filing.

Vedanta has routed that programme through CopperTech Metals, a US-domiciled company that filed for a proposed NYSE listing. ZCCM retains 20.6% of KCM.

At 1,505 m, the orebody grades 2.9% copper. A century after this system began to take shape, Vedanta is making the most serious attempt yet to realise its full potential.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

Magnetometric surveys traced the buried West Wits Line in the 1930s. The reef did not show at surface. It turned out to be the richest gold corridor ever worked. South Deep, featured earlier in this series, lies alongside it on the same reef system.

One mine on it, Driefontein, poured a hundred million ounces of gold. No other mine in history has reached that figure. Kloof, beside it on the West Wits Line, also works the Ventersdorp Contact Reef. It has been given one more year. The gold price will decide the next one.

The ground was Kwena country first. From 1827 to 1836, Maselwane’s community lived in the Lepalong caves on Kleinfontein farm, sheltering from Mzilikazi’s expansion. People lived beneath the same ground that would later be mined from below.

Capital arrived by instrument, not outcrop. GOLD FIELDS backed magnetometry to trace the buried formations and pressure cementation to sink through water-bearing dolomite. It invented neither. Its contribution was to recognise what both made possible, combine them and commit capital when much of the industry would not. West Driefontein milled its first ore in 1952. Gold was first mined on what is now Kloof in 1934.

On 30 August 2005, after fifty-three years, Driefontein poured its hundred millionth ounce. Its lifetime grade averaged 14.2 g/t. By then it employed about 16,800 people.

The ground that gave also took. Men died in a sinkhole in 1962. A vast water inrush nearly took the mine in 1968. It was rebuilt and kept running. Khutsong, built beside the mines and sustained by their labour economy, grew on that gold. Its economy now thins as the mines do.

Gold Fields’ lineage shaped the district for decades. In 2013 it unbundled its mature South African mines into Sibanye Gold. Sibanye-Stillwater runs both today. DRDGOLD Limited, now majority Sibanye-owned, reprocesses West Rand tailings at one of the world’s largest gold surface-retreatment operations.

Sibanye’s 2025 year-end plan runs Driefontein to 2036. Kloof has one year at a time. Driefontein reaches beyond 3,000m; Kloof is also ultra-deep. The rock runs hot and every ounce carries the cost of reaching it. The CEO told the market in February 2026 that Kloof has no long-life reserves and will run while it makes sense.

Its H2 2025 cost was about US$4,080/oz, against spot gold near US$4,340/oz. A court has ordered Eskom to grant the wayleave sought for a solar plant. As elsewhere across the region, electricity prices bear directly on jobs and whether mines can continue to operate. Gold and electricity prices are now central to Kloof’s year-by-year economics.

The West Wits reef system that made Driefontein continues beneath Kloof. Its future now lies in what can still be mined underground, what cheaper power can preserve, and what can be recovered from the tailings.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

Two mines, one company, and neither has finished finding out how deep the ore runs.

The Lamba people had a name for this ground before any of them: Ilamba. Lambaland. They smelted its copper and traded it along routes reaching the Indian Ocean long before any company registered title to it. A Portuguese explorer documented them doing so in 1798. A British scout found them wearing copper bracelets along the Kafue River in 1895, marked the deposit, and moved on. What he filed as a discovery was what the Lamba already knew. Anglo American arrived thirty-three years later.

Anglo American did not come to Nkana for the scenery. The ore graded approximately 2.5 times the contemporary US benchmark. Rhokana Corporation sank the first shafts in 1928, backed by capital from South Africa, Britain, and the United States. On 17 March 1932, the smelter produced its first blister copper.

From first production to 1969, Rhokana Corporation held the mine for thirty-seven years, the largest employer on the Copperbelt. The Matero Declaration brought the Zambian government to 51%. Rhokana was absorbed into state ownership and, by 1982, into Zambia Consolidated Copper Mines. The mine had its second name: Nkana, under ZCCM.

The ore body had not moved.

On 1 April 2000, Glencore, through its Carlisa vehicle, and First Quantum Minerals acquired the Nkana and Mufulira assets from ZCCM. Mopani Copper Mines Plc became the name above the gate.

Ownership changed again. In January 2021, Glencore and First Quantum transferred their combined 90% stake to ZCCM Investments Holdings Plc for US$1, with US$1.5bn of debt assumed. Three years later, ZCCM-IH sold 51% to Delta Mining, a vehicle of UAE-based International Resources Holding, under a US$1.1bn investment commitment. ZCCM-IH retained 49%. A Golden Share remained with the government.

Nkana’s Synclinorium Shaft, at 1,280m, held the depth record for four years. Mufulira’s Henderson Shaft passed it in June 2020, reaching 1,580m. Mindola Deeps, raise-bored to 1,892m, is still being equipped toward its 2,020m target. In Glencore’s final quarter, monthly production averaged 3,412 tonnes.

Under ZCCM-IH from April 2021, it averaged 8,598 tonnes.

The same deposit carries cobalt, mined continuously since the Nkana Cobalt Plant opened in 1982. The energy transition has not changed the ore body. It has changed who wants it.

Zambia’s Ministry of Mines confirmed copper output up 40.27% year on year in 2025. IRH has committed more than US$1.1bn to the recovery and has stated a target of 300,000 tonnes a year by 2029. In June 2026, Mopani declined Zambia’s government-allocated copper concentrate export waiver for the third consecutive time. All output goes to domestic refining.

Two mines, one gate above them both. The concentrate stays in Zambia this time. The ore body has not changed. The decisions around it have.

Pattern Counsel is written for General Counsel, Chief Legal Officers and their teams — and for the law firms, legal operations professionals and technology providers who work alongside them. Originally published on LinkedIn. This work is AI-assisted—read about how I work.

Pattern Counsel reflects my vision of how in-house legal functions could work. Many General Counsel will see it differently—and in the coming months, I’ll invite some of them to share their own visions.

Introduction

Being inside is two things, not one.

A position in the enterprise: where legal sits, who it reports to, which rooms it is allowed into. And a presence across the enterprise’s work: whether legal is there across the phases where the work creates, manifests or exposes legal risk—or only in the phases into which it is invited. Most functions win the seat and assume it guarantees that presence. It doesn’t.

The first instalment, last month, put the model plainly. Legal is an enterprise resource, not a service provider. It compounds value through embeddedness, in the business and across the phases of work that demand it. It trades replicable work out to a governed process. And it is judged by what the enterprise achieves where legal is embedded, not by what legal produces.

One of those four patterns—that value compounds only from inside—depended on when the lawyer was present, not where she sat. I offered it as a pattern I’d noticed, not a theory I’d built. This instalment gives that pattern a method, with the same caution.

Two words, kept apart

The argument depends on separating two things the profession often treats as one.

Positional embeddedness is where legal sits in the enterprise: its place on the org chart, its reporting line and its seat at the table. It is about structure and relationships—who trusts legal, who brings it into the work and how close it is to the people making decisions.

Processual embeddedness is legal’s presence across the enterprise’s work. Enterprise work moves through phases. However those phases are described, one of the central insights of modern management is that work unfolds as a process. Processual embeddedness means having legal present in the phases where that work creates legal risk.

Position tells us where legal sits. Process tells us when legal is present across the enterprise’s work.

To be clear from the outset, this essay is not about legal’s process. The processes discussed here belong to the enterprise. Legal will have processes of its own, but it must design them around the enterprise’s work, not mistake them for that work. When the enterprise must hand its work over to a separate legal process, legal remains outside the work it seeks to shape.

Both dimensions are necessary. Positional embeddedness does not deliver processual embeddedness.

Three answers to where

In my experience, strong positional embeddedness takes three common, often overlapping forms.

The trusted advisor. The General Counsel as counsel to the chief executive, the chair and the executive team. Influence through relationship and proximity to power.

The governance model. Legal placed at the enterprise’s formal gates: the approver of contracts, the keeper of the delegation framework and the desk to which trouble is brought. Authority by mandate rather than relationship.

And the distributed model. From my own experience, this is most developed in financial services. Trusted advisors embedded in business units or product lines, reporting to a group General Counsel who advises the group chief executive. It is the strongest answer the positional axis has produced: a lawyer in every lane, close to every business and senior enough to be heard.

When I first encountered the distributed model during a visit to a long-term insurer in 2015, I was convinced it was the whole answer. It is not. A function can place a trusted lawyer in every unit and still bring each of them in only at execution, after the choices shaping the risk have been made. Structurally present. Temporally absent. Seating a lawyer inside the business settles where she sits, not when she enters the work.

Another common arrangement is a Head of Legal reporting to the CFO. This is a positionally constrained answer to where. Legal sits a step removed from the chief executive and the executive team, with access often mediated through another function. Processual weakness is not inevitable, but it becomes more likely: the further legal sits from those directing the work, the more likely it is to enter only when invited.

All of these arrangements answer where legal sits. None, by position alone, answers when legal is present across the work.

The experiment already ran

Here is the helpful part. Legal is not the first support function asked to move from the back office into the business. Finance did it. HR did it. Risk and compliance did it. Each developed a model, tested it at scale and left a record legal can learn from.

Finance developed business partnering. HR developed the HR business partner model. Risk and compliance developed the three lines model. Each sought, in different ways, to move closer to the business.

In each case, the position was secured but the value did not always follow. The literature on HR business partnering is particularly blunt: the model has often fallen short of its promise. A recurring reason is that position did not become presence. Being at the table is not the same as being in the work the table decides.

This is not peculiar to legal, or to mining. It is what happens when a function treats the seat as the destination.

Finance, HR and risk won the seat. The seat was not the thing.

The lament, and the layer under it

Every in-house lawyer has made the same complaint. I wish they’d involved me earlier.

It is a real signal. The lawyer has felt, without quite naming it, that the work has a shape and that the timing of legal’s presence affects whether it can add genuine value or prevent its loss. Legal keeps arriving at the wrong point in that shape—often after the choices that mattered have been made. That is the processual axis surfacing.

But the complaint turns that insight into a blanket prescription: bring legal in earlier and involve it more. Neither is always the right answer. Sometimes the answer is less legal involvement at execution and more at review. Sometimes it is no involvement at all in a phase legal has attended only out of habit.

Earlier is a direction, not a diagnosis. It tells us to change the timing, but not which phase requires legal’s presence.

There is a subtler error beneath this instinct, and it is where most coverage mapping goes wrong. When a General Counsel maps the function’s coverage, the map is almost always drawn against the org chart. Do we support HR? Do we cover business development? That is the positional instinct applied to a processual question. It produces a picture of which departments legal is near, rather than of the enterprise’s work in which legal is present.

The load-bearing work does not stay within departmental portfolios. It moves through streams that cut across them: a single transaction may draw in finance, commercial, tax and operations at once. Each stream, in turn, moves through phases. The executive committee sees portfolios. A processual map must go two levels deeper: first to the streams that cross them, and then to the phases within each stream. Most coverage maps never reach that second level.

How the work is shaped

A stream of enterprise work has a shape, and the shape has phases. Deming gave a clean version, building on Shewhart before him: plan, do, study, act, a loop in which acting on the lesson becomes the next plan. Last time I ran on three phases, planning, execution and review. This is the same shape read closer. Study and Act separate review into its two real parts: judging what happened and carrying that judgment into the next plan. The distinction matters because carrying the lesson forward is where the compounding occurs.

I use Deming because the cycle is clean, not because it is the only one; a gated model or a lifecycle model would do the same job. The claim is not about Deming. It is about reading the enterprise’s own work into phases so that, at each one, you can ask a single question. Does this phase expose, manifest or create legal risk, and am I present where it matters?

Legal’s task here is rarely to run the work. It is to read it: to understand how it moves through its phases, to recognise where legal risk is created, manifested and carried forward, and to decide where its own presence changes the outcome. That is the whole difference.

Take the work legal is most certain it owns. A major commercial contract.

Plan: The enterprise plans it before there is anything to draft. The commercial team decides to enter the arrangement, settles on the counterparty, fixes the shape in principle: price, volume, duration, exclusivity, the terms of exit. No clause exists yet. Every consequential legal risk does. Whether the counterparty can be relied on to perform. Whether the structure trips merger control. What dependence is being locked in, and on whom. Decided in that room, by non-lawyers, before a word is drafted. This is where being inside pays out. The lawyer who lived the last three deals of this shape knows which counterparty defaulted, where the regulator pressed, what the exclusivity cost when the market turned. She is one of the few whose presence changes the outcome. Not drafting. Interrogating the structure while it is still soft enough to move.

Do: Then the enterprise does it. The deal is papered, negotiated, closed. Here legal genuinely produces, and the drafting is real, skilled work. Here is the trap. This is the one phase where legal makes something, so it is the phase legal mistakes for the whole stream. The risk is manifested here, allocated by the words on the page, but it is barely created here, because the room to create it closed at Plan. Legal is one of many now: commercial holds the terms, finance holds the model, legal holds the paper. Being the drafter at execution tells you nothing about whether anyone was present when the exposure was set. Most functions arrive at this phase and call it involvement.

Study: Then the enterprise studies what it signed, if anyone does. The contract runs. Clauses are invoked, or they leak. The counterparty behaves as the paper assumed, or it does not. The pricing does something when the market moves. This is the business living with its own decision, and reading it with legal eyes is work only someone who was inside the deal can do. One of the few again. Most functions skip it. Close the deal, move to the next.

Act: Then the enterprise acts. The lesson carries into the next arrangement of this shape: the template amended, the position for this kind of counterparty improved, the next deal begun from higher ground. The compounding is the enterprise’s, not legal’s, and legal’s presence at Study is what feeds the better Plan. Act into Plan is where the line becomes a loop, and the loop is the whole mechanism.

A single contract does not loop; it is signed once and done. The class of contract loops. Supply agreements, offtakes and outsourcing deals recur, and the compounding lives at the level of the type, not the instrument. That is where the template and the playbook accumulate.

The result is not another legal framework. It is a diagnostic for reading the enterprise’s work: four questions, asked stream by stream.

• Do we need to be at this phase at all?

• In what form: leading, influencing, contributing, or only watching?

• How much legal presence does this phase demand?

• Are we present enough to fulfil the role we claim as an enterprise resource?

The diagnostic has only one purpose. It tells you where legal presence is genuinely demanded and where it is merely habitual. You do not yet need to decide what to release. You now know where to look.

The questions are allowed to answer no. They are allowed to answer less. A function present everywhere at full weight is not the ideal. It is over-embedded, too close to the business to see it clearly, and its judgment suffers as a result. Calibration is not a concession to a tight headcount. It is the correct shape.

The same shape, without a mine in it

The pattern is not mining’s. It is not the contract’s either. A regulatory affairs specialist in a pharmaceutical company works through the same pattern. A drug moves through a gated lifecycle: discovery, preclinical, trials, submission, approval, and the long tail of market surveillance, each submission resting on the documented history of the one before it. The specialist who was present when the early regulatory strategy was set reads the later submission with knowledge that is not in the file: what the agency signalled and never wrote down, why a design choice was made, where the last application met resistance. One of the few at the front. One of many in the machinery of a filing. One of the few again when the regulator responds and the next submission is shaped. No tailings dam in sight. The same pattern exactly.

A deliberate act

None of this happens by good intention. A function does not drift into presence across the phases that matter. It maps the enterprise’s work and designs its presence against it, or it arrives where habit and invitation put it.

For a mature function, the mapping is an audit, and an uncomfortable one: going stream by stream and asking honestly where we arrive against where the value accrues. Most functions that run it find the same answer. We arrive at Do—present for execution across almost everything and for planning across almost nothing—and we have been calling the execution half coverage. The diagnostic turns a vague unease into a specific question. The map itself is already valuable. Most functions have never seen their work arranged this way.

Designing that presence from the beginning is different work altogether. It belongs to function design rather than function audit, and it is the subject of a later instalment.

Presence across the enterprise’s work is designed, or it is accidental. There is no third case.

What it costs, and what pays for it

I will not pretend this is free. Reading every load-bearing stream at the level of its phases, then ensuring legal is present where judgment compounds, is more work than turning up at execution and drafting well. And it lands at the worst possible moment, when the enterprise is asking legal to cost less, not do more.

The additional work is not overhead. It is what makes legal an enterprise resource, and it is funded the only way it can be: by releasing the replicable work that does not compound, the work an external channel, properly instructed, would do just as well.

The presence you owe is paid for by the delivery you let go of.

Which work belongs in each channel, and how you decide it without losing hold of what matters, is a later question. It is the same diagnostic turned inward, onto legal’s own work. This instalment has used it to read the enterprise; a later one will use it to redesign the function itself.

Concluding this instalment

This is the discipline the series set out to take up: where presence is demanded, and where it isn’t.

Being inside is two things. A position, and a presence. Position is necessary and routinely mistaken for sufficient. Finance, HR and risk each won it and learned the hard way that it did not carry the presence, and legal is about to be asked the same question under AI’s pressure. The method for answering is not a framework to adopt. It is a way of reading the enterprise’s own work into phases and asking, stream by stream, where legal risk is made and whether the function is there when it is. Designed presence, or accidental. There is no third case.

Why presence compounds at all—what kind of knowledge being inside the work produces that being outside it cannot—is the subject of the next instalment.

A note on sources

The word embedded is not mine, and I have tried not to use it loosely. I did not arrive at the idea through economic sociology. I arrived at it through practice, and only later found that others had been working on embeddedness for decades. Looking back, it echoes Mark Granovetter’s work on how economic action is embedded in social relationships, and Brian Uzzi’s work on how those relationships affect organisational performance. Uzzi also shows that embeddedness has limits: beyond a point, it can impair rather than improve judgment. That finding supports this instalment’s argument for calibrated rather than maximum presence.

That legal makes its value upstream, before the problem rather than after, is the heart of the Nordic School of Proactive Law, and of Helena Haapio’s work in particular. It gave a rigorous vocabulary to something I had already been watching.

I learned most from the functions that made this journey before legal: finance business partnering, Dave Ulrich’s HR business partner model and the honest post-mortems on it, and the three lines model in risk. Their failures taught me more than their successes.

Deming and Shewhart supplied the cycle. I borrowed it as a way of reading enterprise work, not as a quality-management method.

One honest gap remains. I could not find a study that puts a number on the cost of late legal involvement in corporate decision-making. The proactive-law literature makes the case in principle; the analogue functions show the pattern in practice; legal’s own number is not yet in, and I will not pretend otherwise.

Continue the discussion

If this reading matches your own experience, or if it points somewhere different, I would value the exchange.

Kevin Lester is an independent counsel and thinking partner. He spent fifteen years as a Country Head of Legal and subsequently as General Counsel at Anglo American. He advises General Counsel, senior executives, and boards at kevinlester.co.za.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

The ore body at Freda Rebecca was laid down in the Bindura-Shamva greenstone belt, an Archaean formation 2.65 billion years old. The mine above it has had seven different owners since Cluff Resources poured the first gold in 1988.

In March 2007, the mine went to care-and-maintenance. Zimbabwe’s inflation was climbing toward an official peak of 231 million percent, and production was impossible. In March 2026, it produced 240 kg of gold, its highest monthly output since 2023.

The Shona had worked gold on the Zimbabwe plateau since at least the 11th century CE. Their mhizha craftsmen traded through the Munhumutapa state into Indian Ocean networks reaching Kilwa and Sofala.

Algy Cluff arrived in Harare just weeks after Zimbabwe’s independence in 1980. He found the Rhodesian Geological Department’s records intact, and unemployed geologists to read them. Everyone else had pulled out. The knowledge had not. Cluff Resources poured its first gold at Freda Rebecca in 1988. At peak, in 2002, the mine produced just over 3 t of gold a year.

Freda Rebecca employs over 800 people directly. Bindura’s commercial life has tracked the mine’s output for nearly forty years. The mine operates a health centre and a primary school. When the mine restarted in 2009, it offered to re-engage fifty-six of its workers on inferior terms. They refused, and were dismissed. In 2022 the Supreme Court found in their favour and sent the case back to be reheard. It remains unresolved.

In 1996, Ashanti Goldfields Zimbabwe acquired it. AngloGold Ashanti absorbed Ashanti in 2004 and said within months it was willing to sell the Zimbabwe mine. Mwana Africa Plc bought it for US$2m, completing in April 2005, and restarted it in October 2009. The mine passed through ASA Resources and a disputed transfer to Sotic International, which bought it and Bindura Nickel out of administration for £23m, before Kuvimba Mining House took control in 2020. Kuvimba was 65% state-owned. It invested US$22m, lifting monthly output from 150 kg to over 300 kg. A presidential decree in September 2023 moved the state’s Kuvimba shareholding, and those of 22 other enterprises, into the Mutapa Investment Fund. The Fund transferred Freda Rebecca to its state-owned gold subsidiary, Mutapa Gold Resources, on 1 January 2026.

Mutapa Gold Resources is raising US$75m through a local debt issuance. The group is targeting 570 kg per month by 2028 and has stated a total capital requirement of US$250m. Freda Rebecca’s share of that is 270 kg a month. Management puts the mine’s remaining life at four years and is spending US$14m on drilling to take it to ten. Construction at the Shamva Hill open-pit begins in August 2026.

The ore body that Cluff Resources first poured gold from in 1988 poured its best month in years in March 2026.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

Newly independent Zambia wanted to reset the economics of mining. Kenneth Kaunda did what many nations have done. He did it with one distinction: he paid.

The Mulungushi reforms of 1968 had already nationalised Zambia’s major industries. A year later the Matero Declaration placed 51% of the Copperbelt’s mines into state hands.

By 1969, the same two houses that had dominated the Copperbelt since the 1930s had restructured, absorbing several of their founding investors into new corporate forms. American Metal Company had become AMAX in 1957. The British South Africa Company had gone into Charter Consolidated in 1965. AMAX held 43% of Roan Selection Trust. Anglo held minority interests in RST’s subsidiaries, and 12.25% of Roan Consolidated Mines.

Compensation was to be settled in bonds redeemable over a decade. In 1973 the government redeemed them early and ended the management contracts under which both still ran their former mines. Both left with money in hand.

Anglo’s own vehicle was Zambian Anglo American. In a statement by that company to shareholders published on 16 September 1971, H.F. Oppenheimer set out the policy: capital released from Zambia by the 51% acquisition would go to Australia and the USA. In Australia, the capital formed Australian Anglo American with about £9.8m. In the USA, investments had already been made into Engelhard.

Zambian Anglo American was redomiciled to Bermuda and renamed Minorco. That company and Anglo American would merge their interests in 1999 to form Anglo American plc.

Roan Selection Trust ran a different route. Selection Trust, the British holding company, passed to BP in 1980, then to Rio Tinto in 1989. The AMAX lineage ran to Cyprus AMAX, Phelps Dodge, then Freeport-McMoRan. Zero presence in Zambia.

Eunomix Research put Zambia’s lost mining rents between 1970 and 2010 at about US$45bn. That is what the rupture cost. Production fell from a peak of about 770kt in 1969 to 250kt in 2000. The world did not stop needing copper.

Anglo’s mines at Nchanga and Konkola passed through privatisation into KCM, now held by Vedanta Group.

Nkana, Anglo’s other great Copperbelt mine, was combined with RST’s Mufulira to form Mopani Copper Mines Plc, now 51% owned by International Resources Holding of Abu Dhabi. RST’s remaining Zambian mines passed to CNMC at Luanshya and CNMC/NFC Africa Mining at Chambishi.

Zambia produced close to 900kt in 2025. The government targets 3Mt by 2031. That ambition is the measure of both the depth of the decline and the scale of the recovery. The rebuilding is now being led by Barrick Mining Corporation, First Quantum Minerals, Vedanta Group and IRH.

What’s missing from that arithmetic isn’t corporate. It’s a generation of Zambians who grew up on the far side of what the copper could have built.

1969 was the lesson. 2031 is the exam.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

My friends know I am a genealogy enthusiast. Today’s story is the wrong kind of genealogy: it traces extinction, not descent.

Five capital families built the flagship copper mines across southern Africa between 1900 and the 1960s. Anglo American. AMAX and Selection Trust. Newmont Corporation. Rio Tinto. And Union Minière du Haut-Katanga.

Anglo ran Zambia’s Copperbelt through Rhokana, Nchanga, and Bancroft. Its Katanga exposure ran indirectly, through Tanganyika Concessions and a board seat at UMHK.

AMAX and Selection Trust ran the RST mines in Zambia and BCL in Botswana. BCL was not solely an AMAX asset. Anglo was a founding party alongside RST, held through its subsidiary BRST. When AMAX exited in 1993, Anglo remained the dominant private partner. Selebi-Phikwe was an extension of the same Copperbelt capital that built Zambia.

Newmont held Okiep in the Northern Cape. At Tsumeb in Namibia, the two American houses met each other.

Rio Tinto ran Palabora for 57 years. Anglo held 16.8% alongside it.

UMHK built a copper state within a state in Katanga, producing around 300,000t a year by 1960.

Then the rupture.

It did not happen once. It happened four times.

January 1967: Mobutu seized UMHK’s Congolese estate. The break was total. August 1969: Zambia nationalised 51% of both the Anglo and RST groups. Management contracts were cancelled by 1973.

Commercial exits followed by different means. Tsumeb reached 1,700m, entered a price collapse, and went into liquidation in 1998.

BCL was placed into liquidation in 2016. More than four thousand workers were made redundant on a single day.

Not one of the five families, nor a descendant, holds a copper asset on the ground in southern Africa today.

Two institutional threads survived the rupture: the post-nationalisation state-owned entities to which the original assets were granted. In Zambia, the Anglo and RST mines passed to ZCCM Investments Holdings Plc, which now holds minority stakes in Mopani, Konkola, and Kansanshi. It operates none of them.

In the DRC, UMHK’s seized assets were granted to Gecamines S.A., which holds minority positions in Tenke Fungurume and KCC. It operates none of them either.

Anglo’s copper is now in Chile, Peru, and the Americas. AMAX fed to Freeport-McMoRan via Phelps Dodge; Selection Trust fed to Rio Tinto via BP Minerals. Neither in southern Africa.

The new holders inherited the ground: International Resources Holding at Mopani, Vedanta Group at Konkola, Ivanhoe Mines and Zijin Mining Group at Kamoa-Kakula, CMOC at Tenke Fungurume. Extraordinary ore bodies. They did not inherit the institutions or the decades between.

The years the rupture cost are not recoverable.

hashtag#MiningIsHuman hashtag#Copper

Though not a capital family like the others in the story, Anglovaal built Prieska in 1971 and closed it 20 years later in 1991. It is now being developed by Orion Minerals Ltd.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

South of the Vaal River, 180 km from Johannesburg, a shaft drops 3 km into the earth. At the bottom sits the Vaal Reef, Neoarchean banket conglomerate laid down 2.7 billion years ago, carrying gold at 8.21 g/t recovered grade. That is roughly four times the global average for operating gold mines. The rock is ancient. The investment decision is current.

AngloGold Ashanti built this mine. Anglo American had worked Great Noligwa on the same ground since 1968, peaking at 483,000 oz a year before structural decline set in and the mine went on care and maintenance in 2015. At its height, the Vaal River complex was producing more than a million ounces a year. By the late 1990s, with Noligwa declining, the deeper ground was the answer. Moab Khotsong opened on a new-order mining right, sinking to 3,106 m. At construction, that was the world’s deepest single-lift shaft. It reached high-grade Vaal Reef below what the older workings could access. Stoping began in November 2003.

Tswana-speaking communities farmed and kept cattle along the Vaal here for centuries before the first colonial survey crossed this ground. Tswana remains the first language of Orkney, the town above the mine today.

As the gold price weakened after 2012, AngloGold Ashanti deferred Zaaiplaats, the planned life extension, citing competing capital priorities across its global portfolio. Four years later, in October 2017, it agreed to sell Moab Khotsong to Harmony Gold Mining Company Limited for US$300m. On those plans, the mine had about seven years of producing life left.

Today, 6,232 people work Moab Khotsong. The Tswana-speaking town was named Orkney by Simon Fraser, a Scots miner from the Orkney Isles, in the 1880s. The streets carry British literary names. In FY25, the mine was seismically active: heightened seismicity disrupted the middle mine; the top mine halted briefly for toxic gas removal. At 3 km below surface, seismic risk and gas management are part of every shift.

Harmony approved Zaaiplaats in 2021. Three declines are now sinking from 101 Level toward 103 Level, the critical-path milestone for production access, at 9 degrees, each 2,915 m long, no new shaft required. Zaaiplaats holds approximately 3.7 Moz of potential additional reserves at around 9 g/t.

Harmony spent R962m on the project in FY25 alone, with gold averaging US$2,622/oz that year. Moab Khotsong produced 198,820 oz in FY25. Production will dip over the next few years as the middle mine exhausts, with FY26 guided at 175,000–178,500 oz. It recovers to 225,000 oz a year at steady state.

The mine life, roughly seven years at acquisition, is now 19 years. The decline is sinking. Nineteen years of reef still to mine.

Orkney — British literary street names, Tswana first language, nothing like the islands it was named for — just secured nineteen more years of school fees and wages.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

The ore body under this low Kalahari hillock, about 75 km from Kuruman, could supply the world’s manganese requirements for 5,000 years. In 2025, the smelter built to process it closed because South African electricity had become too expensive to run a ferromanganese furnace.

The ǀXam San knew this ground long before any geologist mapped it. Their territory ran through the scrublands and river courses of the Kuruman Hills. Wonderwerk Cave, 45 km away, records two million years of continuous human presence in the region. The first geological note on Black Rock appeared in 1908. The first shaft came in 1940.

Assmang was formed in 1935 through a partnership between African Mining and Trust and Anglovaal. In April 1940, they acquired the Black Rock outcrop. Three underground mines now draw from a single ore body: Nchwaning II and Nchwaning III, at depths of 421 and 500 m, and Gloria at 140 m. Nchwaning alone holds 323 Mt at 42.5% manganese. The ore is particularly low in phosphorus: a specification steel mills pay a premium for.

The smelter at Cato Ridge processed Black Rock ore for 66 years and employed 310 permanent workers and 290 contractors. On 31 August 2025, it closed permanently. Electricity costs had risen 930% since 2008. Making ferromanganese in South Africa had stopped making financial sense.

The smelter Assmang built in Sarawak from 2013, operational from 2016, remains open and expanding. It was built there because Malaysian hydroelectric power is reliably cheap. That is why Black Rock ore crosses the Indian Ocean to be processed in Borneo.

Sakura Ferroalloys, the Sarawak plant, is a joint venture with Sumitomo Corporation and China Steel Corporation. It produces approximately 240,000 tonnes of manganese alloys per year. Assmang’s relationship with Sumitomo dates to 1961, when the two companies established a marketing arrangement for Black Rock ore into Japan. In September 2025, Sakura broke ground on a new converter with US$82.5m of committed capital, targeting full production by mid-2027.

South Africa holds the world’s largest manganese reserves and exports almost all of them as ore. The electricity tariff that closed Cato Ridge makes domestic beneficiation uneconomic across the sector. Whether the cost of that has been counted is a different question.

African Rainbow Minerals Limited and Assore Group each hold 50% of Assmang Manganese Black Rock Mine Operations. In 2025, Assmang sold its 54.36% stake in Sakura to Assore for R2.1bn. ARM distributed R900m of those proceeds to shareholders. Assore now owns the Malaysian furnace.

The mines produced 3.7 Mt in 2025. South Africa exported a record 26.2 Mt of manganese in 2025. Assmang is part of a consortium of mining groups, bidding to build a new manganese export terminal at Ngqura, with planned capacity of 16 Mt per year, expandable to 22 Mt.

The constraint was never the ore.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

Ernest Oppenheimer made the largest private capital bet in South African history. It rested on a fixed price. In July 1944, 44 nations signed the Bretton Woods Agreement and pegged gold at US$35 per troy ounce. For a generation, nobody had to guess what gold would be worth.

The land had been Basotho. Free State-Basotho Wars, 1858 to 1868. Treaty of Thaba Bosiu, 1866. British annexation, 12 March 1868. The dispossession created the dependency. The mines would follow 80 years later.

The gold beneath it was invisible. No surface expression. Thick sediment cover. No reef to follow down. Oppenheimer was financing something nobody could see. Anglo American was thirty years old. It had survived the Depression and a world war. Oppenheimer was not consolidating what Anglo had built. He was betting on what it could become.

Eight mines in production by 1955: St Helena, Welkom, Western Holdings, Free State Geduld, President Brand, President Steyn, Erfdeel-Dankbaarheid, Loraine. Eight mines in eight years. Built from featureless grassland at depths between 1,200 and 1,500 metres, with no industrial-scale road, rail or water infrastructure on the surface.

Oppenheimer turned the first sod on 16 May 1947. Anglo built Welkom for these mines. Workers came from Lesotho, Mozambique, Malawi, the Eastern Cape. Matjhabeng, the Sesotho name the municipality carries today, describes a place different nations were forced to assemble.

Anglo maintained for decades that it had sought family housing for Black workers in Welkom, and that apartheid policy blocked it. It made that case again at the Truth and Reconciliation Commission.

In more than a century of Anglo American's history, few decisions match this one in scale. Not the founding in 1917. Not the De Beers consolidation. Not the reach that eventually put two-thirds of the JSE under one family's influence. Nor anything since.

The bet worked. At peak, close to a fifth of free-world gold came from these eight mines. 180,000 workers. Applying the Minerals Council's current one to nine multiplier: an estimated 1.7 million livelihoods.

In August 1971, Nixon suspended Bretton Woods. The fixed price was gone. By 1989, gold had fallen from its 1980 peak of US$850 toward US$400. What had been built for a guaranteed price now faced a falling one.

The institutional heir to that bet was AngloGold Ashanti, formed from Anglo's 1998 gold consolidation and the 2004 Ashanti merger. In 2020 it sold its remaining South African assets and left. Harmony Gold Mining Company Limited and Sibanye-Stillwater remain, working shafts on declared end-of-life timelines. By the mid-2020s, around 30,000 workers where there were once 180,000.

Mining extracts a non-renewable resource. When the economics shift, operators sell. The gold runs out. The jobs and prosperity built around it go with it. The only question is what was left behind.

Welkom is still deciding how to answer that.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

The Kalahari Manganese Field runs roughly 35km through the John Taolo Gaetsewe district of the Northern Cape. It holds more than 75% of the world's manganese. Two mines sit at its northern end, sharing the company town of Hotazel. One cuts from the surface. One goes 300 metres underground.

SA Manganese Ltd formed in 1926 to mine the Northern Cape deposits. In 1937, Dr H.J. van der Bijl, who founded both Eskom and Iscor, incorporated Amcor to process minerals into ferroalloys for South Africa's growing steel industry. Those two companies merged in 1975 to form Samancor Ltd, listed on the JSE.

By 1998, Samancor had become the world's largest ferroalloy producer, its reach extending from the Kalahari manganese fields to the ferrochrome smelters of the Steelpoort Valley. One company held both.

In 1998, Billiton (60%) and Anglo American (40%) bought out Samancor's minority shareholders and took it private. In 2005, BHP Billiton sold the chrome operations to the Kermas Group for US$469 million, separating the Steelpoort assets from the Kalahari ones. The manganese stayed. When BHP Billiton demerged South32 in May 2015, seeking a simpler portfolio focused on its Tier 1 assets, the Kalahari mines transferred with it. Anglo American retained 40%.

Mamatwan commenced operations in 1963. Terrace mining benches cut into primary carbonate ore at surface, currently producing approximately 2.91 million tonnes per annum, with around 15 years of mine life remaining. Dense medium separation and sintering upgrade the ore before export through Saldanha. In August 2025, HMM signed a 10-year export agreement with Transnet.

Wessels is different in every sense. The ore lies 300 metres down, reached through a vertical shaft and two inclines, recovered by mechanised bord-and-pillar mining. What it yields carries its own geological term: Wessels-type ore, the name given to the hydrothermally upgraded high-grade manganese found across the northern Kalahari. Grade: approximately 55% Mn. Reserve: 198.8 million tonnes. The mine gave geology a word.

In December 2024, South32 shut Wessels in response to poor manganese prices, reopening in January 2025. FY2025 combined output was 2.15 million tonnes at a realised price of US$3.71/dmtu FOB. When prices do not run, the underground operation feels it first. Ninety-four percent of South Africa's manganese exports go to Asia. The ore is priced in a market the producer does not control.

The town of Hotazel exists because these two mines do. In June 2024, NUM signed a five-year wage agreement with HMM. At the Hotazel Manganese Mines consortium level, South32 holds 44.4%, Anglo American holds 29.6%, and BEE entities hold 26%.

SA Manganese (1926) to Samancor (1975) to Billiton/Anglo (1998) to South32 (2015). The field was here before any of the names. It is still here.

Originally published on LinkedIn. This work is AI-assisted - read about how I work.

In 2013, GOLD FIELDS handed its South African gold to someone else. Kloof, Driefontein, Beatrix: unbundled into Sibanye-Stillwater. International assets, mechanised operations, growth profile. That was the direction. It kept one SA mine.

South Deep sits near Westonaria, 45km south-west of Johannesburg. BaFokeng and BaKwena Sotho-Tswana people built stonewalled settlements here from at least the 15th century. Mzilikazi's forces depopulated the region in 1823. The Voortrekkers found land that looked empty. It was not. The workers named South Deep's solar plant Khanyisa. In Setswana: Light Up. The prior claim travels in the language.

Gold Fields was founded in 1887 by Cecil Rhodes and Charles Rudd. Johannesburg Consolidated Investments was founded two years later by Barney Barnato. Two of the oldest houses on the Witwatersrand. South Deep came through the JCI line: developed by Western Areas before Placer Dome of Canada took a 50% joint venture in 1999. Barrick acquired Placer Dome in 2006 and sold the stake within a year. Gold Fields paid US$1.525 billion for that 50% in 2006, then bought out Western Areas in 2007. Total: US$2.5 billion. Not a conventional mine.

The Upper Elsburg Massives reef runs up to 120 metres thick. Every other mine on the West Wits Line works reefs averaging less than two metres. The ore sits 2,800 to 3,300 metres down, by design. The seventh deepest mine in the world. At those depths, rock temperatures reach 50 degrees Celsius before refrigeration. The reserve stands at 28 million ounces. South Deep was built for machines. Gold Fields stopped conventional mining in 2008. It had always been the plan. It took twenty years to execute.

Transitioning from labour-intensive to mechanised mining was always going to be difficult. The machine version of South Deep needed fewer people, differently skilled. Targets were missed for nearly a decade. The 2017 Rebase Plan reset the programme. Westonaria is a town where mining accounts for 52% of economic activity.

The model eventually worked. Gold Fields posted 309,000 ounces in 2025, 16% higher year on year, and generated R9.1 billion in free cash flow. Khanyisa provides 24% of the mine's electricity and saves R123 million a year. The life of mine runs to 2101. South Deep is ramping up.

Every other South African asset Gold Fields held, it sold or unbundled. South Deep it kept. It invested in new ore reserves south of a major fault, in renewable power, in depth. That is not an exit position.

When Gold Fields walked away from the conventional Witwatersrand in 2013, it held on to one thing. The thing it had spent twenty years and US$2.5 billion believing it could finally make work.