Da Silva’s Gold Papers: Sibanye's plan to flood the richest reef on earth - at best gold price ever

In this fifth instalment of his Gold Papers series, capital markets strategist Dr Duarte da Silva examines the fate of Sibanye-Stillwater's Kloof mine — once the highest-grade gold mine on earth, now declared a single year of remaining life, at a gold price within a fifth of its all-time high. Da Silva sets Kloof's numbers against two cautionary tales an hour's drive apart — Gold Fields' South Deep, rescued by two decades of patient investment, and Blyvooruitzicht, abandoned and liquidated — before showing three rival companies proving, with their own cash, that deeper and cheaper gold from the same reef still pays..By Duarte de Silva*.T H E C A T H E D R A LI. The Reef That Made a CountryI have stood on the West Wits line and watched a headgear that once lifted the wealth of a nation stand against the sky like a gallows. I know what these reefs were, and I know what they cost. Men went down into heat that would kill you in an afternoon, into ground that detonates without a sound to warn you, three kilometres beneath the veld, and they brought up the richest gold the planet has ever surrendered.Kloof was the proof of it. A first shaft sunk at Venterspost in 1934, on the ground between Westonaria and Carletonville where the basin tips its richest reef toward the day. Venterspost, Libanon, Leeudoorn, Kloof: four old mines folded into one in the year 2000, and behind them more than seventy million ounces hauled to grass across the lifetimes that followed. A reserve grade, in the middle 2000s, of nine and seven-tenths grams to the tonne when the great mines of the world fought over five. A life-of-mine head grade that touched fourteen, and an average of thirteen and a half grams to the tonne sustained across the twenty-six years from 1968. For a generation it was the highest-grade gold mine in the country, and men called it, without embarrassment, the richest on earth, and those who said it were not boasting.The Ventersdorp Contact Reef runs beneath that ground from twenty-five hundred metres down to thirty-seven hundred, a ribbon seldom thicker than a man’s forearm, dipping away into the dark, and we chased it down through five shaft systems because it paid to chase it. At its height this single mine carried close to fifteen thousand people on its books. Nobody ever called it a marginal asset. It was a cathedral of an orebody, and a country was built on ground like it.Gold today trades above four thousand six hundred dollars an ounce. It touched five thousand five hundred and ninety-five in January, the highest price in the history of the metal, fell away from that peak by nearly three tenths by midwinter, and has spent the past month climbing back, up fourteen per cent in four weeks and better than a third over twelve months. Read the number again. Four thousand six hundred dollars, richer than any price this metal ever reached before this year, and near enough double the price at which most of the basin’s reserves are booked. And at that price, with that reef beneath us, the plan on the table is to let Kloof fill with water.Not in a single decision anyone could be asked to answer for. In instalments: a shaft here, a writedown there, a line in a technical statement, a slide at an investor day. That is the matter of these pages, and the question running under all of it is whether the people who own this reef have grasped what is being settled in their name while the price does the opposite of what the plan assumed.T H E V O C A B U L A R YII. Three Words: Endowment, Resource, ReserveEverything turns on three words, and in ordinary use they blur into one another.The endowment is all the gold the rock contains: every ounce laid down when the world was young, whatever the price, whatever the depth, whether or not a living man can reach it. The resource is the slice of that endowment a geologist will certify as having reasonable prospects of one day being mined, at an assumed price, with foreseeable access. The reserve is the smaller slice inside the resource that today’s mine plan can actually take, safely, at the price the company has chosen to assume.Endowment is geology. Resource and reserve are decisions taken about geology.And those decisions rest on a price that is required to look backward. Under the codes the industry reports against, the reserve price approximates a trailing two-to-three-year average. The intention is prudence, and in a flat market it serves. In a market climbing as fast as this one it guarantees that the assumed price trails the real one by years. The cut-off grade follows the same rule: hold the price down and the pay-limit rises, so ground that was ore last year becomes waste on paper this year without a drill being turned. None of that reflects what is in the rock. All of it reflects what the model was told to assume.Follow what that machinery has done at Kloof, and follow it slowly, because the movement is easy to miss inside a results pack. At the end of 2020 the mine carried resources of thirty-two and nine-tenths million ounces and reserves of four and seven-tenths million, with a life that ran into the 2030s. Five years later the declared resource is four-point-four million ounces, the reserve is one hundred thousand, and the life is a single year..Mining did not do that. Kloof in its curtailed state lifts barely a hundred thousand ounces a year, and depletion cannot account for the loss of an empire. The resource shrank because ounces were excluded from it. The 2023 closure of No. 4 shaft alone took six-point-one million ounces off the books. The ground below 35 level went when 7 shaft was shut. The greater part of the high-grade Ventersdorp Contact Reef was removed for rock-engineering reasons, set aside as unreachable by the method the company is willing to use. A further one-point-four million ounces of reserve came down in a single declaration in February of 2026 — three weeks after gold printed the highest price in its history.Thirty-three million ounces to four. The gold did not move. The assumptions did.Kloof is not alone in that ledger. Across the whole South African gold book, in that one year, declared resources fell by a quarter, to thirty-six and a half million ounces, and reserves to nine and four-tenths — and at the operating mines themselves, stripping out the development projects, the resource fell by a third. A single annual planning cycle took a quarter-century of declared endowment off the country’s premier basin, and almost nobody outside the room remarked on it.Exactness matters more than emphasis here, so take the remainder carefully. The four-point-four million ounces still on the books is not money in a vault. At today’s price it is something near twenty billion dollars of contained metal, and the twenty-eight and a half million ounces struck from Kloof’s resource since 2020 is a hundred and thirty billion more. Contained metal is not recoverable value, though, and most of that gold cannot be taken at an acceptable cost in human risk by the method we have. What it truly is, then, is an option: a standing call on the deepest, richest ground in the basin, in a gold market that has re-rated and shows no sign of turning back. Flooding tears the option up to save a one-time bill.Set against the country’s whole inheritance, the scale of it becomes clearer, and the arithmetic has to be done carefully because it is easy to get wrong. On Handley’s reconstruction the Witwatersrand still holds on the order of forty-eight thousand tonnes of gold — and that is a total across every category, not an underground figure with the dumps added to it. Inside it sit some sixteen hundred tonnes in the surface tailings, five and a half thousand lost in mining and transport, ten thousand in pillars and remnants, eleven thousand within the mining leases, and nearly twenty thousand identified outside them by exploration that predates the reporting code. Kloof’s four-point-four million ounces is a hundred and thirty-seven of those tonnes; the resource written off since 2020 is the better part of nine hundred.That estimate is contested, and it should be carried with its counterweight rather than quoted alone. Hartnady’s Hubbert-peak analysis puts remaining economically recoverable reserves nearer three thousand tonnes and calls the basin ninety-five per cent exhausted. The two figures differ by more than an order of magnitude, and they differ because they measure different things: total gold remaining in the rock against the slice of it a mine plan will take at an assumed price. Which is the distinction this paper has been about from its first page. The gap between forty-eight thousand tonnes and three thousand is not a dispute about geology. It is the size of the decision. The basin is not being run down ounce by mined ounce; it is being signed away in annual increments, under a competent person’s signature, while the price argues for the opposite.I N T H E I R O W N W O R D SIII. What the Company Has SaidThe chief executive has been candid about the method, and it deserves to be quoted rather than characterised. Asked about Kloof at the Capital Markets Day in June, he explained that the reserve price is set through the cycle, deliberately below the spot price, and that on that basis the company does not hold long-life reserves at Kloof. The mine would be kept going, he said, only for as long as it stayed profitable and made sense.Read that the way a capital allocator reads it, and read the actual numbers, because they do not say what the through-the-cycle defence needs them to say. In the declaration of February 2026 the company priced Kloof’s Mineral Reserves at two thousand four hundred and twenty-one dollars an ounce, lifted from seventeen hundred and fifty the year before, and its Mineral Resources at two thousand six hundred and fifty. That is a thirty-eight per cent increase in the reserve price in a single year. So an orebody still carrying a six-point-seven gram resource was measured against a materially higher assumption than the year before, and the reserve was cut by ninety-four per cent anyway. The absence of reserves that follows is then reported as though it were a finding about the rock. It is not a finding about the rock. It is arithmetic doing what it was set up to do, and the setting is a choice the company remains free to revisit.And the defence weakens further the moment you put the basin’s owners side by side, because the assumption Sibanye used is not the conservative one. It is the least conservative of the lot. Gold Fields booked South Deep at two thousand dollars an ounce and declared eighty-three years. Harmony booked Mponeng — the very same Ventersdorp Contact Reef, several hundred metres deeper than Kloof’s deepest level — at two thousand two hundred and thirty-seven, and declared nineteen years while committing eight billion rand to go deeper still; in the declaration it published on 27 August it lifted the assumption to two thousand nine hundred and fifteen and the life to twenty. Pan African booked Fairview, a narrow, conventional, hand-held orebody at Barberton, at nineteen hundred and thirty-four dollars, the lowest assumption of any of them, and declared twenty-three years. Sibanye booked Kloof at two thousand four hundred and twenty-one — the highest of the four then on the table — and declared one. Sort the basin by prudence and Kloof came last; sort it by declared life and Kloof comes last still. Those two facts cannot both be explained by caution. And note what the one owner that has since gone above Sibanye on price did with the room: Harmony raised its assumption by thirty per cent and spent it on a twentieth year of mining. Sibanye raised its by thirty-eight per cent and spent it writing the reserve off..A second remark from the same session matters more than it seemed to at the time. Kloof, he acknowledged, produces far less than it was built to produce and carries a very high fixed-cost base, which is why its unit cost looks so poor. Follow that to the bottom. A deep mine is mostly fixed cost: the pumping, the cooling, the hoisting, the ventilation all run whether an ounce is raised or not. Cut the volume and the cost per ounce climbs, not because the ore got poorer but because there is less of it to carry the overhead. The all-in sustaining cost of the division sat at two thousand five hundred and nine dollars, and Kloof’s own at four thousand and eighty — more than double South Deep’s — precisely because sections had been shut and the remaining ounces were left holding the bill. By the first half of 2026 the division’s figure had climbed again, to three thousand one hundred and five dollars. Starve a mine of feed, and the rising unit cost that follows will read back as evidence the mine is finished. It is a consequence of the curtailment rather than an independent verdict on the orebody, and any fair reassessment has to begin by separating the two.And through all of it the till keeps ringing. The South African gold operations threw off twelve and a half billion rand of adjusted earnings in 2025, up a hundred and fourteen per cent on the year, as a thirty-nine per cent rise in the rand price received buried a fourteen per cent fall in ounces sold. The half-year published on 27 August 2026 went further again: a record financial performance from the South African gold operations, adjusted earnings up about eighty-five per cent, as a thirty-five per cent rise in the price received and a five per cent rise in ounces sold more than covered lower production and higher costs. A business generating cash like that at the top of the market is being wound down out of its own cash flow.Ask where that cash goes, and the picture completes itself. Very little of it returns to the reef that produced it. It services the debt taken on for ventures far from here, it funds the dividend, and it is redeployed to platinum, to palladium, to lithium, to copper, on three other continents. Any one of those may be a sound allocation on its own terms. Taken together they describe a gold business funding everything except its own future, and capital allocation has a word for an asset harvested and starved in the same breath. The word is run-off. Run-off is a decision to decline slowly while the declining still pays, and a decision is precisely what it is, not a diagnosis.T H E S A F E T Y C A S EIV. “We Will Not Risk a Life” — What the Safety Case ReachesThere is a defence here, it is a serious one, and it has to be granted every inch before anything is said in reply. The ground at Kloof is treacherous. The reef bursts without warning at depth. In 2023 a load of ballast fell down No. 4 shaft and wrecked it. Seismicity has put whole high-grade blocks beyond safe reach. No ounce of gold is worth a man’s body, and nobody writing about this industry from the surface should pretend otherwise. An executive who says he will not send people into killing ground holds high moral ground of his own, and he should keep it.The question is what that argument reaches. It plainly reaches a decision to stop mining Kloof the way Kloof has always been mined. It does not, by itself, reach a decision to flood the shafts.Those are two separate acts, and the second does not follow from the first. Ceasing unsafe stoping is right, and nothing here argues against it. Drowning the workings, writing twenty-eight million ounces to nothing and foreclosing every method that might one day reach that gold is a different order of decision altogether, irreversible, and the safety case does not license it on its own.Safety is a sound reason to stop mining Kloof the old way. Whether to drown it is a separate question, and it has not been separately argued.There is a principle that ought to govern choices of this shape. Where a harm is irreversible and its full reach unknown, you do not take the irreversible step in order to bank a certain saving against an uncertain loss. Flooding is irreversible. The seismic risk of mining is not; it is managed every day, in mines around the world, by spending money and engineering brains on it. Choosing the irreversible act over the manageable one, and describing the choice as caution, sits awkwardly with the principle it invokes.And the duty being invoked does not stop at the headgear. It runs down the slope to the springs the dolomite will sour once the pumps go quiet, and into the dark where men turned away from the formal industry will go anyway, unventilated and unmonitored, and die. Those bodies have already been counted. At Stilfontein there were ninety-three of them, and there are believed to be more still lying in tunnels nobody has reached. A safety case that weighs only the miners a company declines to employ, and not the people its withdrawal exposes, is an incomplete safety case rather than a dishonest one. Completing it is well within the company’s power, and doing so would strengthen the company’s hand rather than weaken it.Grant the argument every inch of its ground, then, and the conclusion still does not follow. Stopping the mining and drowning the mine are different decisions, and only one of them can never be taken back.T H E C O M P A R A T O RV. The Mine Next DoorHalf an hour’s drive from Kloof, across the West Rand Fault, sits the counterexample. South Deep is the same basin and the same Witwatersrand gold, and it runs at an all-in sustaining cost near eighteen hundred and forty dollars an ounce, a fraction of Kloof’s, on a reserve of some twenty-eight million ounces at its last full technical statement and a declared life of eighty-three years. It is mechanised. In 2025 it lifted three hundred and nine thousand ounces and threw off nine billion rand of free cash, and its owner reversed two hundred and eighty-five million dollars of the impairment once taken against it. On current plans it is likely to be the last gold mine standing in South Africa. South Deep is the subject of the next paper in this series; here it is enough to mark what it proves.The reason for the gap is more interesting than the comparison suggests, and it is largely a legitimate one. South Deep works the Upper Elsburg reefs: thick, bulk-minable ground a machine can be driven into, mined at a reserve grade of only five grams to the tonne but in volume enough to pay. Kloof works the Ventersdorp Contact Reef, a tabular ribbon under two metres thick, three kilometres down, in rock that throws seismic energy like a struck bell. You cannot drop South Deep’s fleet into Kloof’s reef. The mechanisation that saved one mine does not transplant into the other.That is a reason the engineering is hard. It is not a reason the gold is worthless, and two mines in this country are proving the distinction every shift: Harmony wins the same Ventersdorp Contact Reef at Mponeng, conventionally and by hand, several hundred metres below Kloof’s deepest level, and Pan African has been winning a narrow, hand-held orebody at Barberton for a century and has just raised its life again. Narrow-reef mechanisation, meaning reef-boring, low-profile remote stoping and serious seismic management, is the problem South African mining has circled for forty years without solving, largely because it never had to. There was always cheap labour and a closer, shallower reef. Neither is true now. And rather than turn a gold price within a fifth of its record high toward the one frontier that would secure the next century of this industry, the country let its exploration spending fall for a seventh consecutive year, to seven hundred and thirty-eight million rand in 2025 — a fall of more than eighty-five per cent across three decades — under one per cent of the global budget against a stated national target of five, and went fifteen years without sinking a new underground mine at all. We stopped looking down, we stopped building down, and a good deal of the knowledge has gone quiet.South Deep is the proof that the basin yields to engineering when the will and the money are present. It was once written off as a white elephant, and patience and capital turned it into an eighty-three-year mine. The narrow reef is the harder problem, still unsolved, and whoever cracks it owns the future of deep gold on earth. At four thousand six hundred dollars an ounce that brief finally pays for itself many times over. The capital that might write it is instead being posted abroad and sideways: to a mechanised platinum project at Marikana, to palladium in Montana, to lithium in Finland, to copper in Tasmania. Every one of those may be a sound investment. Not one of them is the richest gold reef on the planet, lying under a mine the same company is preparing to flood.And the deepest loss may not be in the ground at all. The men and women who know how to sink a shaft to four kilometres, who can read a seismic network and stope a narrow reef and keep ten thousand people alive in the heat, represent a national skill built across a century, and it is being retrenched, retired and put on aeroplanes. Every shaft that closes exports an expertise the country cannot quickly rebuild. Bury the reef and you lose more than the gold. You lose the only people on earth who still know how to win it, and they carry that knowledge to the deep mines of other countries. We are not merely setting an orebody aside. We are training our successors abroad.T H E S T A K E SVI. What Flooding ForeclosesSet the gold aside for a moment and consider only the water, because the water is the part that does not forgive.These are dolomitic goldfields, and the dolomite is the trap. While the pumps run, the workings stay dry and the great underground water compartments stay where they belong. Stop pumping and the water rises in the dark, shaft by shaft, until it reaches the dolomite, and then it decants: acid, metal-laden, into the springs and streams that feed a region, while the karst above begins to fail into sinkholes that swallow roads and houses and, on the Far West Rand, have swallowed people. None of this is hypothesis. It is the written history of these goldfields, in poisoned water and collapsed ground.Against that the company carries an environmental rehabilitation obligation, with other provisions, of about fourteen billion rand, set against dedicated rehabilitation funds of roughly seven billion. A promise half-funded. The state does not appear to believe even that is enough: on the same West Rand, at Ezulwini, the company was refused authorisation to stop pumping at all, and the Supreme Court of Appeal has since held that it must keep pumping until the Minister issues a closure certificate. The watchdogs put the risk plainly, that a company refused a clean exit may reach instead for liquidation and leave the water, and the bill, to the public. Whatever the intentions behind the closure plan, that is the regulatory ground it is walking onto.The foreclosure is the part that cannot be undone. Pull the pumps and the nine-grams-to-the-tonne Ventersdorp Contact Reef behind those shafts is gone, not for a cycle and not for a generation but for good, because nobody re-enters drowned ultra-deep workings at any price. The option is not written down. It is extinguished.The encouraging part is that the company already holds a clean answer in its own hand. The West Rand Tailings Retreatment Project would reprocess some seven hundred and twenty million tonnes of the old Driefontein, Kloof and Cooke dumps, surface ground with nobody underground and no seismicity to manage, for both gold and uranium. Economics that were marginal at nine hundred dollars are transformed at four thousand, and the uranium is no by-product footnote: the group has already declared a maiden uranium reserve at Cooke into a market lifted by the nuclear build-out. None of this is theory. It is the model DRDGold, half-owned by the very same company, has run profitably for years, drawing gold from old tailings at recoveries the economics now carry comfortably. Reprocess the surface, and let the easy, bloodless ground throw off the cash that holds the deep option open.The water itself can move from the largest line of the liability to a line of revenue. A deep mine on the West Rand pumps and treats millions of litres a day merely to stay alive, and in the most water-starved economy on the continent, treated mine water is a utility with buyers rather than a curse to be abandoned. The same pumping that keeps the reef’s option alive can supply a thirsty region. Walk away and the gold, the uranium and the water go together in a single decision. Stay, and each of them pays something.T H E M E C H A N I S MVII. How a Reef Is Written Off Without a VoteThere was never a moment when anyone announced, in plain words, that South Africa was giving up its richest gold. It arrived in pieces, each one small enough to read as housekeeping.In September 2023 a Section 189 notice went up on No. 4 shaft after the ballast incident, and the union complained that it learned of the process only after the stock-exchange announcement had been published, with workers emailed in the middle of a meeting. The market’s verdict that day was a share price up nearly four per cent. By December the consultation had closed and five hundred and seventy-five people were retrenched, with gold then near two thousand dollars. In the middle of 2024 a second process folded No. 4 into a list of end-of-life shafts, and the southern African workforce fell from eighty-one and a half thousand at the end of 2022 to seventy thousand. In February 2026 the formal writedown appeared in the reserves declaration. In May the annual meeting came and went. In June the Capital Markets Day put the gloss on it. In August the half-year update named what is being done to the richest reef on the continent in four words: planned production rebasing.Consider what kind of process each of those actually was. The Section 189 is a consultation about jobs, conducted under labour law, and even there the workers felt it arrive as a thing already settled. For the matter that weighs most, the sterilisation of a national endowment and the arming of a perpetual water liability, there is no process at all. No shareholder votes on a life-of-mine. Nobody approves a reserve writedown. The owners and the country were never asked, because the rules do not require that they be asked. At the one meeting in the year where the owners convene, the flooding of the country’s richest reef was not an item to be put to a show of hands.The day the writedown was announced, the share price went up.So let the criticism be precise, because precision is what makes it useful. The company hid nothing. It filed the notices, it published the statements, and a competent person signed his name. The problem is not concealment. It is proportion. A decision that floods the best ground in the basin and commits the dolomite for centuries requires less formal process than an amendment to the remuneration policy. That is how something this large passes: not in the dark, but in plain sight, dressed as a quarterly update, while the market applauds the falling cost.And here is the part most worth reopening. The judgement that Kloof was finished even at high gold prices was formed in 2023, when gold sat near two thousand dollars. The price has since more than doubled, and touched nearly three times that level in January. The conclusion has not moved. A view reached at less than half today’s price is being executed at today’s price, and nobody has been asked to test it against the market as it now stands. That is not misconduct. It is the ordinary inertia of a plan nobody owns the job of re-examining, and it is exactly the kind of inertia a new chief executive is well placed to interrupt.T H E C U S T O D I A NVIII. The Inheritance, and the Man Who Now Holds ItThere is a temptation to make this a story about a villain, and it should be resisted, because there is no villain in it.Sibanye-Stillwater exists because one man declined the verdict now being passed on Kloof. In 2012 Neal Froneman, then running Gold One, set out to buy a single mine, Driefontein, from a Gold Fields that wanted its oldest South African gold off its books, and came away, when the tax arithmetic turned against a piecemeal deal, leading the unbundling of the lot: Kloof, Driefontein, Beatrix, assets the market had already stamped high-cost and end-of-life. He turned them around. An eightfold rise in market value and forty billion rand returned to shareholders later, he had been proved right. They called him Mr Fix-It. The founding story of this company is a man who looked at the words end of life stamped on Kloof and took the label off.He did not stop at the balance sheet. Froneman went on to chair the World Gold Council, the body that authored the Responsible Gold Mining Principles, and to co-lead the campaign of business against the criminal economy in South African mining, the economy that floods into the vacuum each time a major walks away from a reef. Responsible gold, and the fight against the men in the holes, were by his own choosing his ground. That is the inheritance, and it is worth holding in view.Froneman retired on the last day of September 2025, and his successor became chief executive the next morning: a geologist by training, a man who reads rock for a living. The writedown was declared on his watch and the Capital Markets Day was his. It is worth being careful about what that does and does not mean. He did not commission the 2023 assessment. He did not order the curtailments that hollowed out the unit cost. He inherited a reserve-price convention the whole industry reports against, and a plan that arrived on his desk already moving, with three years of momentum behind it. The first year of any tenure is the year in which a chief executive has the least room and the most pressure to keep the machine running as he found it.Which is why the second year matters more than the first, and why this is written now rather than afterwards. A geologist is the one kind of chief executive who cannot be told what is in the ground, because he already knows. That is an advantage, not an accusation. The question put here to the office, and to the man who holds it while the choice is still in his hands, is a plain one: is a plan formed at two thousand dollars an ounce still the right plan at four thousand six hundred, and if it is not, who is better placed to say so?The question is not confined to Kloof. Beatrix is carried at six years, Driefontein at eleven, and the Capital Markets Day already describes the future of South African gold as a shallow, surface business, with the deep reef written out of the story. Kloof is not an exception being managed.Kloof is the template. So it is fair to ask, as any owner should, who gains from the template and who carries it. The shareholder banks a quarter or two of harvested cash and a falling cost line. The plan lifts a hard problem off the executive floor. The state inherits the water. The retrenched miner inherits the queue, or the illegal shaft. The country loses its access to the reef. Add that ledger up and ask whether it reads as value created.There is a name for an enterprise that ends up taking a profitable, irreplaceable asset, declining to invest in it, harvesting its last easy cash and handing the cleanup to the public. It is not a mining house by then; it is an estate being wound up. Sibanye was built to be the opposite, the house that saw value where others saw an ending. Nobody in that boardroom set out to preside over the managed extinction of the reefs the company was founded on, and that is exactly why the decision is worth taking again while it can still be taken.T H E P R O P O S A LIX. The Right Return, the Right WayA word, finally, to the people who actually own this. Not the board. The owners.The largest shareholder in Sibanye-Stillwater is the Public Investment Corporation, with about a fifth of the company. The Public Investment Corporation manages the Government Employees Pension Fund, the retirement savings of South Africa’s own public servants. Institutions hold some eighty-five per cent of the register between them. So the capital being redeployed out of South African gold and into Montana and Finland is, in large part, South African money: the pensions of the very state whose ground is being set aside. It compounds a problem this series named at the outset, the 2022 decision to let our pension funds send forty-five per cent of their capital offshore, up from thirty. We loosened the law so the savings could leave, and we are now watching a company those savings own send the money abroad while the reef beneath it is made ready for the water. The capital formation did not fail by accident. It was legislated for.To those owners the argument is not against returns. Gold is above four thousand six hundred dollars, the cash is real, and nobody here is asking anyone to mine at a loss for sentiment. The argument is about the way the return is taken. There is a path that earns it cleanly, and it is the company’s to walk.Hold the option. Do not flood the shafts. Keep them dewatered and breathing, and turn the rising mine water from the largest line of the liability into a revenue line, in a country dying of thirst.Let the surface pay for it. Reprocess the Driefontein, Kloof and Cooke tailings for gold and uranium, clean ground with nobody underground, at recoveries the deep mine could never match, and fund the deep option out of the dumps. It is not a theory: DRDGold, half-owned by this same company, has just reported eleven point two billion rand of revenue and six point five billion of operating profit doing exactly that, and Pan African Resources runs the surface and the deep together, in one company, and is worth more than three billion pounds for it.Incubate the method. Make Kloof the national test bed for the narrow-reef mechanisation this country set aside, and own the engineering no miner on earth has yet been willing to crack. That is the difference between a run-off and a renaissance.Formalise where you withdraw. Where the major retreats from remnant pillars and secondary reefs, do not leave a vacuum for the criminal economy to fill. License the artisanal miner who would otherwise go down illegally and die. Decriminalise the worker, prosecute the launderer who feeds on him, and route the ore through a real plant, where recovery runs to several times what a back-yard circuit manages, instead of the mercury that takes a third and poisons the river. Peru has spent a decade failing to write that law, and we can see exactly where it went wrong. We could write it better. The fifth of these papers set out how.Do that, and Sibanye becomes the house that invented South African gold’s second life rather than the one that managed its close, and it earns a pioneer’s multiple instead of a liquidator’s discount. None of it is charity. It is the most valuable optionality on any balance sheet in this market, and it is about to be flushed underground.I have watched this industry set aside its inheritance before, and the defence afterwards was always that nobody could have known. That defence is not available here. The reconciliation that took thirty-three million ounces down to four was signed, dated, lodged and published, and anyone who needs to read it can. Which is the reason for writing now, while the decision is still open, rather than afterwards.A country that gives up the best of what it owns at the very top of the market is not managing a decline so much as declining to manage. The right return is there for the taking, and there is no argument in these pages against taking it. The question is only how it is taken, and how much of the future the taking has to cost.Everything in this plan can be decided twice. The flooding can only be decided once..THE PROOF AND THE RUIN.South Deep is what commitment bought: two decades of nerve that turned the most mocked write-off in South African gold into the longest-life mine in the country, and, in 2025, into a nine-billion-rand-a-year cash machine whose impairment its owner has begun to reverse. Blyvooruitzicht is what abandonment bought: a liquidation, a poisoned town, and a seam so plainly unfinished that a third company mines it today. This paper walks both endings in full, because Kloof stands between them right now..The first of these papers left Kloof at the edge of its decision. This second walks the two destinations already on the record: the same basin, an hour’s drive apart, and opposite outcomes..A paper that argues from precedent owes its reader both endings in full, and this one gives them in the companies’ own numbers, restated to the latest filings: the capital committed and the cash it now returns at South Deep, the liquidation record and the water at Blyvooruitzicht, and the surface machine that DRDGold, half-owned by Kloof’s owner, has just reported its best year on. Nothing here requires a forecast. Both destinations have already happened, on the same reef Kloof’s owner now proposes to leave to the water, and the reader is asked only to hold them side by side while the third decision is still open..T H E F R A M EI. The Fork in the RoadKloof stands where the last paper left it: the richest reef on the continent, carrying a single year of declared life at a gold price within a fifth of the highest ever recorded, its shafts being prepared for the water. In February its owner went further, three weeks after gold touched five thousand five hundred and ninety-five dollars, and wrote the mine down to almost nothing, cutting the declared reserve from 1.6 million ounces to 0.1 million, closing 7 Shaft and excluding the ground beneath it. The argument in the last paper was that finished is a conclusion drawn from assumptions rather than a finding about the rock. This paper tests that argument against the record, because the choice Kloof now faces has already been carried to its conclusion twice, in the same basin, within an hour’s drive, and both endings can be walked before the third is settled.One mine was committed to. Its owners spent twenty years and a fortune, were mocked for every rand of it, and turned a write-off into the longest-life gold mine in the country. The other was abandoned. Its owners took the profitable decades, sold the liability down the chain and walked away, leaving a poisoned ghost town and a seam of gold so plainly unfinished that a third company is mining it today. South Deep is the proof. Blyvooruitzicht is the ruin. Lay them side by side and you are looking at the two destinations of the road Kloof is on, photographed in advance.P A R T O N E · T H E P R O O F — S O U T H D E E PII. The White ElephantBegin with the humiliation, because the lesson is worthless without it. For the better part of two decades South Deep was the most expensive failure in South African mining, and the industry knew it by the same two words. Gold Fields paid something near twenty-two billion rand across 2006 and 2007 to own the mine outright, buying out the half it did not hold from the wreck of the old Placer Dome venture and then mopping up the rest of Western Areas, and fed in the better part of another ten billion trying to make it work.It would not work. The grand promise of seven hundred and fifty thousand ounces a year was cut to five hundred, and the mine missed even the smaller number year after year while the cost line climbed. Under Nick Holland, who ran Gold Fields for thirteen years to 2021 and owned this asset’s worst stretch, South Deep was through the first half of 2018 losing on the order of a hundred million rand a month, and nearer a hundred and sixty across the strike-hit year. The company wrote four point eight billion rand off its carrying value after tax that year — six and a half billion before it — and that August moved to cut as many as fifteen hundred and sixty jobs. The National Union of Mineworkers struck for six weeks. The mineral resources minister, Gwede Mantashe, accused the company of treating consultation as a box to be ticked. The market had long since reached its verdict and never softened it: a white elephant, the hole into which a great mining house poured a fortune and pulled out humiliation. Roughly thirty-two billion rand, all in, sunk before the mine turned a clean profit. On the numbers as they stood that year, no asset on the Witwatersrand more plainly deserved the stamp end of life on cost..T H E C O M M I T M E N TIII. The TurnAnd then, slowly, it stopped failing. Not because anyone struck new gold, since the gold had been there since the world was young, but because the company changed how it took the gold out and refused to quit the asset at the bottom of its fortunes. From 2019 South Deep delivered its first full year of positive cash flow. It lifted two hundred and twenty-two thousand ounces that year, two hundred and twenty-seven through the worst of the pandemic, then two hundred and ninety-three, then three hundred and twenty-eight. The cash came up behind the ounces, and faster: a couple of hundred million rand of free cash in 2019 had become more than three and a half billion by 2023, and nine billion rand — half a billion dollars — in 2025..The mine that bled a hundred million rand a month became the single largest generator of free cash in Gold Fields’ South African book, and with the company’s retreat from the rest of its South African gold, effectively the whole of it. The 2025 year settled the argument. Output recovered from two hundred and sixty-seven thousand ounces in 2024 to three hundred and nine thousand, up sixteen per cent, and free cash flow tripled to nine billion one hundred and forty-four million rand. In the same results Gold Fields reversed two hundred and eighty-five million dollars of the impairment it had once taken against the mine — the write-down itself written back. The wrinkles deserve precision, because precision is the discipline of this series. The cost line is not flat: all-in sustaining costs ran near eighteen hundred and forty dollars an ounce in 2025, above the group average, and the company guides them higher still this year. The turnaround is real and it is unfinished. Its shape is not in dispute, though, and neither is its lesson. An asset the market had buried as dead was, with patience and capital, made to live and to pay. The men who bedded the recovery down, Martin Preece through 2023 and Mike Fraser from the start of 2024, inherited a cash machine the executive who endured the worst of it had already begun to build.T H E M E T H O DIV. Why Machines, and What Five Grams BuysThe thing that turned South Deep was mechanisation, and the reason it worked there is exactly the reason it cannot simply be lifted across the fault. South Deep mines the Upper Elsburg reefs, not the thin tabular ribbon of the Ventersdorp Contact Reef that Kloof chases, but thick, bulk-minable ground, in places scores of metres deep, that a machine can be driven straight into. Around 2008 the company abandoned conventional hand-held mining there altogether and rebuilt the operation around trackless, mechanised methods: low-profile drill rigs, long-hole stoping, a fleet working the orebody underground. It remains the only deep-level gold mine in the country run that way at scale, three kilometres down, on a main shaft that drops in a single hoist to nearly three thousand metres.At that depth a machine is not a convenience. It is the line between an economy and a graveyard. Mechanised mining moves far more rock per worker, and, in the part that matters most, it puts far fewer bodies into the seismic zone in the first place. Lower the number of people in the killing ground and you lower the cost and the risk in a single motion. That is how South Deep can run a reserve grade of barely five grams to the tonne, about half Kloof’s historic grade, and still pay handsomely: it makes up in safe, mechanised volume what it gives away in richness. Kloof’s defenders are right that you cannot drop this fleet into a sub-two-metre reef that bursts. They are wrong to conclude that the gold is therefore worthless. The two statements are not the same, and the case for closure depends on treating them as though they were.T H E L E D G E RV. The Eighty-Three-Year Number That Tells the TruthNow to the number that gives this part its name, because it exposes the machinery better than any other. South Deep declares a life of mine of eighty-three years and attributable reserves of some twenty-eight million ounces. Kloof declares one year, and one hundred thousand ounces. Both numbers are produced by the same instrument, the through-the-cycle reserve price set deliberately below the market, and the gulf between them is not geological. It is a difference in what each owner decided to assume about the future.Watch the instrument work, because in the 2025 declarations it stops even pretending. Gold Fields books South Deep’s reserves at an assumed gold price of two thousand dollars an ounce, lifting that to two thousand three hundred only for the wider resource. Sibanye-Stillwater books Kloof’s reserves at two thousand four hundred and twenty-one dollars, and its resources at two thousand six hundred and fifty. Read those two lines again. Kloof’s owner used the more generous gold price of the two — four hundred dollars an ounce above the price Gold Fields used — and still wrote one point four million ounces off the mine, leaving a hundred thousand ounces and a single year. The conservative-price defence collapses here. The gap between eighty-three years and one year cannot be a difference in prudence, because the shorter life was declared at the higher price. Nor is it a difference in grade: Kloof’s remaining measured and indicated resource still runs at six point seven grams a tonne, above South Deep’s reserve grade of about five. And spot, as this is written, is above four thousand six hundred dollars — nearly double the price Sibanye used to decide the gold was not worth taking out — and three weeks before that declaration was signed, gold printed five thousand five hundred and ninety-five dollars, the highest price in the history of the metal. One company used a conservative price to underwrite eighty-three years of mining. The other used a less conservative one to support closing the richest reef on the continent. The number was never the gold. The number was the decision.South Deep books eighty-three years at two thousand dollars an ounce; Kloof books one year at two thousand four hundred. The mine with the shorter life used the higher price. The gap is not in the rock, and it is not in the prudence. It is in the decision.T H E O P T I O NVI. The Frontier We Will Not FundThere is one real gap in the South Deep parable, and it must be named because the industry hides behind it. The mechanisation that saved South Deep solves the thick-reef problem, not the narrow one. The Ventersdorp Contact Reef at Kloof, the Carbon Leader at depth: these are ribbons, sometimes under a metre wide, three and a half kilometres down, in rock that throws seismic energy like a struck bell. No machine yet built mines them at scale. The problem has a name, narrow-reef mechanisation, and it has been circled for forty years.It was not circled for lack of imagination. A little over a decade ago AngloGold Ashanti spent some thirty to forty million dollars on a reef-boring programme it summed up in a slogan, all the gold, only the gold, all the time, safely, built to bore a reef-width hole, gather the chips in a sealed system and backfill the void to calm the seismicity. It produced its first gold in 2013, out of a shaft pillar previously written off as too dangerous to enter, and the company spoke of adding thirty million ounces to its reserves and taking a fifth of its South African output that way within five years. It never reached scale. AngloGold lost patience with deep South African gold altogether, sold the last of it, Mponeng among the assets, to Harmony in 2020, and the reef-borer went quiet with the company that had built it. The frontier sits exactly where it was left: unsolved, and unfunded.And mark what is not being funded. At four thousand six hundred dollars an ounce the brief to crack the narrow reef pays for itself many times over, set against ground that has just been written down rather than mined: the four million ounces of measured and indicated resource Sibanye still declares at Kloof are worth some eighteen and a half billion dollars of contained metal at spot, and that is only what survived the write-down — a year earlier the same declaration carried fourteen and a half million ounces, better than sixty-seven billion dollars’ worth, and the resource struck from Kloof since 2020 runs to a hundred and thirty billion. The country’s answer has been to let exploration spending fall for a seventh consecutive year, to seven hundred and thirty-eight million rand in 2025, a fall of more than eighty-five per cent across three decades; to hold under one per cent of global exploration budgets against a stated national target of five; to go fifteen years without sinking a new underground mine at all; and to ship the cash its gold still throws off out of the basin entirely. South Deep proves the deep basin yields to engineering and patience. The refusal to build the narrow-reef version of it is the clearest evidence we have that finished was never a fact of the rock. It was a decision not to write the cheque, and the same decision is now on the table at Kloof, with the price arguing harder against it every month.P A R T T W O · T H E R U I N — B L Y V O O R U I T Z I C H TVII. Happy ProspectDrive further west, deeper into the Far West Rand, and you reach the other ending. Blyvooruitzicht is old Afrikaans, and it means happy prospect, the glad thing seen ahead. The mine took the name from the farm, and for a generation it earned that name as no mine on earth has earned one before or since. First gold in 1942; the first mine sunk on the West Wits line; and within a generation the most profitable gold mine in the world, working the Carbon Leader Reef at in-situ grades the great mines of the world could not touch. By the turn of the 1970s they said, in the company town that grew up around the headgear, that it had the highest earnings a head of any place in the world, and the men who said it were not boasting. This is the mine that abandonment finished, and it is not a forecast. It is a place, with an address, and people still living in its wreckage..T H E C H A I NVIII. Pass the ParcelWatch the asset travel down the food chain, because the journey is the warning to Kloof. Rand Mines built Blyvoor and worked its golden decades. By the late 1990s it had passed to DRDGold. In 2012 DRDGold sold it on again, down to a smaller company called Village Main Reef, at a price engineered to be small: the claims changed hands for some hundred and fifty million rand settled not in cash but in shares, and the seventy-four per cent equity in the mine itself carried a nominal consideration of a single rand, one rand, a leg of the deal that in the event was never even completed before the whole arrangement collapsed into dispute. The reason the price was cut is the part to hold onto. On the seller’s own account, it was lowered so the buyer could top up a rehabilitation fund that held thirty-four and a half million rand against a requirement of seventy-seven million. Read that twice. The asset changed hands cheap precisely because the cleanup it carried was worth more than the gold left in it.This is not a freak event. It is a recognised move, and the people who watch South African mining have named it. A large house mines the profitable years, then sells the ageing mine, with its mounting closure liability, to a smaller one that pulls the last feasible ounces and folds before the rehabilitation is ever done. The environmental campaigner Mariette Liefferink calls it pass the parcel, and the under-resourced buyers scavenger companies. The pattern is all over the record.Pamodzi Gold, a junior that had rolled up President Steyn, Orkney and the East Rand operations at Grootvlei, collapsed into liquidation in 2009 with some fifteen thousand jobs at risk, and the viable ounces were carried off by the majors while the shell and its obligations went to the liquidators. Mintails, a surface-retreatment outfit on the West Rand, went into final liquidation in 2018 leaving a rehabilitation bill its own documents put between three hundred and three hundred and thirty-six million rand against a provision of twenty-five and a half — and which was reported at four hundred and sixty million against twenty-eight by the time the final order was granted. DRDGold had itself once forecast Blyvoor profitable past 2030. Some eighteen months after the sale, seller and buyer alike were calling it finished.T H E A B A N D O N M E N TIX. The Morning the Lights Went OutAugust 2013. Village Main Reef stopped funding the mine, the board voted it into provisional liquidation, and then began the spectacle that tells you everything about how the parcel is passed. Neither the company that had sold Blyvoor nor the company that had bought it would admit that it was the owner. While the lawyers argued the point, the people on the ground fell through the gap. More than a thousand workers, by most counts close to seventeen hundred, were left in a place that was neither open nor shut: not formally retrenched, and so owed no retrenchment pay, and not employed, and so paid nothing at all. No package. No wage. A gate, a dispute, and a liquidator.Around that gate stood a company town of some six thousand people, a town whose water, electricity, sanitation and security the mine had always quietly carried, because that is what a company town is. When the mine stopped, the bills stopped, and the place began to die in slow motion. By 2017, lawyers documenting the collapse found roughly three-quarters of the village without work and most households short of the money for enough food. It took a court order in 2015 to get the village’s water reconnected, and a second, a year later, to force a plan for keeping it on, over an account nobody would own. The richest gold mine the world had ever seen had become a settlement that needed a judge to keep its taps running. A rights group called it, without exaggeration, a catastrophe, and laid it exactly where it belonged: at the meeting point of private abdication, a weak closure law, and a state that would not enforce the one it had.T H E A F T E R M A T HX. What the Water RemembersSet the gold aside and consider only the water, because the water keeps no one’s secrets and forgives nothing. Blyvoor sits in the Far West Rand, on dolomite, above the Wonderfonteinspruit, a catchment already notorious internationally for the uranium its gold mines left behind in the streams. These are the goldfields where the state once authorised pumping the dolomitic water table down by as much as a kilometre to keep the mines dry, and where the disturbed ground answered with sinkholes: well over a thousand of them across fifty years, one of which opened beneath a house at Blyvooruitzicht one night in 1964 and swallowed a family of five. When the pumps stop and the water climbs back through the dark, it returns acid and metal-laden, and it decants to the surface to sour the springs that feed a region.None of that bill vanished when the companies did. It moved to the state, which carries a national rehabilitation liability for derelict and ownerless mines last estimated at some forty-nine billion rand across roughly six thousand sites — a figure now a decade old and regarded on all sides as too low. Set against it, the department’s 2026/27 budget for rehabilitating those mines is one hundred and forty-one million rand. At that rate the arithmetic is not a programme; it is a gesture, and it runs for three and a half centuries. Blyvoor is one line in that ledger. The gold was privatised through its golden years and the poisoned water nationalised at the end, and the human vacuum was filled the way it always is. The shafts nobody would own filled with illegal miners, the zama-zamas, working the remnant gold by hand and by stolen machine in ground nobody was keeping safe. The town was stripped for its copper and its bricks. The turf wars over the holes ran up a body count the residents counted in the dozens. It is the same vacuum, filled by the same economy, that killed ninety-three men underground at Stilfontein, with bodies still believed to be lying in tunnels nobody has reached. When a major walks away from a reef, the reef does not fall silent. It fills with the desperate, and then it kills them out of sight.T H E E V I D E N C EXI. The Gold Was Still ThereAnd here is the fact that ought to end every argument about whether any of this was geology. The gold was never gone. A new company, Blyvoor Gold, bought the principal shaft out of the wreckage, reopened it around 2019, and from 2022 has been mining the same orebody again. It now employs close to two thousand people, is hiring a thousand more, and takes better than nine in ten of those new hands from the communities the liquidation stranded. It declares proved and probable reserves of five point one million ounces at five and a half grams a tonne, a thirty-four-year reserve life on the Carbon Leader alone, and a resource base it describes as running to a century. Everything the liquidation swore was finished is being hauled to grass today, by a third owner, at a profit.Be sober about that second life, because it is thin and hard-won, and it carries its own warning. The mine is small: about twenty-five thousand ounces a year against the hundred and fifty thousand it talks of building toward. A seismic event south-west of Carletonville in January 2025 brought down ground at the mine, killing one man and injuring twenty; a second miner died there that June. Its flagship plan to list in New York collapsed in November of that year, when the deal meant to carry it there — approved by regulators and shareholders alike — was terminated by mutual agreement, and the shell that was to take it public dissolved and handed its investors their money back. A year on it is still unlisted. None of that touches the central fact. The ore two listed companies swore was worthless, that they shut, flooded and walked away from, is paying a third company to dig it back up. Finished was never the rock’s verdict. It was a line in a model, the model was wrong, and a town paid for the error in wages and water and lives while the world waited for someone to come back for what had been left.T H E S E C O N D H A R V E S TXII. The Irony in the DumpsThere is a final turn to the Blyvoor story, and it bears directly on the logic of retreat. The company that passed Blyvoor’s underground liability down the chain, DRDGold, did not vanish. It pivoted. Having decided that deep South African gold was someone else’s problem, it rebuilt itself entirely around surface tailings, reprocessing the old slimes dams of the East and Central Rand through its Ergo operation, and the West Wits dumps around Carletonville through Far West Gold Recoveries, for the gold the old mills had left behind. It worked spectacularly, and it has never worked better than in the year just reported. In the year to June 2026, announced on the nineteenth of August, DRDGold turned over eleven billion one hundred and fifty-nine million rand, up forty-two per cent, and made an operating profit of six billion four hundred and fifty-two million, up eighty-three. Headline earnings rose eighty-nine per cent to four and a quarter billion rand. Free cash flow rose eighty-five per cent to two billion two hundred and sixty-six million. It poured a hundred and fifty-five thousand five hundred and seventy-seven ounces, above its own guidance, at an all-in sustaining margin of fifty-three per cent against thirty-nine the year before. It declared a final dividend of a hundred and twenty cents a share, three times the previous year’s, in its nineteenth consecutive year of paying one, and closed the year with two billion seven hundred and seventy million rand in the bank and no debt at all — while spending three and a half billion rand of capital, most of it on Vision 2028, a ten-billion-rand programme to lift throughput from two point one five million tonnes a month to three million. Nobody underground. No fresh seismicity to manage.Hold that beside Kloof, because the ownership closes the circle. DRDGold is today a little over fifty per cent owned by Sibanye-Stillwater, the company that owns Kloof and is preparing its shafts for the water. So the surface-retreatment model Sibanye half-owns, and that prints money on old West Wits tailings a few kilometres from Kloof’s own dumps, is a clean, bloodless and profitable answer sitting in the group’s own hands while the deep reef is made ready for flooding. The company that left the deep ground grew rich on the surface.And on the day those results landed, DRDGold’s chief executive, Niël Pretorius, made the point this series has been making, in his own words and about other people’s mines. Rather than let a major close an operation and pull down its plant, he told News24, “instead of closing your mine and dismantling all of your capital infrastructure, let’s repurpose it” — DRDGold bringing the capital in return for a share of the proceeds, and letting the waste pay for the cleanup. He was talking about tailings, and about counterparties in Africa and Latin America. But the sentence is the whole argument of this paper, spoken by the half-owned subsidiary of the company that is about to dismantle Kloof.I said as much to News24 the same day, and I will repeat it here, because the second half of the remark is the part that bears on Kloof. “It’s so good to see that there is life in the South African gold industry,” I told them — and then, on the share, which is up more than half over the year and rose again on the day and is still, to my eye, cheap: “to be trading at a discount to your peers that have so many more risks from a geological point of view is just something I don’t understand.” That discount is the same error as the one in the reserve statements, seen from the other end. The market marks down the producer that has no geological risk left to run, and applauds the company preparing to drown the reef that holds the most gold. In both directions it is declining to look at the rock.It is the model the next of these papers takes up in full; here it is enough to mark the irony. The cheapest gold in the country is lying on the surface in Gauteng, already mined once, and the people preparing to flood the deep reef already own the machine that mills it — a machine that has just posted the best year of its life, and whose own chief executive is publicly asking other miners not to do what his majority shareholder is doing at Kloof.S Y N T H E S I S · T H E T W O D E S T I N A T I O N SXIII. No Third DoorSo lay the two endings side by side, with Kloof’s unmade choice between them, because they are the same gold and the same decision caught at three moments of its life. South Deep is what commitment bought: a white elephant turned, over twenty patient years, into an eighty-three-year mine, a nine-billion-rand-a-year cash engine and a reversed impairment, proof that the deep basin pays anyone willing to build into it. Blyvoor is what abandonment bought: a stranded workforce, a town that sued for its own water, a poisoned catchment handed to the public, and a seam of gold so obviously unfinished that a third company is mining it while the lawyers still argue over the corpse..There is no third door marked managed decline that leads anywhere else. You either build, as South Deep was built, or you make a ruin and hand the bill to the state, as Blyvoor was made. Those are the destinations, and the basin has now demonstrated both, an hour apart, inside a single working lifetime. There is no clean way to abandon the richest reef on a continent. There is only the way that takes twenty years of nerve and the way that takes a single August.Kloof stands between these two endings right now, at above four thousand six hundred dollars an ounce — within a fifth of the highest price this metal has ever reached, and climbing again — and the market is applauding the course that led, before, to Blyvoor. The owners, and the pensioners whose savings own the company, are entitled to ask why the road to South Deep is the one not being taken, and why the answer that works on the surface is being offered to strangers abroad and not to the reef next door. If the water does finally close over Kloof, nobody will be able to say the country had not seen, in advance and in full, exactly where each road led..THE DEEPEST GOLD AND THE CHEAPEST.T H E F R A M EI. Two Roads Still OpenThe first two of these papers mapped a choice and its destinations. Paper 16 set out the decision being taken at Kloof, the richest reef on the continent carrying a single year of declared life at a gold price within a fifth of the highest ever recorded — a life since reaffirmed, and the reserve behind it cut in February 2026, three weeks after gold printed its record, from one and six-tenths million ounces to one hundred thousand. Paper 17 walked the two places that decision leads, on the same basin: South Deep, the mine that was committed to and now books eighty-three years while pouring nine billion rand a year of free cash, and Blyvooruitzicht, the mine that was abandoned and became a poisoned ruin.Now turn to where the road could lead instead, because there are three proven ways forward for deep South African gold and all three are being walked profitably, two of them within an hour of Kloof. You can go down, three hundred metres below Kloof’s deepest level and heading for more. You can go up, to the surface, and gather what the deep mines left behind. Or you can decline the choice altogether and run both, letting the safe ounces pay for the hard ones — which one listed company has been doing, profitably and by design, since 2013. All three pay handsomely with gold above four thousand six hundred dollars an ounce, within a fifth of the record it set in January. None of them is the road currently mapped for Kloof, and the company that would have to choose already owns a share of one of them.P A R T O N E · D O W N — M P O N E N GII. The Deepest Gold on EarthForty kilometres from Kloof, in the same Far West Rand goldfield, men go to work four kilometres beneath the veld. Mponeng is the deepest mine on earth, and nothing else is close. It grew out of the old Western Deep Levels complex, the shaft sunk in the early 1980s and the sub-shaft deepened in 1993, and took its present Sesotho name, Mponeng, look at me, in 1999. Its workings now reach three thousand seven hundred and forty metres below surface, on a shaft sunk to three thousand eight hundred and ninety-one, and a project under way is driving them toward four thousand one hundred.Understand what that depth means, because it is the whole measure of the difficulty. The virgin rock at that level is hot enough to kill an unprotected man in an afternoon, and the mine pumps slurry ice down the shaft to chill the air to something a body can survive. The journey from surface to the deepest face runs to the better part of an hour before a single ounce is touched. The weight of nearly four kilometres of rock bears down on every face, and the ground answers with seismic events on a scale no other mine on earth has to manage. This is the hardest gold mining done anywhere on the planet, and it is done, every shift, in the same basin we are told at Kloof is finished.T H E E X I TIII. The Mine AngloGold FledIt is worth remembering who walked away from this place, because the words they used are the words being used at Kloof now. For most of its life Mponeng belonged to AngloGold Ashanti, for a time the largest gold miner in the world, and the company, as an earlier paper recorded, that spent thirty to forty million dollars on a reef-boring programme meant to crack the narrow reef, produced its first gold from it, and then let the effort lapse. In 2020 AngloGold sold the last of its South African gold, Mponeng, the worked-out TauTona and its surface-treatment plants, to Harmony Gold, and left the country entirely.Look closely at the price, because its structure tells you how little AngloGold believed in what it was selling. The headline was about three hundred million dollars: two hundred million in cash on completion, and the rest deferred as a payment of two hundred and sixty dollars an ounce on production above two hundred and fifty thousand ounces a year, plus twenty dollars an ounce on gold mined below the existing infrastructure. Harmony itself valued the package nearer three hundred and sixty million once the deferred payments were projected. AngloGold took a modest cheque up front and a contingent tail it plainly did not expect to amount to much. Its reasoning was precisely the reasoning heard around Kloof today: the deep South African mines could not compete for capital against shallower, cheaper ounces in Ghana, Australia and the Americas, the money was better spent elsewhere, the costs would never come down. So it concluded that the deepest gold on earth was someone else’s problem, and went..T H E G E O L O G YIV. The Same Reef Kloof Is Writing OffBefore going further, fix one fact that ought to settle the Kloof argument on its own. The reef Mponeng mines, its principal producer, the ground it has worked for a generation and is now spending billions to chase deeper, is the Ventersdorp Contact Reef. It is the same reef Kloof works. It is the same reef Sibanye has just declared finished, written down to a single year of life and slated for the water.And Mponeng does not even mine it the easy way. There is no bulk-mechanised fleet here as there is at South Deep. The Ventersdorp Contact Reef at Mponeng is a narrow, tabular ribbon, won by conventional deep-level methods, men and rock-drills working a seam often little more than a metre thick, four kilometres down, in ground that bursts without warning. It is, in other words, exactly the kind of mining Kloof’s owner says cannot pay: narrow reef, ultra-deep, conventional, unmechanised. Harmony is doing it now, at a profit, on the same reef, in the same basin, three hundred metres below Kloof’s deepest level and driving deeper. When a company reports that the Ventersdorp Contact Reef is finished, the plain answer is that a rival is mining that very reef deeper than anyone has dared, and calling it the best asset it owns.A rival is mining the Ventersdorp Contact Reef deeper than Kloof has ever reached— and calling it the best asset it owns.T H E P R O O FV. Now Harmony’s Best MineFive years on, the buyer has been vindicated by the price, and the contingent tail AngloGold shrugged off has come due. Mponeng’s output climbed three years running, two hundred and thirty-nine thousand ounces in the year to June 2023, two hundred and eighty-one thousand the next, three hundred and thirty-three thousand in the year to June 2025, with the recovered grade rising alongside it from eight grams a tonne to better than eleven. It threw off roughly seven billion rand of free cash in that last year, the single largest cash generator in Harmony’s entire portfolio — a portfolio that, it is worth saying now rather than later, is not only a deep one. The deepest mine on earth had become the best mine in the company, and it has not given the title back. Through the first half of the year to June 2026 Mponeng lifted a hundred and sixty-six thousand ounces at ten point eight five grams, at an all-in sustaining cost of sixteen hundred and twenty dollars an ounce, and generated four billion eight hundred and seventy-seven million rand of adjusted free cash — up thirty per cent on the same half a year earlier, on a cash margin near half. The audited figures, published on 27 August 2026, settled the argument. Mponeng lifted three hundred and seven thousand ounces over the full year, inside guidance, at a recovered grade of ten point six seven grams — eight per cent fewer ounces than the year before, on slightly less tonnage and a slightly softer grade. And its adjusted free cash flow rose thirty-four per cent anyway, to nine billion four hundred and forty-four million rand. One mine, on the hardest ground on earth, now generates more than half of Harmony’s record group free cash flow. The group met production guidance for an eleventh consecutive year at one million four hundred and thirty thousand ounces, lifted headline earnings a share by eighty-seven per cent, and declared the highest final dividend in its history..Be exact about what did that, because it is the same lever this series keeps returning to. It was the price. At an all-in sustaining cost near eighteen hundred dollars an ounce against a price received of nineteen hundred, Mponeng was running on a margin of barely a hundred and thirty dollars an ounce as recently as the year to June 2022. The ore did not get richer by some accident of geology; the metal got dearer, and a marginal mine became a gusher. That is precisely the re-rating that has happened to the gold beneath Kloof, since the Ventersdorp Contact Reef is worth at four thousand six hundred dollars what it could never be at one thousand, and precisely the re-rating Kloof’s plan has not yet been revised to reflect. The distance between a mine harvested for its last cash and a mine declared finished and flooded is, on this evidence, a decision rather than a cost curve.T H E P O S T U R EVI. Extend, Not HarvestHere is the part that ought to settle the matter. Harmony is not merely milking Mponeng for its last easy ounces. In February 2024 its board approved a deepening of the mine below its 120 level, some seven point nine billion rand of capital, funded entirely out of Mponeng’s own cash flow at roughly a billion rand a year, to drive the workings down through a further stretch of the Ventersdorp Contact Reef and into the Carbon Leader below it, convert more than three million ounces into reserve, and extend the life of the deepest mine on earth to at least 2044 — turning a seven-year mine into a twenty-year one. The project carries a cash contribution the company puts near two and a half billion rand a year at a real gold price of one point one million rand a kilogram, and a steady-state target near two hundred and sixty thousand ounces a year, and at its most recent update, in May 2026, Harmony reported it still on schedule. In the results published on 27 August the life-of-mine was extended by a further year, to about twenty.Read that against Kloof one more time. One company, at this price, is spending the better part of eight billion rand of its own money to go deeper into the Ventersdorp Contact Reef, and books a twenty-year life on it. Another, at the same price, has written the same reef down to a hundred thousand ounces and a single year, and is preparing to open the pumps. Both are listed gold miners, both can read a spot screen, and both employ competent persons to sign reserve statements. When Harmony last declared against a lower gold price than Sibanye’s — two thousand two hundred and thirty-seven dollars an ounce against Kloof’s two thousand four hundred and twenty-one — it got nineteen years and Sibanye got one. In the declaration Harmony published on 27 August it raised its own assumption to two thousand nine hundred and fifteen dollars, well above Sibanye’s, and spent the room on a twentieth year of mining. Sibanye raised its assumption by thirty-eight per cent and spent the room writing the reserve off. The same lever, pulled in opposite directions. The single variable that differs is the willingness to commit capital to deep South African ground, and that variable is not set by geology. It is set in a boardroom, and it can be set either way.T H E A L T E R N A T I V E SVII. Every Alternative, and Still the Deep ReefThe obvious objection is that Harmony simply has nowhere better to put the money. The opposite is true, and it is the most telling part of the story. Harmony is in the middle of the largest strategic shift in its history, a pivot into copper on two other continents, and in the past year it has stopped talking about that pivot and started paying for it. In October 2025 it completed the purchase of the CSA copper mine in New South Wales for about a billion dollars, and has already run it for a part-year at eighteen thousand tonnes of copper. In November 2025 it took the final investment decision on Eva Copper in Queensland, a build of one and a half to one and three-quarter billion dollars over three years, now under construction and targeting first production at the end of 2028. It still holds half of Wafi-Golpu in Papua New Guinea, one of the world’s great undeveloped copper-gold deposits, awaiting its mining lease. Group capital spending, up nearly a third in the year to June 2025, came in at seventeen billion one hundred and six million rand for the year just ended — seventy per cent higher again. Harmony also has, inside its own portfolio, the cheap and bloodless alternative this paper is about: a surface and retreatment business that lifted two hundred and twenty-one thousand ounces last year at a forty-six per cent cash margin, bought rather than built. It has every competing call on its capital that AngloGold ever pleaded, newer and more fashionable ones besides, and a safer way to make money in its own back yard — and it is still funding the deep reef at Mponeng, because, in its chief executive’s plain words, it is the gold business in South Africa, where it all began, that is starting to show the world that gold mining in South Africa is a good business.That is the sentence to hold against Kloof. The company with the strongest conceivable alternative use for its cash, an entire copper future on two continents, looked at the deepest and hardest gold in the country and chose to spend eight billion rand going deeper into it. The company that owns Kloof is also redeploying its gold cash into platinum and palladium and lithium and copper, on three continents, and has reached the opposite conclusion about the very same reef. Both cannot be reading the price correctly.T H E C A V E A TVIII. The Necessary CaveatGrant the difficulty its full weight, because it is real and it is lethal, and the past eighteen months have made that point without mercy. This is not cheap gold and it is not safe gold. The rock bursts. A seismic event in March 2020 killed three men at Mponeng; a fall of ground killed two more in February 2025; a shaft engineering incident killed two more in May 2026; a man died at the neighbouring TauTona service shaft that July. In the year to June 2026 Harmony lost six people across all of its operations, against eleven the year before, on the lowest lost-time injury rate it has ever recorded. The direction is right, the company says so itself, and six is still six. Cooling and ventilation alone consume a large share of the mine’s electricity. The engineering consensus is that four kilometres is close to the practical floor of how deep a human being can be sent to win a narrow reef with the methods we have. Mponeng’s revival is a function of an extraordinary price, not of any breakthrough that made the deep cheap or the deep safe.All of that is true, and none of it is a reason to flood Kloof. It is an argument for the two things this series has pressed throughout. Take the gold while the price is paying for the risk. And spend a sliver of the windfall solving the narrow-reef problem, the reef-boring and remote stoping AngloGold set down, so that the next four kilometres need not cost in blood and rand what this one does — a case the fatality record makes more urgent, not less. Harmony is doing the first with conviction, and it is engineering at the margins of the second: Mponeng has moved to a five-day week on twelve-hour shifts to cut the share of a shift spent travelling, and chairlifts are being sunk at other operations to get crews to the face faster. Nobody in this country is yet seriously doing the second at the scale the problem needs. What cannot fairly be said, though, with Mponeng running every shift a few kilometres away on the very same reef, is that the deep basin is finished. It is not finished. It is being mined deeper than ever before in human history, at a profit, right now.P A R T T W O · U P — D R D G o l dIX. The Cheapest Gold in the CountryNow turn the other way, not down into the rock but up onto the surface, to the second road forward and the company that took it. DRDGold mines no reef at all. It reprocesses the great slimes dams of the Witwatersrand, the flat-topped mountains of finely milled sand a century of underground mining left lying across the Highveld, each one still holding a little gold the old plants were too crude to catch. It washes those dumps down with high-pressure water, pipes the slurry kilometres to a central plant, and pulls the gold out at grades that would be a joke anywhere else on earth: under two-tenths of a gram to the tonne, against the thirteen grams Kloof was built on and the eleven Mponeng now recovers four kilometres down.It runs two operations. Ergo works the old dumps east of Johannesburg, around Brakpan and Germiston. Far West Gold Recoveries works the West Wits tailings around Carletonville, a few kilometres from Kloof itself, on the same goldfield. There is nobody underground. There is no seismicity, no rock hot enough to kill a man, no hour-long descent into ground that bursts. There is sand, water, and the gold somebody else already paid in sweat and blood to bring to surface. It is the cleanest and safest gold mined in South Africa, and at this price it is among the most profitable..T H E B U S I N E S SX. Getting Rich on the DumpsFor at this price the surface is very nearly a licence to print money, and the year just reported was the best in the company’s history. In the year to June 2026, announced on 19 August, DRDGold turned over eleven billion one hundred and fifty-nine million rand, up forty-two per cent, and made an operating profit of six billion four hundred and fifty-two million, up eighty-three — a margin near fifty-eight per cent, against forty-five the year before and thirty-three the year before that. Headline earnings rose eighty-nine per cent to four billion two hundred and fifty-four million rand. Free cash flow rose eighty-five per cent to two billion two hundred and sixty-six million. The all-in sustaining margin reached fifty-three per cent. It declared a final dividend of a hundred and twenty cents a share, three times the previous year’s, in its nineteenth consecutive year of paying one, and finished the year with two billion seven hundred and seventy million rand in the bank and no bank debt at all — having spent three and a half billion rand of capital in the same twelve months. The average price it received for a kilogram of gold rose to some two and three-tenths million rand.The market moved on the day — the share rose on the result, and is up more than half over the year — and it is still, to my eye, mispriced. Asked about the numbers by News24, I said that “it’s so good to see that there is life in the South African gold industry,” and that the discount the company trades at makes no sense to me: “to be trading at a discount to your peers that have so many more risks from a geological point of view is just something I don’t understand.” Hold that beside the subject of this series. The market marks down the producer with no geological risk left to run, and applauds the owner preparing to drown the reef that holds the most gold. In both directions it is declining to look at the rock..Read the chart slowly, because it tells the truth of the model. Revenue is up eighty per cent in two years and operating profit has more than tripled, while the gold poured has gone from a hundred and sixty-one thousand ounces to a hundred and fifty-five and back to a hundred and fifty-six. The margin widened regardless. This is not a mining business in the ordinary sense, where you earn more by lifting more metal. It is a leveraged bet on the gold price, laid over a fixed stock of waste, and the bet is paying spectacularly because the price has run. The same force that lifted Mponeng from break-even to gusher has lifted DRDGold from a scrap-recovery curiosity into one of the most profitable gold operations in the country. The metal is doing the work and the company is supplying the leverage.T H E S C A L EXI. Moving MountainsMake no mistake about the scale of the thing, because the thin grade hides it. To win a hundred and fifty-five thousand five hundred and seventy-seven ounces from sand holding less than two-tenths of a gram a tonne, DRDGold put better than twenty-five million tonnes of old tailings through its plants in a single year, moving, washing and reprocessing a literal range of artificial hills. The economics turn entirely on volume and price, and the ore body is itself a wasting asset, since the richest and most accessible dumps are cleaned up first. The year just reported shows how finely the two are balanced: throughput fell from 25.6 to 25.07 million tonnes while the yield rose from 0.189 to 0.193 grams, and the ounces barely moved. The chief executive likes to say the company sits on six million ounces of gold, though, as he is careful to add, it is in waste. That sentence is the whole model in six words.T H E R E M E D I A T I O NXII. Cleaning Up as a BusinessWhat makes the surface road genuinely attractive, beyond the margin, is that its by-product is a public good. Every dump DRDGold reprocesses is a legacy hazard removed, a source of windborne dust, acid seepage and uranium-bearing sediment lifted off the Highveld and re-deposited, lined and engineered, in a single modern facility. The cleanup is not a cost of the business. It is the business, and it arrives with the regulatory wind at its back, in a country desperate to be rid of a century of orphaned tailings.And the company is ploughing the windfall straight back into the ground, without a cent of bank debt. At Ergo it has built a sixty-megawatt solar plant with a hundred-and-sixty-megawatt-hour battery, around three billion rand, which the chief executive calls a prepaid electricity facility covering about half of Ergo’s power, and has already cut Ergo’s reliance on the grid by about a seventh. At Far West it is part-way through a multi-billion-rand expansion. The whole of it is branded Vision 2028: some ten billion rand over four years, of which better than five billion is already spent, to lift combined throughput from two point one five million tonnes a month to three million and take annual output toward six tonnes of gold. Three and a half billion rand of that went in during the year just reported. The Daggafontein tailings facility came into operation in July 2026 and the second Driefontein elution circuit poured its first gold in the same month; the regional tailings storage facility, eight hundred hectares fully lined and engineered for eight hundred million tonnes, is about two-thirds built, while the Withok facility has slipped to the end of 2029 after test drilling found ground needing more than the standard liner to stop seepage, with the revised timetable resting on the outstanding approvals landing by the end of this year. Management has been candid that uranium is not part of this plant’s future: gold recovery by carbon-in-leach runs alkaline and uranium leaching runs acid, and chasing both would compromise each. This is what a gold business reinvesting in its own future looks like, and it is worth holding the image beside a deep mine whose cash is being posted offshore while its shafts are made ready for flooding.T H E O W N E R S H I PXIII. The Ownership, and the CeilingTwo facts complete the picture, and the first closes the circle of this whole series. DRDGold is majority-owned, just over half, by Sibanye-Stillwater, the company that owns Kloof. The holding was built in two steps: in 2018 Sibanye traded its West Rand tailings-retreatment assets into DRDGold in exchange for a stake of about thirty-eight per cent, and in early 2020 it exercised an option to lift that to fifty point one. Far West Gold Recoveries, the profitable, growing West Wits surface business, is in plain terms Sibanye’s own West Rand project, spun into a vehicle that has made it shine. The company preparing to flood the deep reef already owns the surface model that proves how good the clean alternative can be. And the transition is further along than the plan admits: in the six months to June 2026, surface material accounted for thirty-six per cent of all the gold Sibanye’s South African operations produced, surface output rising thirteen per cent while the deep mines fell away.But be as clear-eyed about the surface road as about the deep one, because it has a ceiling written into its premise. The gold DRDGold recovers exists only because somebody else mined it first and threw the tailings away. It is finite, since there are only so many dumps and the best are cleaned first, and it is wholly derivative, producing nothing the deep reef did not produce before it. At a hundred and fifty-six thousand ounces a year, guided to between a hundred and sixty and a hundred and seventy thousand next year and reaching for something near a hundred and ninety by 2028, DRDGold is not even the largest surface producer in the country — Harmony’s tailings and dump operations lifted a hundred and eighty-six thousand ounces last year — and the entire surface business is a fraction of what the Witwatersrand’s deep mines once lifted in a single year — less than half of what Mponeng alone brings up from four kilometres down. Surface retreatment can clean up the past and mint extraordinary cash from a high price. What it cannot do is replace the deep ounces that made the dumps in the first place. It is a magnificent way to harvest what has already been won. It is not, on its own, a future for the basin.P A R T T H R E E · B O T H — P A N A F R I C A N R E S O U R C E S T H E T H I R D R O A DXIV. The Company That Refused to ChooseThere is a third way through this, and one company has been walking it for more than a decade. Pan African Resources runs a narrow-reef, hand-drilled, conventional underground mine that has been working since the 1880s, and three tailings retreatment plants that put nobody underground at all, inside the same balance sheet — and the surface half is nearer half the business than a sideline. It did not pick a side, and it did not arrive at the position by accident or by acquisition. It built the surface half itself, one plant at a time, until the cheap ounces could carry the dear ones. Harmony, the deep specialist of this paper, has since arrived at the same answer from the other end, and is now the largest surface gold producer in the country: its surface and retreatment operations lifted two hundred and twenty-one thousand ounces in the year to June 2026, of which a hundred and eighty-six thousand came off tailings and waste rock dumps, a fifth more than DRDGold, the company that does nothing else. But Harmony is a hybrid by accretion. The plant at the centre of that business, Mine Waste Solutions, came into the group in the same 2020 package as Mponeng, sold by an AngloGold on its way out of the country, and surface is a thirteenth of what Harmony mines. Pan African is the hybrid by design, and it was the hybrid first. The arithmetic of that difference is now visible in the one place the argument cannot escape: the cost line.In the year to June 2026 the group lifted about two hundred and seventy-five thousand ounces, up some forty per cent, at an all-in sustaining cost near eighteen hundred and seventy dollars an ounce. Set that beside the others. DRDGold, which mines nothing but surface and carries no underground risk whatever, came in at nineteen hundred and eighty-six. Harmony, carrying the deepest mine on earth, a surface business bigger than DRDGold’s and a copper build besides, came in at two thousand one hundred and ninety-five — the honest qualification being that depth, scale and the copper spend all sit inside that number, so what buys a low cost line is not breadth by itself but the method of running it. Sibanye’s South African gold operations, the book that owns Kloof, ran at three thousand one hundred and five across the first half of this year, and that figure already carries its DRDGold stake; strip the surface business out and its managed gold mines ran, in the first quarter, at three thousand four hundred and eighty-six. The company that does both, on the oldest and narrowest ore in the country, is cheaper than the company that does only the safe half, and cheaper by forty per cent than the company that does only the hard half. That is not what the retreat thesis predicts. It is close to the opposite of it..T H E U N D E R G R O U N DXV. Barberton, and the Ore That Was Never Allowed to Be FinishedStart underground, because that is where the argument bites. Barberton sits in the Mpumalanga lowveld, on the oldest exposed greenstone on the planet, and the town was founded on gold in 1884. Sheba was incorporated in 1885 and is still hoisting; the company calls it, with some justice, one of the oldest working gold mines in the world. Fairview alongside it has run for more than a hundred and thirty years. These are not the kind of assets a modern portfolio manager is supposed to want. They are old, they are small, they are narrow, and they are mined by men with hand-held drills on semi-mechanised up-dip cut-and-fill and room-and-stick stopes, to sixteen hundred and sixty metres below adit at Fairview and a kilometre and a half at Consort. Roughly three in ten recovered ounces come from contractors sweeping ground that was already mined once. Thirteen hundred employees and nearly a thousand contractors bring up about seventy-two thousand ounces a year, which is about thirty-one ounces a head. By every metric the retreat thesis uses — age, scale, narrowness, labour intensity, absence of mechanisation — Barberton should have been written off thirty years ago.It was not, and the reason is the whole point. The grade is extraordinary and the owner kept paying to find more of it. Barberton underground recovers about six and a half grams a tonne across the complex, but the Main Reef Complex at Fairview is a different animal: a block declared at forty-two and a half grams a tonne, platforms assayed at ninety-nine grams over three and a third metres and forty-six grams over two, and a Rossiter orebody whose free gold runs from twenty-nine grams to a hundred and sixty-five. Drilling those extensions did not shorten Fairview’s life. It lengthened it, to about twenty-three years. And here is the figure that ought to be read twice by anyone defending the Kloof decision: Pan African books those reserves at nineteen hundred and thirty-four dollars an ounce. That is the most conservative reserve price of any of the four owners in these papers — five hundred dollars below Sibanye’s — and it is applied to the narrowest, oldest, least mechanised orebody of the lot, and it still yields twenty-three years. Sibanye applied the highest price in the basin to the richest reef on the continent and yielded one.T H E S U R F A C EXVI. The Wreck They Bought for Fifty Million RandNow the surface half, and here the story doubles back onto the previous paper in a way that ought to stop the reader. Elikhulu came first: a one-and-three-quarter-billion-rand plant built over the old Evander slimes at Kinross, first gold poured in August 2018, treating better than a million tonnes a month and producing fifty-six and a half thousand ounces last year at an all-in sustaining cost, in the year before, of one thousand and seventy-seven dollars. The Barberton Tailings Retreatment Plant had come earlier still, approved in 2012 at three hundred and twenty-five million rand, built for three hundred and five, and pouring its first gold in June 2013. Neither is glamorous. Both are among the cheapest ounces produced anywhere in this country.But it is the third one that closes the circle. Mintails — the West Rand surface-retreatment company whose 2018 liquidation the last paper cited as a warning, leaving a rehabilitation bill its own documents put between three hundred and three hundred and thirty-six million rand against a provision of twenty-five and a half, and reported at four hundred and sixty million against twenty-eight by the time of the final order — did not simply evaporate. Its Mogale and Soweto Cluster assets went into provisional liquidation, and in November 2020 Pan African agreed to buy them for a maximum of fifty million rand. Fifty million. For two hundred and forty-three million tonnes of tailings carrying two point three six million ounces of gold: about a dollar and forty cents an ounce in the ground. It then spent a hundred and thirty-five million dollars building the Mogale Tailings Retreatment plant on the site, brought it in under budget and ahead of schedule in fourteen months, commissioned it in October 2024 and had it at steady state by that December. It produced about fifty-two thousand ounces in the year just ended, on an eighteen-year life, and a feasibility study completed in November 2025 sets out a further six hundred thousand tonnes a month next door on the Soweto dumps, adding thirty to thirty-five thousand ounces a year for fifteen years at an all-in cost of a thousand to twelve hundred dollars.Read that sequence against the argument of Paper 17. The same wreck that was offered there as the endpoint of pass-the-parcel — a scavenger company folding and leaving the public the bill — was bought out of the liquidator’s hands for the price of a Sandton townhouse and turned into the best new asset in a listed miner’s portfolio, on the West Rand, an hour from Kloof, cleaning up the very ground the liquidation abandoned. The gold was not gone at Mintails either. It was merely owned by people who had run out of money and will. And it is worth recording that in December 2021 Pan African was also examining the surface resources at Blyvoor, the other ruin in that paper. The ounces the industry writes off do not disappear. They wait, at a discount, for someone with a balance sheet and a longer horizon..T H E L E D G E RXVII. The Full Ledger.Be as hard on this case study as on the others, because it is not a fairy tale and its difficulties run in exactly the directions this series has been careful about. Sheba, the oldest of the underground mines, has been the problem child: production down a tenth two years running, gold theft so entrenched that sixty people were caught on the property and hundreds more were trespassing, and a Section 189 process opened in February 2025 over seven to eight hundred jobs, with the chief executive saying flatly that Sheba was not contributing. Barberton’s underground all-in sustaining cost ran to two thousand five hundred and ninety dollars an ounce in the first half of this financial year, against two thousand and sixty-five for the year before. Two employees died in the group’s underground operations in the 2025 financial year and a third shortly after it, at Evander in July. The Australian acquisition at Tennant Creek delivered about thirty-four thousand ounces against a plan near fifty, at a cost above two and a half thousand dollars. Group all-in sustaining cost guidance for the current year has been raised to between two thousand and seventy-five and two thousand one hundred and seventy-five dollars, and the capital budget lifted from two hundred and sixty-seven million dollars to three hundred and twenty-four in the space of four months. The chief executive’s own summary of the industry is worth quoting: an all-in sustaining cost of seventeen hundred dollars an ounce, he said in February, is the new thousand.None of that damages the argument; it sharpens it. The surface operations recorded no lost-time and no reportable injuries at all, while every fatality in the group occurred underground. That is the risk asymmetry stated as plainly as a company can state it, and it is precisely why the hybrid works: the half of the business that cannot kill anyone generates the margin that lets the half that can be run properly, ventilated, supported and supervised, instead of being starved until its unit cost condemns it. Pan African did not solve narrow-reef mining. It bought itself the time and the money to keep doing it while the price re-rated, and it raised Fairview’s life while doing so. Sibanye had the same option available in its own group — it half-owns the most profitable surface retreater in the country — and drew the opposite conclusion.T H E M E T H O DXVIII. Built, Not BoughtNow put the leadership question directly, because a size table hides it. Three plants in eleven years, two of them brought in under the figure the board approved and two built on ground somebody else had walked away from — that is not a portfolio outcome. It is a method, and it was adopted deliberately, by a company small enough that getting it wrong once would have finished it. The turn is on the record. In May 2017 the chief executive said Pan African was done with underground mining; by September of the same year he was saying that the sweet spot was to have the operations combined. What changed in between was the arithmetic of Elikhulu — that a surface plant, built cheaply and quickly, need not replace a narrow reef but can carry it. Read the capital allocation and that is what it says, though the company has never put it in those words: each plant has widened the margin that keeps the oldest gold mines in the world working.That record matters because of who has since followed it. Harmony is the larger surface producer today, but that business was bought with Mponeng, not built, and what Harmony says it will do next is the giveaway. The West Wits and Free State reclamation projects, better than six million ounces of resource between them, are described by Harmony itself as an application of “the proven MWS operating model” to legacy tailings. They are feasibility studies. Pan African’s equivalents have been pouring gold since 2013, 2018 and 2024, and it assembled the third out of a liquidation nobody else would touch. On the model itself — deep ounces and surface ounces carried on one balance sheet, the half that cannot kill anyone funding the half that can — the largest deep miner in the country is the follower, seven years behind, and the mid-tier is the pathfinder. The difference in conviction is the difference between a thirteenth of output and nearly half of it.And the leadership shows up where it counts for an owner, which is in the balance sheet rather than the tonnage. Running the two halves together has taken the group from geared to fully degeared with something near two hundred and twenty million dollars of cash projected at year end, on half-year headline earnings a share up five hundred and twelve per cent, an adjusted EBITDA margin of fifty per cent, and a maiden interim dividend — while the underground half, the part every retreat thesis says to close, was having its life extended rather than shortened. That is the whole argument of these three papers rehearsed inside one small company: the surface pays, the deep reef keeps its option, and the option is worth more each year the price holds. Sibanye holds the identical pieces on a far larger scale — Kloof, Kloof’s dumps, half of DRDGold — and is preparing to put the deep half under water. The difference is not endowment, and it is not capital. It is the choice about what the cheap ounces are for.The market has noticed, which is the part a board answers for. Pan African is worth something over three billion pounds, sits in the FTSE 250, moved from AIM to the London main market in October 2025 and added an Australian listing in July 2026, and has returned about a hundred and twenty per cent over the past year. One of its larger institutional shareholders, which has held it since 2009, puts the case in a sentence this series would happily borrow, paraphrasing Warren Buffett: superior long-term wealth is built by owning the goose rather than the golden egg. A company with a hundred-and-forty-year-old hand-drilled mine and a pile of somebody else’s waste has out-returned every large gold house in this country — not by outspending them, which it cannot, but by refusing the one choice they all insisted had to be made.S Y N T H E S I S · T H R E E R O A D S , O N E O W N E RXIV. Five Postures Toward One BasinStand back, and the basin resolves into five postures toward the same gold. AngloGold left it, selling the deepest reef on earth for a modest cheque and a contingent tail, and went to mine easier ground abroad. Harmony bought that reef and committed to it, spending eight billion rand to go deeper even as it pays a billion dollars for one copper mine and takes a final investment decision on another — and it is also the country’s largest surface gold producer, drawing seven tonnes a year off tailings and dumps at a forty-six per cent cash margin, though that business came to it in the 2020 package with Mponeng and amounts to a thirteenth of what it mines. DRDGold set the deep reef aside altogether and grew rich cleaning up its leavings on the surface, and has just reported the best year of its life. Pan African chose the posture the others have been arriving at piecemeal, and chose it first: it has run the narrow reef and the tailings together for more than a decade, built every surface plant it owns rather than buying one, bought a liquidator’s wreck for fifty million rand and made fifty thousand ounces a year of it, books the most conservative reserve price in the basin, and is the cheapest producer of the four. And Sibanye, which owns Kloof, and Kloof’s dumps, and half of DRDGold, and the West Rand project DRDGold’s surface business was built from, is preparing to flood its deep reef while holding, in its own two hands, every one of the proven ways forward..The table sets the three roads side by side, and the third of them is the standing proof that the first two were never exclusive. Note who is walking it, and in what order. The pathfinder is not some marginal mid-tier hedging its bets but the cheapest producer of the four, which has carried the narrow reef and the tailings in one balance sheet since 2013 and built every plant that makes it work. The convert is the largest deep miner in the country, which added the surface to a deep book by acquisition and is now studying how to do six million ounces more of it. One designed the model and one bought into it, and they arrive at the same answer from opposite ends. Neither road is a free lunch: the deep road is lethal and dear and leans on the price, the surface road is finite and derivative and cannot, alone, replace a reef. But all of it is being walked, profitably, at a gold price within a fifth of the highest ever recorded, and two of the three are being walked within an hour of Kloof. The deep road says the Ventersdorp Contact Reef still pays, and a rival is proving it on that very reef. The surface road says the cash is there to fund the wait. The hybrid says you were never obliged to pick one — and it has said so, in audited numbers, for longer than anyone else in this basin. Put the three together and you have the outline of an answer for Kloof that is neither flooding nor pretending: harvest the surface for cash, hold the deep reef as the option it plainly is, and spend a sliver on the engineering that would make the next four kilometres cheaper than this one.T H E C O N C L U S I O NXX. The Options Still on the TableThat is the case, stated plainly, and it has run through all three of these papers. The plain path was never a mystery. Run the surface dumps for cash, as the company’s own half-owned subsidiary does so profitably — six and a half billion rand of operating profit in the year just reported. Hold the deep option open rather than flooding it, as Harmony’s example shows pays at this price, on this reef. Run the two together, as Pan African has been showing for more than a decade inside one balance sheet and one cost line — the model Harmony has since adopted and is now preparing to apply to six million ounces of legacy tailings of its own. And spend a fraction of the windfall on the narrow-reef engineering that would secure the next century. Flooding Kloof does none of these things. It takes none of the roads forward. It takes the third road of the paper before this, and that road ends at Blyvoor.Three ways forward, all proven, two of them within an hour of Kloof, all of them reporting the best years of their lives at a gold price within a fifth of the record: one by going deeper than human beings have ever gone, into the very reef Kloof calls finished; one by stooping to gather up what was dropped; and one by doing both at once, first, and beating each of the specialists at its own cost line. The richest reef on the continent sits in the gap between them, waiting on a decision that has not yet been made irreversible. This is not a country out of options. It is a company that has not yet taken the ones it already holds, and a reef that will give no second chance once the water is in it.P O S T S C R I P T · A D D E D 3 S E P T E M B E R 2 0 2 6XXI. The Owner Begins to WalkThis paper was finished in the last week of August 2026. On 1 September, after it was written, Sibanye-Stillwater published its results for the six months to 30 June, and the commentary on its South African gold operations contains a sentence this trilogy has been waiting for. The supportive gold-price environment, the company said, “provides an opportunity to assess the potential economic extraction of additional reserves at Kloof beyond 2026, including through the support of appropriately priced hedging mechanisms.” Read that for what it is and for what it is not. It is not a reversal. No Mineral Reserve has been restated; the same filing confirms that there were no material changes to the Mineral Resources and Mineral Reserves declared at 31 December 2025, which means Kloof still carries its hundred thousand ounces and its single year, and its ore reserve development and sustaining capital are now expensed against that short life rather than capitalised. On the results call management said that remaining reserves were under assessment, that nothing was committed, that any additional extraction “has to clear returns and affordability”, and that it was weighing “whether a hedge book is the right way to underwrite a specific block of work”. What the sentence records is a company beginning to walk toward the reconsideration these three papers have argued for: an owner that in February wrote the reef down to a year, and in September is asking, in its own filing, whether there is more to be mined beyond it.No reserve has been restated and nothing is committed — but the owner that wrote the reef down to a year in February is now asking, in its own filing, what lies beyond it.Be as exact about Kloof’s own numbers as this paper has been about everyone else’s, because they are the reason the sentence is hedged. In the same six months Kloof produced gold at an all-in sustaining cost of four thousand three hundred and sixteen dollars an ounce against a realised price of four thousand four hundred and sixty-four — a margin of about a hundred and forty-eight dollars, the thinnest of any operation in Sibanye’s South African gold portfolio. Set it beside the neighbours. Driefontein’s margin ran near fourteen hundred and forty dollars an ounce, Beatrix’s near fourteen hundred, DRDGold’s near two thousand six hundred and fifty, and the South African gold book as a whole, at an average price received of four thousand five hundred and ninety-seven dollars against a cost of three thousand one hundred and five, near fifteen hundred. A hundred and forty-eight dollars is within twenty dollars of the margin Mponeng was running in the year to June 2022, before the price re-rated a marginal mine into Harmony’s best — with the difference that Mponeng had a plan, and seven years of reserve, to spend the re-rating on, and Kloof at present has neither. That is almost certainly why the company pairs “economic extraction” with “appropriately priced hedging mechanisms” rather than treating the matter as a spot-price decision. A hundred and forty-eight dollars of margin at a rebased mine is not a cushion a board commits capital into off a spot screen; it is a margin that has to be locked in, block by block, before a single one of those blocks is planned. Say so plainly. This postscript does not report a vindication. It reports the first step, taken in the company’s own words, and the arithmetic that makes the step a cautious one.None of that changes the argument of the trilogy; it dates it. The three roads set out above are still open, the reef is still in the gap between them, and the water is not yet in it. What has changed since August is that the owner has said, for the first time in a regulatory filing, that the question is worth asking. The decision Paper 16 watched it defer is now, on the company’s own account, under assessment. The rest of this paper stands as written..*Dr Duarte F da Silva · Capital Markets Strategist · Managing Director, Northbound Processing · Germiston, 2026.Sign up for your early morning brew of the BizNews Insider to keep you up to speed with the content that matters. The newsletter will land in your inbox every morning on weekdays. Register here.Support South Africa's bastion of independent journalism, offering balanced insights on investments, business, and the political economy, by joining BizNews Premium. Register here.If you prefer WhatsApp for updates, sign up to the BizNews channel here.

Da Silva’s Gold Papers: Sibanye's plan to flood the richest reef on earth - at best gold price ever

Related Stories

Dr Duarte da Silva names SA mining's builders, leavers and breakers. Brilliant.

Duarte da Silva's Gold Papers: Who held the gold? - the numbers

Stroke victim loses R450,000 disability money on online trading platform

Dr Duarte da Silva: SA gold mine owners’ windfall funded departure from Reef