Dr Duarte da Silva: A diamond is forever. De Beers was not.

A diamond's value was never in the stone. It was in a company's willingness to lose money to protect it. The family that built that willingness into a system gave it up between 2000 and 2010 and sold the company at the top in 2011. The professionals who bought it paid a price that valued De Beers at thirteen billion dollars and ran it as a mine. They are reported to be selling it for about a billion, on terms not yet signed, to a consortium led by the man who was its chief executive under the family. A gold strategist sets the stone beside the metal, and finds that the one with no custodian kept its value and the one with a custodian lost him .By Dr Duarte F da Silva*.The stone was sold to the world as forever. For eighty years a company made the promise good: one channel for most of the world's rough, a stockpile that bought in every downturn, and two hundred million dollars a year of advertising. This report says who dismantled it, what it cost, and who paid. The Oppenheimers renounced the custodian's role in 2000. They sold the stockpile by 2004 and shut the generic advertising by 2010. In 2011 they sold their forty per cent to Anglo American for US$5.2 billion, at the top. Anglo's professionals thought it cheap and carried it at or near the purchase figure until 2024. In 2016 they said in writing that the laboratory stone would not have enduring value. In 2018 they launched a brand to sell it. They closed that brand in 2025, and in August 2026 they conceded that the signals had been missed. They have written off US$6.8 billion and are reported to be selling to a consortium led by Gareth Penny, on terms not yet signed. A separate section tests the reading that the family was the trader and Anglo the patsy. It finds that reading true in outcome, unproven in intent, and sharper than it sounds: Anglo paid a price the market called low for a thing the seller had already dismantled. Four calculations carry the argument. What did Anglo's shareholders get back on the 2012 cheque? What would the same cheque be worth in gold? What would it cost the buyer to hold the price? Where has the margin on a one-carat ring gone? The report ends with the fortunes made and lost, and with the people who believed the stone was a store of value. It sets out what a custodian would have done from 2012, decision by decision and priced. It closes with the lesson for the platinum, chrome and manganese South Africa still dominates .The method is simple and it is stated first. Every figure carries its date and its source, and most of the sources are Anglo American's and De Beers' own results, releases and filings, read in full for 2011 to the third quarter of 2026. The numbers sit in the tables and the charts; the text says what they show. Each decision is placed against the name of the person who made it and the chairman who approved it, with the pay where it is public in the remuneration reports. Where a figure rests on press reporting of an unsigned transaction, the text says so. Where a figure is an estimate, it is called one. Quotations are verbatim, with the publication and date in the references. The judgements are mine. The record is theirs. .THE VERDICT · WHY A GOLD MAN · THE METHOD · THE DIAMOND FIELDS · RHODES'S IDEA · WHAT THE FAMILY DISMANTLED · THE CHEQUE · THE FAMILY'S TRADE · THE OTHER EXIT · THE RECORD · THE BUILDERS · THE STONE · THE PRECEDENTS · THE SALE · BOTSWANA · THE PEOPLE · THE FORTUNES · NOT A REAL DIAMOND · THE BELIEVERS · THE CHAIR, OCCUPIED · WHAT SOUTH AFRICA STILL HOLDS · THE BEAR, MET · THE VERDICT · WHAT RHODES AND ERNEST WOULD SAY .BEFORE THE VERDICT A line written late at night In 1947 a copywriter at N.W. Ayer in Philadelphia, Frances Gerety, had the De Beers account and a deadline. The agency's own history, and the account she gave of it in later years, tell the same story. She had finished the night's work without the line the campaign needed. She went to bed, got up, and wrote four words on a slip of paper so that she would not forget them. In the morning her colleagues did not think much of them. The client ran them in 1948 and in every De Beers advertisement after, until the generic campaign was shut in 2010. "A diamond is forever" was named the slogan of the century in 1999. Gerety never married. The line was not a description but an instruction, and the company that paid for it did the work the line required. It bought every diamond the world could not absorb, stored it, and waited. The aim was that nobody who had been given a diamond would ever learn what it was worth to anyone who had to sell it. That work had a cost. A family bore it for eighty years because it intended to be there in eighty more. The line outlived the work. This report is about what happened to the stone when the people who had paid for the promise stopped paying. It is also about the metal that was never promised anything and did not need to be. A Diamond Is Forever. De Beers Was Not. · Research Exposé, November 2026 Page 3 THE VERDICT I. What was built, and what the professionals did with it A diamond's value was never in the stone. It was in a company's willingness to lose money to protect it. When the owners stopped being willing, the value went with them, and no buyer since has had the balance sheet or the patience to bring it back. That is the thesis of this report, stated once. Every section after this one tests it against the record, with the names and the dates, and the reader is asked to keep score. De Beers was not killed by a laboratory. It was dismantled by its owners between 2000 and 2010. It was sold at the top in 2011 to buyers who, on everything they have published since, did not know what they were buying. Those buyers ran it as a mining company for fourteen years. Then it met the laboratory with none of the three things that had protected the price for eighty years. Those were control of the supply, a buffer against the cycle, and a story told loudly enough to hold the price. The family's part is stated first because it came first. Nicky Oppenheimer and Gary Ralfe announced in 2000 that De Beers would stop being the custodian of the diamond price. They sold two-thirds of a century's stockpile into a rising market in four years, repaid the debt of the 2001 buy-out with the proceeds and paid dividends from what was left. Their successors redirected the generic advertising in 2008 and shut it in 2010. They did this for reasons that were respectable and are set out below. The effect was that the mechanism Rhodes had built and Ernest Oppenheimer had perfected was gone before the first colourless laboratory stone reached a shop window. Then the family sold. Anglo approached them. They had refused for years. In 2011 they said yes, at the top, for cash. They have not been wrong by a dollar since, and they have said nothing. Anglo's part is the larger charge, because Anglo paid for the chair and did not sit in it. Sir John Parker made the approach. Cynthia Carroll signed the agreement in November 2011, was appointed chairman of De Beers in September 2012, and four weeks later announced that she would leave Anglo. Mark Cutifani chaired De Beers from July 2013 to 2022, nine years. In those years a two-billion-dollar shaft, approved by the boards the year before he arrived, was built and did not deliver. A brand was launched to sell the stone the company had said would never matter. And the business was carried in Anglo's accounts at or near the 2012 figure while the realised price per carat fell by a third. Duncan Wanblad took three write-downs, decided the company did not fit, and is trying to sell it in the worst diamond market since 2009. Between them the owners employed a railway executive, a lawyer and an oil man to run the business. None of the three had sold a diamond before he arrived. The laboratory stone did what De Beers' own executives predicted in writing: it became cheap. What they did not predict is that cheap would win. In six years it took the centre of the American engagement ring, the product De Beers invented. The mined stone's realised price, with no buffer and no voice to defend it, fell by half. Anglo's chief executive conceded in August 2026 that "we might have wanted to have been a little bit more aggressive on the signals we were getting". The signals were in his own company's files. De Beers was not killed by a laboratory. Its owners sold the scarcity, its buyers never acted as if it was there, and the stone from the factory walked in through the open door. The report goes further than the sale, because the record does. It sets the stone beside gold, the obvious control in the experiment, because the author is a gold man and the comparison is the point. It goes back to Kimberley, because diamonds funded every mining house in this country. It sets Rhodes's idea beside the commodity businesses that still run on it. It counts the money: who made fortunes between 2000 and 2026 and who lost them, and what the people who believed the stone was a store of value were left with. It sets out, decision by decision and priced, what a custodian would have done in Anglo's chair from 2012. And it ends on what South Africa still dominates, which is run today the way De Beers was run after 2011. For the reader who holds South African mining shares, every one of those companies is now being tested on the same four requirements. The next five sections elaborate what they are. A Diamond Is Forever. De Beers Was Not. · Research Exposé, November 2026 Page 4 WHY A GOLD MAN II. The metal and the stone The two stores of value grew up in the same houses. Ernest Oppenheimer formed Anglo American in 1917 to buy gold ground on the Far East Rand. He took control of De Beers in 1929. For the next thirty years he ran gold and diamonds as one system from the same boardrooms on Main Street. The diamonds carried the gold when gold was weak, and the gold carried the diamonds in the Depression. In those years De Beers shut every mine it had and held the price. For eighty years this country sold the world its two great symbols of lasting value from the same office, and the men who ran one had usually run the other. I grew up on the goldfields, among the headgears of the Witwatersrand, and spent about thirty-five years in capital markets watching what the market did to each. That is the standing I bring to this report and the reason I wrote it. Then history ran the experiment. One of the two had a custodian: a single seller, a stockpile, an advertising budget and a slogan. The other had none. There has never been a gold cartel that worked, no stockpile manager and no "gold is forever". The only large holders who could have played the role, the central banks, spent the 1990s selling. Between 2000 and 2026 the custodian walked away from the stone, in the stages Section VI sets out. The metal never had one to lose. What happened to each is in the exhibit. The price of a mined diamond, on De Beers' own index, is below where it stood before the financial crisis. The price of gold is about fourteen times what it was in 2000, when the family began selling the stockpile. It is about two and two-fifths times what it was when Anglo's cheque for the family's forty per cent cleared. .The metal and the stone, 2011 to 2026. Gold at the LBMA PM fixes of the dates shown and US$3,874.40 at the close of 30 October 2026; rand at R16.71. The diamond figures are from the indices and lists named in the references; the sale terms are Bloomberg's and unsigned. Author's arithmetic; the working is in the note to the table..The lesson for a gold reader is the one the table makes without help. Gold's value rests on something no company owns. It has five thousand years of use as money. Central banks hold it as a reserve and have been net buyers in every year since 2010. No producer anywhere can make more of it on demand. A diamond's value rested on scarcity that was organised by a company, and could be sold off by one. Once the organiser left, the price had nothing under it. That is why the stone fell when the custodian walked, and the metal rose while nobody guarded it. In August 2012 Anglo American paid US$5.2 billion for the right to sit in Ernest Oppenheimer's chair. Had it bought gold that day, it would hold about US$12.6 billion of metal this morning, before custody and insurance. It would have needed no chief executive, no stockpile and no slogan to keep it. Why it matters here, now, is the rest of the report. The country that lost the custodian of the stone is the same country now deciding, mine by mine, how its gold, platinum, chrome and manganese are to be run, and by whom. The diamond is the first case and the clearest, because the mechanism was visible and the people who dismantled it said so at the time. A South African investor who understood why the stone fell and the metal did not has the only test that matters for the platinum belt. Section XXI applies it.The metal had no custodian and did not need one. The stone had one, and lost him. Anglo's 2012 cheque, in gold, would be worth US$12.6 billion this morning; in De Beers it is worth under half a billion on the reported terms. THE METHOD III. How each decision is scored Hindsight is cheap, so the scoring is narrow and the burden is on this report. Each decision is placed at its date, with the information the decider had, the person who made it, the chairman who approved it, and the pay where it is public. The outcome is measured in De Beers' own figures: production, sales, the realised price per carat, earnings before interest, tax and depreciation, capital spent, and the value Anglo carried the company at. All of these come from Anglo's results 2012 to 2026 and De Beers' own for 2011. The laboratory stone is measured by the industry's own counts. These are The Knot's surveys of American couples, which are surveys and are called that; the wholesale price indices; India's export statistics; and Bain's reports. The claims the industry makes are tested against the regulators who have ruled on them. Three objections are met in advance. First, that the family could not have held the single channel at forty per cent of the market. Conceded in part. The stockpile and the advertising were not the channel, and that is where the scoring falls. Second, that nobody saw the laboratory stone. Section XII answers it, in the company's own words and dates. Third, that the 2011 price was fair by the brokers' standards of the day. Conceded as to the brokers. But the sellers were the Oppenheimers. A buyer who does not ask why the people who made the market are leaving it has not done the one single piece of diligence that matters. Two cautions on the figures. A realised price per carat is an average of what was sold in the period, and it moves with the mix as well as with the market. Where the text says the price fell, the index that measures like for like is given beside it. And a carrying value in an acquirer's accounts is not a market price. It moves with depreciation, investment, distributions and impairment, and the chart in Section X keeps the three kinds of figure apart. Currencies are those of the transaction at the rate of the day. The 2026 sale terms are Bloomberg's; Anglo has confirmed no price. Nothing is adjusted for inflation, which flatters every seller in the report. THE DIAMOND FIELDS IV. Diamonds funded everything Barney Barnato arrived at Kimberley in 1873 with forty boxes of cigars to sell and no money. Sixteen years later he wrote his name under the largest cheque the world had seen. The men who made South African mining learned it at Kimberley, and they paid for the rest of it with Kimberley's money. The table is the list. Cecil Rhodes arrived at the diamond fields in 1871, aged eighteen. Alfred Beit came in 1875 as a diamond buyer for a Hamburg firm, and brought Julius Wernher. J.B. Robinson, Solly Joel, Lionel Phillips, Hermann Eckstein: the roll of the Randlords is a roll of Kimberley diamond men. When gold was found on the Witwatersrand in 1886, these men bought the farms, floated the companies and sank the shafts. They did it with the capital and the London connections the diamond fields had given them. Rhodes and Rudd formed Gold Fields of South Africa in 1887. Barnato formed Johannesburg Consolidated Investment in 1889 on the strength of his De Beers shares. Beit lent Robinson the money for his first Rand farms in 1886. Then, with Wernher and Eckstein, he built the Corner House, which floated Rand Mines in 1893 and opened the deep levels. The gold houses were diamond houses that had moved three hundred miles north. The fortunes were private, and the table says whose. De Beers as a company did not finance the Rand; its shareholders did. The next link in the chain was forged by the same families. Ernest Oppenheimer arrived in Kimberley in 1902 as a diamond buyer. He formed Anglo American in 1917 to buy gold ground on the Far East Rand. He used Anglo to buy the German diamond fields of South West Africa in 1920, took control of De Beers in 1929, and ran the two companies as one system for the next thirty years. Then the gold houses financed the platinum. JCI, Barnato's house, consolidated the Merensky Reef discoveries into Rustenburg Platinum in 1932 and controlled it for sixty years. Union Corporation, a gold house, founded Impala in the 1960s, and Gencor, a gold house, built it. When Anglo took JCI's platinum in 1995 it became Anglo American Platinum. Diamonds built the gold houses. The gold houses built the platinum. The chain runs from the Big Hole to Rustenburg without a break, and the first link was the one Rhodes forged. That is why De Beers matters more to this country than its revenue suggests. It is the origin of the system. The method was this: a South African mining house held a price through a cycle, bought at the bottom with money earned at the top, and lasted long enough for the ground to pay. Rhodes invented it at Kimberley. Oppenheimer refined it at De Beers. It was then applied to gold, to platinum, to coal and to copper. This report scores what the professionals did to the first link. The gold this country lost between 1970 and 2024 was not lost to a substitute. It was lost by the same route: the sale of control by people on a short clock. The diamond is the clearest view of the mechanism, because the sale was a single cheque on a single day. .The chain from the diamond fields to the platinum belt: who financed what, from the company histories cited in the references. The fortunes were the diamond men's own; the gold houses they founded then built the platinum..Every Randlord was a Kimberley man. Gold Fields was founded by Rhodes and Rudd and JCI on Barnato's De Beers shares; the gold houses then built the platinum. The chain runs from the Big Hole to Rustenburg without a break. RHODES'S IDEA V. Consolidation, and what it needs to work In 1888 there were too many claims at Kimberley, too many sellers and a falling price. A thirty-four-year-old with a weak heart and a Rothschild loan bought them all. Rhodes bought out Barnato's Kimberley Central and amalgamated the fields into De Beers Consolidated Mines on 13 March 1888, with a trust deed that let the company do almost anything it liked. The cheque for the Central's assets, five and a third million pounds, was the largest written to that date. It followed in July 1889 and is still framed. Within two years De Beers produced nine-tenths of the world's diamonds and sold them through one syndicate in London. The price stopped falling. Rhodes's reasoning, as Epstein records it from his letters, was that the supply of diamonds should be matched to the number of engagements in America and no more. A stone that was plentiful was a stone that was worthless. Ernest Oppenheimer wrote the same rule down before he owned a mine: "the only way to increase the value of diamonds is to make them scarce, that is to reduce production." The idea is now the organising principle of the commodity world, and nobody calls it a cartel when it works. Three companies ship most of the world's seaborne iron ore and their discipline sets the price. A dozen states run oil through OPEC, with two price wars to remind the others what discipline is worth. Two selling consortia ran potash until one walked out in 2013 and the price fell by a quarter in a week. Concentration is not control, and a producer does not set a price by existing. But everywhere a mineral is concentrated in few hands, the hands do what Rhodes did, and the price holds for as long as they hold together.What the idea needs is also simple, and De Beers is the best-documented case of what happens when the needs are not met. It needs control of enough of the supply to matter. It needs a buffer, which is either a stockpile or spare capacity, so that the controller can absorb a downturn instead of passing it to the price. It needs a story, told at scale, so that the demand exists to be matched. And it needs an owner with time. The whole method is to lose money in the bad years in order to hold the price for the good ones, and no hired manager paid on three years' results will do that twice. Rhodes had the time; he was building a company to outlast him. Ernest Oppenheimer had it; he was building for his son and his grandson. Harry Oppenheimer held the price through 1981 and 1982, when the Central Selling Organisation's sales fell by more than two-fifths in a year. He did it by buying and storing stones that nobody wanted, halving the dividend, and taking the loss. The professionals who ran De Beers after 2011 had none of those four things in full. The sections that follow show which of them each man gave up, and how much of each was left. .Rhodes's method as four requirements, and where each stood in 1999 and in 2026. Sources: Section VI, the NDC budget series and the references..Consolidation needs four things: the supply, a buffer, a story, and an owner with time. De Beers had all four in 1999. By 2010 its owners had given up most of three. Its buyers never knew about the fourth. WHAT THE FAMILY DISMANTLED VI. Nicky Oppenheimer, Gary Ralfe and Gareth Penny, 2000 to 2010 In July 2000 Nicky Oppenheimer sat down with Martin Rapaport in London and said the sentence that ended a century. "We are moving from being the seller of last resort to being the seller of choice." The family took the machine apart before it sold it. The family said so at the time, in its own words. After two years of Bain & Company's review, Oppenheimer went on: "Our priority has shifted from controlling supplies to driving demand." The company's spokesman put it more plainly a year later: "We do not ever see us returning to the full custodian role." Gareth Penny had run the review and would run the company. In 2004 he said what the review had concluded about the 1990s: "We absolutely destroyed shareholder value." The 1990s had been hard. Argyle had left the channel. The Russians had leaked billions of dollars of rough. Canada's new mines sold outside. The Asian crisis had cut sales by a quarter. The United States had indicted the company and kept its executives out. The stockpile had grown in nine of ten years to nearly five billion dollars. The share price was at a ten-year low. A review was due. What the review decided was to stop being De Beers. The stockpile went first, and the pace was the point. Nearly five billion dollars at the end of 1999. Under a billion by the middle of 2004, which the company called working levels. The organisation had bought diamonds in every downturn since 1934. In four years it sold two-thirds of a century's buffer into a rising market. The proceeds repaid the debt of the 2001 buy-out and funded dividends of US$400 million in 2003 and US$450 million in A Diamond Is Forever. De Beers Was Not. · Research Exposé, November 2026 Page 8 2004. Nicky Oppenheimer had called the stock "a burden". It was also the only thing that had ever held the price. Harry Oppenheimer had used it in 1981 and 1982 to do exactly that. The advertising went next, and the reasoning was recorded. In 2008 De Beers' share of world sales was down to about forty per cent. It moved its marketing budget to its own brand, Forevermark. Its executives told the trade press why: with that share they no longer wished to advertise other companies' diamonds. By the end of 2010 the American generic campaign, the one that had run "A diamond is forever" since 1948, was shut. The decision predated the crisis. As one analyst wrote at the time, the crisis changed the timing, not the strategy. For five years nobody told the world that a diamond was forever. The producers formed an association in 2015 to say it again. Its first-year budget was six million dollars, three per cent of what De Beers alone had spent at the peak. The budget can be dated, and the dates settle who cut it. De Beers spent about US$220 million on marketing in 2007, the last year before the money moved to Forevermark. In 2018 it spent US$166 million. That is a quarter less in nominal dollars, over a decade in which the company earned more than a billion dollars in seven years out of ten, and 2018 was the year it launched Lightbox. For this decade it has published no figure. In 2025 it announced its "largest category marketing investment in more than 10 years", and gave no number; a company that had the number would print it. Two figures, eleven years apart, are the bookends of the story. The family cut the generic campaign on a share of forty per cent because the money advertised other people's stones. The professionals inherited a budget that was already falling and let it fall further, while the stone that needed a story was arriving in the shop window. Paul Zimnisky, the independent analyst whose series this report uses, put it this way in November 2026: "Diamonds are more than just an asset you have to mine. You have to generate demand." Over the past decade, he said, "it doesn't seem like generating demand was the priority." It was not the priority because it was not in the plan, and nothing Anglo has published shows that it knew there had been one. .The three parts of the method and when each went: De Beers' share of world rough sales; the stockpile; and the generic advertising budget. Sources: Gems & Gemology (2005); De Beers and Anglo American releases 2000–04; JCK and IDEX on the marketing, 2008–10; references at the end..The share can be followed past 2008, and the curve is the argument of Sections VI and X in one line. Paul Zimnisky's estimate of the share of world diamond supply distributed by De Beers, by value, runs from about eighty per cent in 1990 to about a third in 2026. It fell by half in the family's last twenty years, from eighty to forty. It fell by a further quarter in Anglo's first decade, from forty to the low thirties. It rose to about fifty-four per cent in 2022–23, which coincides with the sanctions on Alrosa; that reading of the cause is mine, not Zimnisky's. Then it fell to about twenty-three per cent in 2024–25, the lowest point in the series, as the buyers refused the goods and De Beers held them. The windfall of 2022 was the one year in fourteen in which Anglo was handed something close to the custodian's share. It did nothing with it. By 2024 the share was lower than it had been before the windfall. .De Beers' estimated share of world diamond supply by value, read from the Financial Times' chart of Paul Zimnisky's series, 2 November 2026. Values are read from the chart to the nearest two or three points and are approximate. The events in the third column are the paper's own, dated from Sections VI to XIV..The scoring is this. Ralfe and Penny carried out the family's decision, and on its own terms they did it competently. The debt was cut. The regulators were satisfied. When the crisis came in 2009 Penny halved production, shut mines and raised a billion dollars. Mmegi wrote when he left that cutting the company's debt by forty per cent "may prove to be Penny's greatest legacy". The decision itself is the charge, and it is the family's. A company with forty per cent of the market cannot run a cartel. It can still hold a buffer and still tell the story. The family gave up the buffer and the story because each cost money in the short run. In doing so it turned the company Rhodes had built into a mining company with a famous name. That is what it then sold. For the shareholder in a South African producer, the lesson is this. The stockpile and the advertising were the two things a forty per cent producer could still have afforded. They were cut first, because they were the two that showed in the next year's accounts. The family stopped being the custodian in 2000, sold the buffer by 2004 and shut the story by 2010. Then it sold the company. It converted De Beers into a mining company with a famous name, and Anglo bought the name. THE CHEQUE VII. Parker and Carroll buy the chair, and do not sit in it Nicky Oppenheimer was at his desk in September 2011 when Sir John Parker came to see him. Two weeks later he had an offer on the table for the thing his family had held since 1929. Anglo asked. Both sides of the table say so. James Teeger, the family's managing director, said on the day: "We didn't approach Anglo, Anglo approached us." Oppenheimer said: "I was extremely surprised when John Parker came to see me about two weeks ago and put an offer on the table." Reuters reported that the family "had long rebuffed informal approaches from Anglo American", including during the crisis. Parker "tried again in September 2011, and Nicky Oppenheimer was more receptive". Why he was more receptive is on the record too, and none of it is a laboratory. The family had put two hundred million dollars of loans into De Beers in 2009 and four hundred million into a rights issue in 2010, borrowed against its Anglo shares. Venetia needed about fifteen billion rand to go underground. A family holding company is not the balance sheet for that. Mary Slack, Nicky's sister and the other half of the family vote, wanted the profit taken after the losses of the crisis. And Anglo had refused to give Jonathan Oppenheimer a seat on the De Beers board. That closed the route by which he would have succeeded his father. "Do I have an ambition that Jonathan should succeed me as chairman of De Beers?" Nicky asked that week. "Absolutely!" Anglo held an effective veto over the chair, and it had already answered. Those are the reasons the family and the reporters gave. What the family thought beyond them is not on the record and is not supplied here. Ten months earlier he had said a sale "seems to me an unlikely event". On 4 November 2011 he called it "a momentous and difficult decision" taken "for reasons that I am not prepared to go into". The family has said nothing about De Beers since, and it did not bid. It saw a top in its own market, a balance sheet that could not carry the next shaft, and a succession the buyer had blocked. It did what the Oppenheimers had done in every generation. It read the cycle and acted before the people reading the quarter. What Anglo's board thought it was buying is the heavier charge. The Mail & Guardian reported from inside Anglo that week that "Anglo's board reportedly doubted Nicky Oppenheimer's ability to fix De Beers, and Anglo CEO Cynthia Carroll was reportedly instructed to end the arrangement". The same report had a former director calling De Beers "a misfit", and head office thinking itself "stuck with this asset anyway". Carroll called the purchase "a unique opportunity for Anglo American to consolidate control of the world's leading diamond company". Parker called it "fully aligned with the Board's strategic priorities". The brokers thought it cheap. On the simple arithmetic, US$5.1 billion for forty per cent values the whole at US$12.75 billion. That is about seven and a half times the earnings before interest, tax and depreciation of the best year in the company's history. The brokers used lower multiples, because the price included the family's shareholder loans. A proper enterprise multiple would separate the loans, the debt and the minorities. Anglo's release did not do that. Liberum said the deal would add eight per cent to Anglo's earnings per share. RBC said it was thirty per cent below its own valuation. Anglo's shares rose on the day. Not one contemporaneous note called it the top. Fifteen years on, the people who were near the room say what the brokers' notes did not. "Nobody on the Anglo board really understood it," a person familiar with the board's thinking in recent years told the Financial Times in November 2026, noting that the market for diamonds was "radically different" from the markets for industrial metals. A former De Beers executive gave the verdict in seven words: "Anglo has destroyed a very good business." Another said the company had been "caught asleep at the wheel" on the storytelling that sells a natural stone to a younger buyer. All three are anonymous and are weighed as such. But they describe in 2026 the purchase this report describes in 2011: a board that bought the chair of the diamond market on a mining multiple, from the family that had set the price, and left no record of asking what the chair was for. Sir John Parker, who made the approach, called diamonds "an exception in the gloom" for Anglo during the commodity downturn of the mid-2010s. They were. The exception was the stock of goodwill the family had built, running down. The table has the terms. The verdict is short. Anglo paid the people who had set the price of diamonds for eighty years for the right to set it, and then did not set it. Carroll was appointed chairman of De Beers on 28 September 2012. On 26 October Anglo announced she would step down as its chief executive. Parker signed five more years of accounts at or near the purchase figure. The man running De Beers, Philippe Mellier, had been chosen by the family in 2011, a year before completion. Anglo kept him. .The purchase. Sources: Anglo American's releases of 4 November 2011, 31 July 2012 and 16 August 2012, the Annual Report 2012 and the Directors' Remuneration Report 2011. Also the reporting of Reuters, Miningmx, the Mail & Guardian and Business Report cited in the references..Anglo asked. The family said yes because it had just put six hundred million dollars in, Venetia wanted fifteen billion rand more, and Anglo would not seat the son. The brokers said cheap. Nothing on the record shows anyone asking why the people who made the market were leaving it. In July 2012 Botswana declined its pre-emption: a quarter of De Beers for about US$1.25 billion. That is the stake it is now reported to be taking for a fraction of that price. A state that reads the cycle is worth more to its people than one that reads the broker's note. Section XV says whether this state is reading either. THE FAMILY'S TRADE VIII. Was the family the trader, and Anglo the patsy? The thesis tested On 14 July 2000 Gary Ralfe told a hall of gemmologists that the market placed "almost no value" on De Beers' US$4.5 billion of stockpile. He said the shareholders "haven't had a good ride in the 1990s". He said there would be "no sacred cows". Eleven years later the family sold for about four times what its stake had cost it in 2001, on the unadjusted figures. This section tests whether that was the plan. Did Nicky Oppenheimer and Ralfe set out in 2000 to convert a custodian that the market would not pay for into a mining company that a buyer would? Did they abandon a mantra that had run out of room at forty per cent of world supply, push the price to a record, and sell to the one buyer who could not say no? The verdict comes first. In outcome, yes, and the numbers are below. In intent, the record does not prove it and in places points the other way. And the patsy was Anglo, though it did not overpay. It paid what the market called a low price for a thing it did not understand. That is the only way to be a patsy at a fair price. The sequence is confirmed step by step, and the table dates it. The custodian role was renounced in 2000. The stockpile fell from about US$4.5 billion to working levels by 2004. The proceeds went first to repay the US$3.5 billion of debt the 2001 take-private had loaded onto the company. Net debt fell from US$3.17 billion at the end of 2001 to under a billion by the end of 2003. The European Commission adopted a favourable position on Supplier of Choice in January 2003. De Beers pleaded guilty in Ohio in July 2004, paid ten million dollars, and its executives could enter the United States again. The American class action was settled for two hundred and ninety-five million dollars in 2008. The family appointed an outsider to run the company in mid-2011. Rough prices rose by twenty-seven per cent in 2010 and by about a third in the first half of 2011, the best half-year in the company's history. The sale was agreed in November 2011, five months after the peak. The terms accrued interest to the seller, paid the seller if completion slipped, and paid the seller a share of any uplift if Anglo listed the company within two years. Every one of those steps raised the price a buyer of a mining company would pay, and none of them is what a custodian does. .The family's steps from 2000 to 2012 and what each did to the price, from Anglo American's and De Beers' releases, the EC and DoJ records and the contemporary reporting cited in the references. The return is the author's arithmetic on the sourced inputs..What the record does not support is also stated. The family did not take the stockpile out in cash. The proceeds repaid the buy-out debt. The ordinary dividends were modest after 2005 and nil from 2009. The family put about six hundred million dollars back in 2009 and 2010. Its net draw before the sale was therefore of the order of a few hundred million. The debt was not cut and held. It was rebuilt to four billion by 2007, and the de-gearing before the sale was funded by the shareholders' own billion. And no document, interview or speech has been found in which Oppenheimer, Ralfe or Penny links Supplier of Choice or the sale of the stock to an eventual exit. In January 2011 Oppenheimer said a sale "seems to me an unlikely event", that the company "lends itself to being a private company", and that it "will give us a proper return". In June he said he had heard nothing of Anglo buying. The approach came from Parker in September. The people closest to the decision name Mary Slack, the blocked succession and the 2009 cash call. One associate, unnamed, said that in June 2011 "prices were insane, at a peak". That is the whole of the evidence for timing as intent, and it is not enough to print as fact. Then the price, which is where the thesis is usually thought to stand and where it falls. The market of November 2011 did not think Anglo overpaid. RBC put the price at twenty-five to thirty per cent below its own valuation. Cadiz said: "I don't think Anglo overpaid." An analyst quoted by Bloomberg called it "on the low side, but for the Oppenheimers there was only one buyer". Botswana held the pre-emption. Anglo held an effective veto over the chair. A private shareholders' agreement between three parties left no room for a Chinese or sovereign buyer. The structure depressed the price, and Anglo paid a fair one by the standards of the day. If the family was running a trade, it did not get a trader's price. It got the price of the only exit it had, at the only moment the exit was open, and it took it. So the thesis survives in a stricter form than the one it started in. The mantra had run out of room. At forty per cent of the world's rough the custodian could not be funded. The family's own executive said in 2004 that the 1990s had "absolutely destroyed shareholder value". Everything the family did from 2000 was what an owner does with a business it may one day sell. Make it look like the businesses buyers value. Settle what a buyer cannot carry. Defend it through the crisis. Go when the door opens. That is foresight, or a trader's reflex, or an old man's exhaustion, and the record cannot say which. What the record can say is who was left holding the thing. The family bought its stake for about US$1.2 billion, mostly in paper. It sold it for US$5.2 billion, in cash, at a return of about fourteen per cent a year on that unadjusted cost. Anglo bought the same stake at a price its brokers called cheap. On the account in Section X, Anglo has lost about twelve to fourteen per cent a year on it, and about ten billion dollars of cash and value on the whole eighty-five per cent. The patsy in a trade is the one who does not know what he is buying, whatever he pays. The family knew, because it had just finished taking it apart. The family sold at about four times its cost, at a price the market called low, to the only buyer it had. Anglo was not ripped off. It paid a fair price for something the seller had already dismantled, and its record shows no sign that it knew. For the buyer that is worse. THE OTHER EXIT IX. The family leaves Anglo first: the register, 2001 to 2017, and what the cash did not buy In March 2017 Nicky Oppenheimer gave the Sunday Times the only account he has given of why the family let go of the company his grandfather founded. Anglo, he said, "was becoming less strange and more normal." The ethos in which business partners were also friends had ended when the company moved to London in 1999. Of the shares the family still held: "if we thought the price was high we would sell them." By then the holding was nine-tenths of one per cent. The previous section tested whether the De Beers sale was a trade. This section tests what the De Beers sale hides: that it was the last step of a withdrawal from Anglo American that began the year the professionals took the chair, and that what the family did with its Anglo shares, before the cheque and after it, says what no interview does. The verdict first. The family did not time Anglo. It left it. It sold near two highs and near the low, and when it was handed US$5.2 billion in cash it did not buy one share of the company it had founded. It sold more. The register, step by step. Before 2001 the family's position in Anglo was not a shareholding but a structure. De Beers held 35.4 per cent of Anglo, Anglo held 32.2 per cent of De Beers, and the family controlled De Beers. Bloomberg later put the family's Anglo interest in 2000 at eight per cent; the figure can only be a look-through, because the direct holding was never that large. The 2001 take-private unwound the cross-holding, and the family put about 10.4 million Anglo shares into DB Investments as part of its price for forty-five per cent of De Beers. Anglo's own notice of meeting records what was left. At 22 February 2005 E. Oppenheimer & Son Holdings held 52,250,206 shares, 3.49 per cent of the 1,493,839,387 in issue. In November 2006, two weeks after Anglo named Cynthia Carroll chief executive, the family sold 17 million of them, a third of the holding, to Larry Yung's China Vision Resources for about £425 million, around £25 a share. Oppenheimer said the proceeds would go into "sectors that are traditionally counter cyclical" to mining. Between 2006 and 2010 about eight million more left the register without an announcement: the 27.3 million shares Business Day later reported as the holding before December 2010 is eight million short of what the 2006 sale left, and those were the years in which the family borrowed against its Anglo shares to meet De Beers' cash calls. In December 2010 it sold 2.1 million at £30.44, within sight of the all-time high; James Teeger of E. Oppenheimer & Son said there was "no intention of selling further shares". In February 2011 Anglo announced that Nicky Oppenheimer would leave the board at the April meeting after thirty-seven years, the first time since 1917 that no Oppenheimer sat on it; the Financial Times reported that his son Jonathan had been "blocked from taking his place for corporate governance reasons". The De Beers sale was agreed nine months later, for cash. The release of 4 November 2011 offers no share alternative and records that EOSIL's 25.2 million Anglo shares, 1.9 per cent, would not be voted on the resolution. Then the last step. In the December quarter of 2016, ten months after Anglo had touched 215.55 pence, the lowest price in its history, the family sold half of what remained, about 12.6 million shares, in a quarter in which the stock traded between R149 and R218, about £11 to £12. Teeger said the sale was driven by valuation. The holding was 0.9 per cent. There is no public record of it after March 2017. .The family's withdrawal from Anglo American, 2001 to 2017. Shares and percentages from Anglo American's notice of meeting of 2005, its release of 4 November 2011 and the reporting cited in the references; prices from the reported proceeds and the contemporary price record; the last column is the author's arithmetic at 3,704 pence, the London price of 30 October 2026, before dividends. The eight million shares of 2006 to 2010 are a difference between two reported holdings, not a reported sale..Three prices and one direction. The family sold at about £25 in 2006, at £30 in 2010 and at about £11 in 2016. The first two were near the top of their cycles. The third was ten months after the bottom of the worst one, in a year Anglo's shares rose by more than three times and became the best stock in the FTSE 100. A trader does not sell the thing he has held through a ninety per cent fall as soon as it has trebled off the low; he waits for the rest. At the price of October 2026 the 2016 sale alone left about £330 million on the table. Whatever the family was doing, it was not timing Anglo. The three sales have nothing in common but their direction. "If we thought the price was high we would sell them" is the sentence of a holder, not of a believer. The family was not optimising Anglo. It was voting with its feet, in instalments, and the instalments track the professionals' arrival: a third of the holding within a fortnight of Carroll's appointment, the board seat given up when the succession was refused, the De Beers stake within the year, and half the remainder once the 2016 collapse had shown what the professionals' balance sheet looked like. What the cash did not buy. Anglo paid in cash and offered no shares, so the family never had to choose between paper and money at the table. But it had the money afterwards, and Anglo's shares were on the market every day. In August 2012 they traded at about £18 to £19.50. Five billion two hundred million dollars would have bought between 170 and 190 million of them, twelve to fourteen per cent of the company, more than the Public Investment Corporation held, and the largest single shareholding in Anglo since the cross-holding was broken. The family could have rebuilt, with Anglo's own cheque, the position its grandfather had built. It bought none. It sold. What the shares would have done is also recorded, because the record cuts both ways. At the low of January 2016 the stake would have been worth between five and six hundred million dollars, a fall of about ninety per cent. At 3,704 pence on 30 October 2026, with about US$15 a share paid in dividends since 2012, it would be worth between eleven and twelve billion dollars: two and one-tenth to two and three-tenths times the money, five and a half to six per cent a year. That is below the gold line of Section II, which turned the same cheque into about US$12.5 billion, and it was reached only in September 2026, on a record set by the merger with Teck's copper, not by anything the board of 2012 bought. On Anglo paper alone, then, the cash was roughly a draw against the market and a catastrophe avoided against 2016. The family chose neither. It put the money where it could hold the chair: the Temasek venture, the family office, the London investment company whose public portfolio returned about a quarter in dollars in 2025. The fortunes are listed in Section XVII. What matters here is the shape. The family liquidated what it could not control and kept what it could. The grandfather did the opposite, and the opposite is the point. Anglo American became De Beers' largest shareholder in 1926. Ernest Oppenheimer bid for control in May 1927 and took the chair in December 1929, with Rothschild money introduced by Morgan Grenfell, two months after the Crash. He formed the Diamond Corporation in February 1930 as demand disappeared, and he held through the decade in which De Beers' mines were shut and the stock sat in London unsold. He bought control going into the worst years the stone has had, because the chair was going to be his and his son's. His grandson sold all of it, for cash, five months after the best half-year the company had, and sold the founding company alongside it at whatever price the day gave. The symmetry is exact. Ernest bought what he could control at the bottom. Nicky sold what he could not control regardless of the price. The difference between them is not a reading of the cycle; both read it. The difference is who was going to sit in the chair. In 1929 it was an Oppenheimer. In 2011 Anglo had said it would not be. For the reader who holds South African producers: the people who knew De Beers best, and who had known Anglo from the inside for ninety-four years, were the first out of the company that bought it, and they left in the order in which the professionals arrived. Every write-off in this report, the US$6.8 billion of Section X, the mines shut and the brand paused of Section XI, the reported sale at a tenth of the price in Section XIV, came after the family's last Anglo share but one had been sold. The family said nothing. It did not need to. Its register is the statement, and it reads, in nine lines, that the custodians would not be minority holders in a company run by its managers, that they would stay only as partners with the chair, and that when the chair was refused they liquidated and went where they could hold it. The test of whether they were right about Anglo is the record of the next section, and the companion paper on Anglo itself. The family sold Anglo near two highs and near the low, and when it was handed five billion in cash it bought none. It was not trading Anglo. It was leaving it, in the order in which the professionals arrived. Ernest bought control at the bottom of 1929. His grandson sold what he could not control at whatever price the day gave. THE RECORD X. Fourteen years, from Anglo's own accounts In July 2015 the sightholders came to Gaborone, looked at the boxes De Beers had priced for them, and left most of them on the table. The record of De Beers under Anglo is one curve, and the chart draws it. Earnings rose for two years and peaked in 2014. They fell by half in 2015 when the buyers refused the goods. They recovered to about the 2011 level in the two years after the pandemic. Then they collapsed: a profit of seventy-two million dollars in 2023, a small loss in 2024, and a loss of half a billion in 2025. The loss in the first half of 2026 was smaller than the loss in the first half of 2025. That is the one improvement in the series, and the table shows it. The realised price per carat fell by half between 2015 and the middle of 2026. In the third quarter of 2026 alone it was eighty-four dollars, a further fall. Part of that is mix, because De Beers sold cheaper goods in 2026 than in 2015. Part is the market. The like-for-like measure is De Beers' own rough price index. It is published against a base of a hundred at the end of 2006. It stood at sixty-eight in the first quarter of 2026. A diamond, on De Beers' own index, is worth less today than it was before the financial crisis, before the Chinese boom, before anyone had heard of a laboratory stone. That is not a cycle, and Anglo's chief executive said as much in August 2026. An independent series on a different base reaches the same place. Zimnisky's rough diamond price index, set at a hundred on 31 December 2007, rose to about two hundred at the peak of early 2022 and stood at about ninety-eight in the last quarter of 2026, read from his chart. De Beers' own index, on a 2006 base, is at sixty-eight. Two indices, two bases, one finding: the mined stone is priced below where it stood before the financial crisis, and the whole of the post-pandemic recovery has been given back with interest. The Financial Times put the company's own decline in the same terms this report uses, "a fraction of the almost $13bn it was valued at in 2011", with the carrying value at US$2.3 billion and underlying earnings from US$1.4 billion in 2016 to a loss of about half a billion in 2025. Those are Anglo's figures; the table below has them by year. .What De Beers earned and what it got for a carat, 2011 to the first half of 2026. Bars: underlying EBITDA, US$ million, coloured by chief executive. Line: realised rough price, US$ a carat, which moves with the mix sold as well as the market; the like-for-like index is in the text. Sources: Anglo American results 2012–26; De Beers 2011 results..Three things in the table decide the argument. The business Anglo bought was a good one for a decade. It earned more than a billion dollars in seven of the ten years to 2022. It paid for Jwaneng's new cut, Venetia's shaft, a Canadian mine and a Namibian ship. The asset was not a dud, and this report does not pretend it was. The second thing is what happened in 2023. The buyers stopped buying. De Beers had renounced the custodian's role in 2000. It now discovered that it was still the buyer of last resort of its own production. The inventory rose to about two billion dollars by the end of 2024, the largest since 2008. In 2025 the company cut it by nine hundred million dollars, selling stock bought at higher prices at a trading loss of four hundred and twenty-four million. Anglo said in February 2026 that the midstream's stocks had "broadly normalised". The table dates the stock. The stockpile the family had called a burden and sold had come back, unbidden, and nobody was holding it on purpose. .The third thing is the carrying value, and it is the board's. Anglo consolidated De Beers at US$12.9 billion of net assets in 2012. A carrying value is not a market price. It moves with depreciation, investment and distributions. It is tested each year against the cash the business is expected to produce. In Anglo's accounts the test was passed in every year to 2022. It was passed through the 2015 collapse, the fall in the realised price and the arrival of the laboratory stone in a third of surveyed American rings. The first charge, mostly against goodwill, came with the 2023 results in February 2024, after Cutifani had left. Three write-downs in three Februaries total US$6.8 billion, as the schedule sets out. The reported sale price is below the written-down figure, though Anglo describes the carrying value as an enterprise value and the basis of the reported price is unknown, so the two are not strictly like for like. The chart keeps the three kinds of figure apart: what was paid, what was carried, and what is bid. .What De Beers was worth by three different measures, which the chart does not join. First, the prices paid in 2001 and 2011, grossed up. Second, the carrying value in Anglo's accounts at the end of 2012 and after each impairment. Third, the value implied by the price reported for the 2026 sale. US$ billion. Sources: Anglo American; CNN Money, 2001; Bloomberg, 29 July 2026..What did Anglo's shareholders get back? The write-downs do not answer that. An impairment is a mark and a return is cash. The exhibit does the calculation. It takes the US$5.2 billion paid in August 2012 for the forty per cent. It adds forty per cent of every dividend De Beers has paid to its shareholders since, which is the bought stake's share. It subtracts the same share of every dollar put back in. It sets the result against what that forty per cent would fetch on the reported sale terms. Anglo does not publish a year-by-year De Beers dividend, so the exhibit shows every assumption from nil to break-even. The finding does not depend on the assumption. Break-even would have needed De Beers to pay out about US$11.8 billion in dividends in ten years whose whole operating profit before capital spending was US$11.8 billion: every dollar it earned, with nothing left for tax or the mines it built. It built two mines, a shaft and a ship in those years. On any defensible dividend assumption the cheque returned well under half its money and lost between seven and fifteen per cent a year; the next two paragraphs narrow that to about a fifth of the money back and twelve to fourteen per cent, and extend it to the whole eighty-five per cent. That is the number this report puts at the head of the account in Section XVII. The chart shows a valuation contraction. The number in Section XVII is an estimate of the cash return on the cheque, on assumptions that are stated and can be changed. For the reader of any South African company's accounts: a carrying value that has not moved in a decade while the product's price has halved tells you one thing: the test has not been applied with the facts in the room. It does not tell you the asset is stable. .The dividend assumption can be narrowed, and when it is the loss sits at the bottom of the range, not the middle. De Beers does not publish its dividend, but Anglo publishes the segment's capital spending every year, and the two series together show where the US$11.4 billion of operating profit from 2013 to 2025 went. US$6.7 billion went back into the ground and the ship: Venetia's shaft, Gahcho Kué, Cut-9, the Benguela Gem, Lightbox's plant, Chidliak. Finance costs and tax took about a quarter of a billion a year in the profitable years, on Anglo's own segment lines for 2014 to 2016; call it two billion over the period, less in the loss years. The stock absorbed most of the rest: about two billion of rough by the end of 2024, of which nine hundred million came back in 2025 at a trading loss of four hundred and twenty-four million. What was left for the three shareholders over thirteen years was of the order of one to two billion dollars. The bought forty per cent's share of that is US$0.4 to 0.8 billion; with its share of the reported sale proceeds, about US$0.9 to 1.3 billion comes back. That puts the return on the cheque between the first and second lines of the exhibit above: about a fifth of the money back, at a loss of twelve to fourteen per cent a year. This is a reconstruction from the segment figures, not a cash ledger. Anglo does not publish one, and the exhibit is open to anyone who has it. One thing the capital spending is not. It is not a further loss on top of the write-offs. The shaft, the Canadian mine and the plant are inside the carrying value the board impaired, and to add them to the US$6.8 billion would be to count them twice. What they are is the reason the dividend line sits where it does: the asset's own cash was spent on things the next buyer will pay a tenth for. The cheque is not the whole cost, and this is where the account so far has been too kind to Anglo. Anglo did not buy forty per cent of De Beers in 2012. It owned eighty-five, and the forty-five it already held was marked by its own transaction at about US$5.8 billion. The whole eighty-five per cent stood at about eleven billion on the price Anglo itself set, and it is now bid at about one billion. That is ten billion of value gone from Anglo's shareholders, of which six and eight-tenths has been written off and the rest sits in a carrying value the market will not pay. To it must be added what Anglo has carried since the earnings turned: an operating loss of about six and a half hundred million from 2024 to the first half of 2026 and a billion of capital spending in the same period, about US$1.4 billion at Anglo's share, from a business that could no longer fund itself. Anglo does not disclose how it funded that, and the exhibit says so. And the stock's carry: two billion dollars of rough financed for about two years at the group's cost of debt is of the order of two hundred million, before the trading loss already counted in the earnings. The all-in account is in the exhibit. Three different measures are kept apart in it and should not be read as one: the change in value on the eighty-five per cent from the 2012 mark to the reported bid, which is the ten billion; the cash return on the cheque, which is the exhibit above; and the impairment in Anglo's accounts, which is the US$6.8 billion. The second line of the account is a valuation mark, not cash. On those measures it is a loss to Anglo's shareholders of about ten billion dollars on the eighty-five per cent, in cash and value together, before any opportunity cost. The gold line of Section II sits beneath it as an illustration of the alternative, not as a further loss. The number at the head of Section XVII is the cheque's. The number here is the company's, and it is the one a shareholder should carry. .Anglo carried De Beers at or near the 2012 figure through the 2015 collapse, the halving of the realised price and the arrival of the laboratory stone in a third of surveyed American rings. Then it wrote off US$6.8 billion in three Februaries. The asset's own cash went into a shaft, a mine and a plant the next buyer will pay a tenth for, so the cheque got about a fifth of its money back, at a loss of twelve to fourteen per cent a year, on the stated assumptions. On the whole eighty-five per cent, the loss to Anglo's shareholders in cash and value is about ten billion dollars. THE BUILDERS XI. Mellier, Cleaver and Cook: what they sold, what they built, what they broke In 2014 a 29.6-carat blue stone came out of a mine De Beers had sold six years earlier for about a hundred and fifty million dollars. The rough sold in February 2014 for US$25.6 million. The 12-carat polished stone cut from it, the Blue Moon of Josephine, sold at Sotheby's in November 2015 for about US$48 million. The suspicion about any company that has lost nine-tenths of its value under professional management is that the managers sold the good assets and kept the bad. The record says the opposite, and the opposite is worse. De Beers kept the best mines it had. Jwaneng and Orapa, the richest diamond ground in the world, are half Botswana's and were never for sale. Debmarine Namibia, which recovers the highest-value stones per carat in the group, got a new ship. Venetia is the one large mine in which De Beers has the controlling interest: seventy-four per cent through De Beers Consolidated Mines, with Ponahalo holding the rest. It got a two-billion-dollar shaft. What was sold, between 2007 and 2016, was the bottom of the list. The old South African mines went to Petra Diamonds, Namaqualand to Trans Hex, and the Kimberley tailings to a local consortium. The sales brought in about four hundred million dollars all told. The buyer did not prosper. Petra's shares have lost ninety-nine per cent of their value since 2014 and its bondholders took the company in 2021. The one exception tells the story. Cullinan is the only mine in the world that produces large blue diamonds with any regularity, and that was known when it was sold. Several of its stones have since sold for more than the mine cost. In 2020 and 2021 De Beers itself, in partnership with Diacore, paid US$80.6 million for two parcels of Cullinan blues from Petra; how the partners split the cost is not disclosed. Exceptional stones do not make a mine's economics, and the sale may still have been right after costs. But the company that sold the mine was, twelve years later, a buyer at auction prices of the stones it produces. Where the money went is the other half of the account, and it went into building. Snap Lake in the Canadian Arctic, De Beers' first mine outside Africa: nearly two billion dollars spent, never profitable, written down in 2015 and flooded in 2017. Voorspoed in the Free State: built in 2008, closed in 2018. Chidliak, bought in 2018, suspended. Lightbox, the laboratory brand, with a ninety-four-million-dollar plant in Oregon: the brand closed in 2025, and what became of the plant is in the table. Gahcho Kué, the second Canadian mine: De Beers took its partner's half for its debts on 16 October 2026, four years before the mine closes. And Venetia underground. The De Beers and Anglo boards approved it in 2012. It was launched in October 2013, in the first year of Anglo's control. The spend has passed two billion dollars, with nearly a billion to go. First ore came in 2023 and output has not risen since. Anglo now puts steady state in "the mid-2030s". De Beers paused it for two years on 13 July 2026. The pause is a decision about price, not a verdict on the rock. The pause has a headcount, and it is the first figure in this report that is not a dollar. De Beers concluded its union consultations in mid-September 2026 and told its 3,500 direct employees at Venetia what their status would be. The contract workers, thousands of them on the Financial Times' count, were not consulted; they wait to hear whether a one-year contract is renewed. Venetia is South Africa's last major diamond mine. For two years the country in which Rhodes amalgamated the fields will mine no diamonds at scale, and a two-billion-dollar shaft will stand still under a market that will not pay for its ore. A shop owner opposite the gate, who has watched the mine for more than thirty years and whose trade has slowed "to a trickle" in the past few months, said: "It was good years ago. Now, you can't even compare." She is describing the custodian's absence from a distance of a hundred metres. The shaft's delays are one thing. The market that will not pay for its ore is another. The table keeps them apart. About four hundred million dollars came in from everything sold. About four billion went out on things that are shut, flooded, suspended, closed or paused. The professionals did not sell the family silver. They melted it down to build things that did not work, or that work in a market that will not pay for them. .The three men who ran it in those years were chosen by the owners, and each is scored on his own decisions. Philippe Mellier came from Alstom's train division. The family chose him and appointed him from 1 July 2011, before the sale was announced. Nicky Oppenheimer said "we had the feeling this was the time to do something different". He said Mellier had "no mining experience, no experience of diamonds and no experience of southern Africa". Anglo kept him. He delivered the record year of 2014 by pushing rough prices up faster than polished. In 2015 the buyers refused the goods. Most of the July sight went unsold. De Beers cut prices by about a tenth in August and let clients defer three-quarters of their allocations. The company sold a third less that year. His view of his customers, January 2015: "I have a business. I am not a banker. I am not a supporter of customers." A reasonable view for a company that sells trains. It is the opposite of the one De Beers was built on: the whole chain had to make money, or the price would not hold. Bruce Cleaver was a lawyer who had been at De Beers for eleven years. He made the two large bets of the decade. One was Lightbox, which the next section scores. The other was Botswana, which took five years and was signed by his successor. He was candid in 2023 that "bridal has been affected" and that "significant volumes of sales" were laboratory stones. That was more than the company had said while he ran it. He then tried to buy De Beers in 2025 and withdrew in July 2026 because he could not see "an appropriate return on investment over the short term". A man who ran the company for seven years looked at it as a buyer and declined. Al Cook, a geologist from BP and Equinor, inherited the collapse in 2023. He has managed it as an oil man would: cut the overhead, close what does not pay, settle with the host government, and tell the market that supply is running out. The last is true, and it is the mined stone's best argument. His job since May 2024 has been to sell the company. He is not responsible for the purchase, the stockpile, the advertising, Lightbox or Venetia. He is responsible for the price at which they are all being sold. A board that hires from outside the trade is buying a view. The view it bought in 2011 was that a customer's margin is the customer's problem. That is the one view a custodian cannot hold. A railway executive, a lawyer and an oil man. The first told the buyers their margin was their own problem. The second built a brand to sell the stone the company said would never matter. The third is selling the company. None of the three had sold a diamond before he arrived. THE STONE XII. What De Beers said about the laboratory diamond, and what it did On 17 November 1959 Harry Oppenheimer called a press conference in Johannesburg to announce that De Beers had made a diamond. The company that was beaten by the laboratory stone could make the laboratory stone, and had been able to for sixty years. The first synthesis was in September 1958 at the Adamant Research Laboratory in Johannesburg. The Springs plant was running by 1961. The industrial business was renamed Element Six in 2002 and is, in its own words, a global leader in synthetic diamond. Every reactor in Surat and every press in Henan is a descendant of work De Beers paid for. That fact changes how the rest of this section reads. Whether the stone could be made was never in doubt. Who would sell it, at what price, and under what name, was. They saw it. That is the finding. The company's own words and dates, set out in the table, make it. De Beers' finance director described the laboratory stone accurately in 2007. The company instructed its clients in 2013 never to call it a diamond. Its head of brands, Stephen Lussier, had run its marketing for a generation. He told JCK in June 2016 that the stone was not a threat "today", and that "it is not inherently rare, and as a result, it won't have enduring value". Asked whether De Beers' own Element Six would ever make gem stones, he said: "I don't think so". Twenty-three months later De Beers launched Lightbox, a brand of laboratory jewellery made by Element Six. It announced a ninety-four-million-dollar plant to grow two hundred thousand carats a year. Bain, the company's own consultants, forecast in 2019 that the stone would take five to fifteen per cent of the market's value by 2030. It was inside that range by 2022. That does not disprove the forecast, but it does say how fast the lower bound arrived. The Knot surveys American couples every year. It put the stone in twelve per cent of surveyed engagement rings in 2019 and sixty-one per cent in 2025. The survey is of couples who answer it, not a census. The trend is what matters. The warning was older than Lussier's interview, and it came from the company's own consultants. According to two people with knowledge of the work, Bain & Company advised De Beers during its review of 2000 to 2002, the review that produced Supplier of Choice, not to do anything that would legitimise laboratory-grown stones. Set that beside Section VI. The same review, in the same two years, told the family it could stop holding the price and told it never to bless the substitute. The family took the first advice because it saved money. The professionals broke the second in 2018. It cost nothing to keep, and nothing they have published shows they knew it existed. "The cataclysmic thing that happened was De Beers launching Lightbox," one former executive said in November 2026. It "legitimised" the category. "It was catastrophic." De Beers was "surprised" by the retailers' appetite for the stone once Lightbox had put the company's name on it, according to a former executive with knowledge of the rollout. The retailers piled in and stayed, "even if it hurt their natural [diamonds] business", because the margin was theirs. That is the margin exhibit below, seen from the shop floor two years before this report priced it. The summary this report has been working towards was given by one of them. "De Beers didn't automatically have a right to exist. It created a right to exist via the mystique it created around diamonds. As soon as that mystique was shattered, it was nothing more than an intermediary." An intermediary is what Anglo bought in 2012, on a mining multiple, and did not know it. The mystique was the thing the family had already sold. "There wasn't enough marketing, yes," another said. "But you can only market something if you know what you're selling." Nothing in Anglo's record shows that it knew. That is the finding of Section VII, in the trade's own words. The consultants who told the family in 2000 that it could stop holding the price told it in the same review never to legitimise the laboratory stone. The family kept the advice that saved money. Anglo broke the advice that cost nothing, and nothing on the record shows it had been told the advice existed. De Beers was right about the price. The wholesale price of a laboratory stone has fallen by about nineteen-twentieths since 2018. India's average export price for a polished one fell from nearly four hundred dollars a carat in 2022 to about seventy in 2026. Cook put the cost of making one at thirty to a hundred dollars. What De Beers got wrong was the conclusion. It assumed a cheap stone would stay a small one. The stone's price collapsed and its share rose at the same time. That is what happens when a product becomes cheap. The mined stone did not get the customer back. Its own like-for-like index fell by a third from the 2022 peak while the laboratory stone was falling by nine-tenths. The Knot's figures show the average natural engagement ring unchanged in size and the average spend on all rings falling. .The stone they said would not matter. The chart shows the share of surveyed American engagement rings with a laboratory-grown centre stone (The Knot) and the wholesale price of a laboratory stone against 2018 (Golan index). The dates of De Beers' own statements are marked. Sources in the references..Lightbox is the episode in one brand. Cleaver launched it on 29 May 2018 at eight hundred dollars a carat, a tenth of the mined price, as "affordable fashion jewelry that may not be forever, but is perfect for right now". Five years later he told JCK: "it's one of the scariest things I've ever done. If we launched this product too soon, we may legitimize a whole category quicker than we should have." The price was cut twice. Production for jewellery stopped in May 2024. On 8 May 2025 De Beers announced the closure of the business. Cook, asked whether it had met its goals, said: "Kind of." Avi Krawitz's verdict in August 2026: "In hindsight it was probably their biggest mistake. The cat is out of the bag." Then the admission. On 4 August 2026 Duncan Wanblad presented results in which De Beers had lost money again. The Financial Times asked him whether the company had misread the stone. "Hindsight would probably show that we might have wanted to have been a little bit more aggressive on the signals we were getting," he said. "Mostly diamond markets, when they go through dips, recover very rapidly. And I think that there was a full expectation that might be the case here." And: "Looking in the rear-view mirror is a much more perfect science than trying to do it in real life." The signals were not in a rear-view mirror. They were in the company's own client note of 2013 and its own marketing director's interview of 2016. They were in its own plant of 2018 and its own consultants' forecast of 2019. They were in the survey every jeweller in America reads each year. The company saw the stone. It said in writing that the stone would not matter. It sold the stone, stopped, and then said it had not been warned. The exhibit shows where the margin went, and that is why the stone won. A diamond has a chain. The mine sells rough, the cutter makes polished, the dealer sells it on, the retailer sets it. The exhibit takes the industry's own pipeline figures for the whole chain, in 2011 and at the 2022 peak, and sets a one-carat stone beside them. The pipeline figures are the only matched series. On them, polished sales were about 1.6 times rough sales in both years, and retail sales about three times polished. The ratios between the rungs barely moved while the chain grew. Since 2022 every rung of the mined chain has roughly halved. The money did not move up or down the mined chain. It left it. At retail, a one-carat laboratory stone sells for about a seventh of a mined one of ordinary grade, on the medians the price aggregators publish, and costs its maker less than a hundred dollars a carat. Edahn Golan put the retailer's gross margin on it at about sixty-four per cent against forty on a mined stone. That is why Signet and Pandora sell it. The one-carat figures in the exhibit are not a matched basket: each is the best dated figure at its stage, and the exhibit says which. A producer that does not know where the margin in its chain sits cannot defend it. The platinum producers have the same chain, from the smelter to the showroom, and the same blindness. .De Beers made the first laboratory diamond in Johannesburg in 1958. In 2013 it told its clients never to call one a diamond. In 2016 it said it would never make one for a ring. In 2018 it built a plant to do so. It closed the brand in 2025, and in 2026 it said that the signals were missed.THE PRECEDENTS XIII. The pearl, the nitrate, the indigo, the feather and the tin In 1921 the jewellers of Paris put notices in the press warning the public that the pearls a Japanese firm was selling were counterfeit. In 1924 a court in the Seine told them to stop. De Beers is not the first natural monopoly to be broken. The industry treated the laboratory stone as unprecedented, and that habit is part of why it was so slow. The precedents are in the table. They do not all share one mechanism. Three were broken by a factory. One was broken by fashion. One broke itself. Each is used here for what it shows. The natural pearl was the diamond of its day, and the Gulf was its Botswana. Kokichi Mikimoto had patented a method of culturing round pearls in 1916. In the 1920s he sold them in Paris and London at a fraction of the natural price. The Paris trade said they were not real. Mikimoto's agent sued. In 1924 the Tribunal civil de la Seine found the campaign to be unfair competition and left the science to the scientists. The scientists settled it the same way within the decade. Bahrain's pearl exports fell by more than two-thirds in a year at the turn of the 1930s. Its fleet went from three thousand boats to three hundred. This is the first mechanism, and it is De Beers' own. A factory makes the same product, and the market does not care who made it. Chilean nitrate was the state's diamond, and the second case of the same mechanism. Nitrate taxes were about half the Chilean government's revenue for forty years. In September 1913 BASF opened the first Haber–Bosch plant at Oppau and made nitrogen from the air. Chile's exports fell by seven-eighths between 1929 and 1932. The state formed a joint company with the Guggenheims in 1931 to hold the price. It collapsed within two years. A state that lives on a mineral cannot hold its price against a factory. The attempt costs it the treasury it was trying to protect. Botswana should read it. Indian indigo was the planter's diamond, and the third. In 1897 BASF, again, put a synthetic indigo on the market. World output of the natural dye fell from about nineteen thousand tonnes to about a thousand within fifteen years. The planters campaigned that the synthetic was inferior. It was not, and the campaign did nothing. A natural product with a chemical identity is a chemical, and the chemist wins on cost. Oudtshoorn's ostrich feathers were South Africa's own case, and the one broken by fashion and not by a factory. In 1913 feathers were the country's fourth export. In 1914 the fashion turned, the motor car took the hat off the head, and the war closed the markets. The trade never recovered. The mechanism is different. The lesson is the one the diamond industry should hear most clearly: a luxury whose value is a story can lose the story in a season. Tin was the cartel's diamond, and it broke itself. The International Tin Council held the price of tin for thirty years through a buffer stock and export controls, in the way De Beers held diamonds. On 24 October 1985 it ran out of money to buy. It defaulted on its banks and brokers for about nine hundred million pounds, and the London Metal Exchange suspended tin trading until 1989. The price halved. No factory was involved. The mechanism is the one the Oppenheimers understood in 2000 and Anglo did not understand in 2011. A buffer stock works only for as long as the holder has the capital and the will to keep buying when nobody else will. The day it stops, the market finds out what the stock was worth without the buyer. Of the five, the one South Africa's platinum producers should study is not the pearl but the tin. A producer that cuts output to hold a price without a funded buffer is running the Tin Council's book without the Tin Council's money. .Three of the five fell to a factory, one to fashion, and the tin cartel to its own balance sheet. De Beers faced the factory with the balance sheet of the Tin Council after 1985. Its owners had sold the buffer before the factory opened. THE SALE XIV. Penny buys at the bottom with other people's money, and what it would cost to hold the price On 17 July 2026 Botswana's minister stood up in parliament in Gaborone and told the country who was buying the company it lives on. He did so before the seller had. The sale is not signed as this report goes to press, and the terms are Bloomberg's. The minister said that Anglo had run "a competitive process involving three short-listed bidders" and chosen the Global Diamond Consortium, led by Gareth Penny. The consortium's money comes from the diamond trade, with the Antwerp house Pluczenik among those confirmed, and from the governments of Angola and Namibia. Richemont and Signet were reported in October to be in talks to join. The reported price is about a billion dollars for Anglo's eighty-five per cent, three-quarters upfront, with earn-outs and half a billion of new capital from the buyers. It has not been reported whether the billion is an equity value or carries the company's debts, leases, closure obligations and the former Canadian partner's liabilities with it. The table marks what is known and what is not. Botswana, with first refusal, is expected to go to about a quarter. The African states together are to stay below half, which the private backers required. Anglo's third-quarter report of 20 October said only that it would "continue to work towards a transaction in a challenging market". Cleaver's rival group had withdrawn in July. If the billion is for the equity, it values De Beers at about US$1.2 billion. On a simple gross-up that is about a tenth of the thirteen billion the 2011 price implied, a fall of about ninety-one per cent in fourteen years. The Oppenheimer forty per cent, bought for US$5.2 billion, would be worth under half a billion. Those are the seller's losses, and Section XVII counts them. This section turns to the buyer. Is he being handed a hospital pass? The table sets what he gets against what he takes on. Penny has run the company through a worse year than this one. He was chief executive in 2009, when sales halved and the company nearly ran out of cash. He brought it out with the debt cut and the mines open. He negotiated the predecessor of the Botswana agreement he is now buying into. On price, he is buying from a seller who has written the asset off and wants to be rid of it, with the traders' and the governments' money. On price alone that is the Oppenheimer method. Whether it is the Oppenheimer method on anything but price depends on what the consortium intends to do, and what that would cost. That is the rest of this section. .The consortium has not published what it would cost to hold the price. This report does the calculation in its place, and states every assumption so that the reader can change it. Take the company as it is. De Beers' guidance for 2026 is 25 to 28 million carats, raised in October when it took full ownership of Gahcho Kué. Not all of that is De Beers' to sell: Botswana's own company takes a rising share of Debswana's output, and in 2025 De Beers sold 20.9 million carats against 21.7 million produced. Call De Beers' own saleable volume about 24 million carats. At the first-half realised price of US$105 a carat, a year's rough is worth about US$2.5 billion. A custodian holds supply back until the price recovers. Keep the mines running, sell two-thirds and warehouse a third for two years, and the receipts forgone are about US$1.7 billion at today's price. The stones go into a vault that already held about US$1.4 billion at the end of 2025, by the author's arithmetic from Anglo's statements. That stock is financed already, by Anglo, and the buyer takes its financing over rather than funding it twice. The new stock builds up over the two years, so the interest on it, at 8 per cent, a fair rate for a private diamond company with a loss-making record, is about US$0.14 billion. The mines themselves lose money even at full sales: about US$0.2 billion a year on the first-half 2026 run-rate, which had trading profits on the old stock in it, and US$0.5 billion on 2025's record, which carried a US$424 million trading loss. Over two years that is US$0.4 to 1.0 billion. The custodian's bill for two years is therefore about US$2.3 to 2.8 billion before a cent of marketing: call it US$2.5 billion. The consortium's reported new capital is half a billion. The gap is about US$2 billion. That number, not the headline price, decides what kind of owner the consortium can be. .The second number is the free rider. The Oppenheimers controlled more than four-fifths of the world's rough when they held the price. So four-fifths of the benefit of every carat withheld came back to them. De Beers today sells about a quarter of the world's carats. It is the largest seller by value, with Alrosa its equal. Every carat it holds back props up Alrosa's price, Angola's, Canada's, and the Russians' sanctioned goods moving through Dubai and India. On a share of about a third, roughly seven dollars in every ten of price support go to competitors. A custodian at a third is paying for a party to which everybody else is invited. The third number is the ceiling, and it is new. When Harry Oppenheimer bought stones in 1982 there was nothing above the price he was defending. A buyer who wanted a diamond paid his price or waited. Today a one-carat laboratory stone sells at retail for about a seventh of the price of a mined one of ordinary grade. Its maker's cost is thirty to a hundred dollars a carat. A stockpile can hold a price up. It cannot hold it above a substitute that is the same product at a seventh of the price. My judgement, and I give it as mine: the price of 2022 will very probably not return, whatever the buyer does. The ceiling is now set in Surat and Zhengzhou and not in London. That leaves the consortium three plans, and the table scores them. The harvester runs Jwaneng, Orapa and the Namibian ships for cash at whatever the price is. It spends nothing on the story, lets Venetia sleep, and pays the governments and the traders their dividends. It needs the half-billion and little more. At a billion for the company it very probably works. It is the bargain of the decade for the people who put the money in. The brand house takes the Hermès model the bear raises in Section XXII. It owns the name, the jewellers and the top of the market, controls no supply, and spends heavily and steadily on the story. It needs a marketing budget in the hundreds of millions a year for a decade and a retailer's patience. That is why Richemont's and Signet's names are in the room. The producers' council puts Botswana, Namibia and Angola inside one selling organisation with De Beers. It is the closest thing to a custodian on offer. It needs the states to agree to withhold their own revenue in a bad year, which Chile's nitrate cartel of 1931 could not do for two years. It also needs a competition lawyer in every jurisdiction it sells into. On 2 November 2026 the Financial Times reported, on one person familiar with the matter, that Richemont, which owns Cartier, "was part of the consortium". Not in talks: in. The report is unconfirmed by either company and is given that weight. If it holds, it moves the reading of the three plans by one degree. A jeweller in the room is a vote for the brand house, because a jeweller buys the name, the top of the market and the story, and has no use for Jwaneng's run-of-mine. It is not a vote for the custodian, and it is not money for one. Cartier does not hold stock to defend a rough price; it buys polished at whatever the price is. The same report has De Beers itself saying, in a statement, that it has seen "growing industry-wide recognition of the critical importance of a shared approach to natural diamond marketing through collective investment", with early efforts "delivering promising results". That is the company asking in 2026 for the thing its owners shut in 2010 and its buyers never restored. Anglo declined to comment. The bear's fifth point, that the mines will be worth a great deal when the market turns, has a witness. Zimnisky said in November 2026 that "a swath of mine closures has taken out a chunk of natural diamond supply", so that any uptick in demand "could result in a swift rebound in prices". Argyle is shut, Venetia is paused, Canada is closing, and Section XI has the list. The point is fair and the report concedes it in Section XXII. But the arithmetic of the free rider above says who collects a rebound: on a share of about a third, seven dollars in ten of it goes to Alrosa, Angola and the Russian goods in Dubai. Supply running out is the harvester's hope. It is not a custodian's plan, and the consortium has published neither. .The verdict: for a custodian it is a hospital pass; for a harvester it is the bargain of the decade. The consortium has the money for the second and has not shown the money for the first. Its only public remark since the bid is Penny's, on a podcast in June, on the duties of a chairman: "The chair should be the last to speak." For the South African reader: the harvester's plan is the one every hired manager on the platinum belt is running now. It pays its owners well at the bottom of a cycle. What it never does is defend the price. A country that holds seven-tenths of a metal and leaves its price to the harvesters has chosen the plan without knowing it. For a custodian it is a hospital pass; for a harvester it is the bargain of the decade. The custody would cost about US$2.5 billion over two years. The consortium has shown half a billion. There is one more thing the bidder had in front of him, and it may be the best reason of all to bid low: the seller's record. Anglo American has sold or given away eight large assets since 2001. Every one of them is worth more today than Anglo received, on the same shares or the same mines under their new owners. Most are worth two or three times more. The chart puts the eight side by side. The gold mines of the Free State went to Harmony in 2001 and 2002 for US$339 million. Harmony's share of them is worth US$1.9 billion. A fifth of Gold Fields went to Norilsk in 2004 for US$1.2 billion and is worth US$3.5 billion. The AngloGold stake went in three pieces, in 2006, 2007 and 2009, the last to Paulson at US$32 a share. US$5.2 billion was received; the shares are worth US$10.8 billion today. Rustenburg and Kroondal went to Sibanye for US$0.8 billion and are worth US$1.6 billion. Thungela was given away with a cheque attached and is worth US$2.5 billion with the dividends since. A fifth of Valterra was placed at R845 in 2025 and trades at R1,165. In all, US$9.9 billion was received for what is worth US$24 billion. That is before the US$2.6 billion of dividends the sold shares have paid their new owners. De Beers itself is the ninth line. It was bought at US$5.2 billion for forty per cent in 2012 and is reported to be selling at about a billion for eighty-five per cent in 2026. Two cautions on the table. Thungela was demerged to Anglo's own shareholders, so its value went to them rather than away from them. And the Harmony and Sibanye figures are estimates from the buyers' present earnings, not observed prices for unchanged mines. .The seller's record. What Anglo American received for each large asset sold or given away since 2001, against what the same shares or mines are worth at the closes of 22 October 2026, US$ billion. The exhibit is reproduced from the author's audit of Anglo American, with its sources. It is included here because a bidder for De Beers in 2026 had the same record in front of him. .Read that table as a bidder would. Here is a seller that has never once held out for price, under four chief executives and three chairmen. Its board sold gold at a fifth of its price and platinum a year before its earnings multiplied. It announced in May 2024 that De Beers would be sold or demerged, and said in October 2024 that there was "no way De Beers is a fit in the portfolio". The three bidders knew the seller was leaving before they knew anything else. A buyer facing that seller does not need to bid against the asset's value. He bids against the seller's need to be gone, and he knows from the record that the seller will take it. The family, in Section VIII, sold to a buyer that did not know what it was buying. Penny is buying from a seller that did not know what it was selling, and the record says it has never known. That is the incentive. Anglo's own history was the strongest card in the consortium's hand, and the seller dealt it. Anglo has sold eight large assets since 2001. Every one is worth more today than it received: US$24 billion against US$9.9 billion. A bidder for De Beers did not need to value the stone. He needed to value the seller's record, and it was public. BOTSWANA XV. The country that lives on diamonds buys more of the company that sells them On 1 March 1967, five months after independence, a De Beers geologist named Manfred Marx found kimberlite at Orapa. The country then had twelve kilometres of paved road, twenty-two university graduates and an income per head of about seventy dollars a year. Botswana is the only country in the world whose economy is a diamond company, and the chart is the argument. Diamonds are about seven-tenths of its exports and a third of government revenue in a normal year. Debswana's sales fell by more than half between 2022 and 2024. Mineral revenue averages about twenty-five billion pula a year. It is budgeted at about ten billion for 2025–26. The economy shrank in 2024 and again, slightly, in 2025. The deficit ran at seven to nine per cent of output for two years. The reserves fell from US$7.5 billion in 2017 to under four billion. Moody's cut the country in October 2025 and S&P in March 2026, to its lowest rating since the agency began covering it. The state's own stockpile of unsold rough stood at twelve million carats at the end of 2025, nearly twice its allowable inventory. .Botswana's exposure: Debswana's rough diamond sales, 2022 to 2024, US$ billion, and mineral revenue as a share of national income, 2022 and 2024. Sources: Debswana and De Beers releases; Bank of Botswana and IMF figures cited in the references. .Into that, the government of Duma Boko, elected in November 2024, has decided to buy more of De Beers. The minister of minerals said in July 2025 that Botswana wanted "full control over this strategic national asset and the entire value chain including marketing". He said that any sale "without our support will be difficult to achieve", and that financing "is not an issue". The International Monetary Fund's mission in November 2025 "cautioned the authorities against increasing their stake in De Beers". The president answered on 15 December: "We want to take a bigger share and become the real owners. The IMF are not Batswana, they are not voters.We said we want De Beers, and we are going to take it." He wanted the headquarters moved to Gaborone. By October 2026 the reports were that Botswana would settle for about a quarter. That is the stake it had declined for US$1.25 billion in 2012 and can now have for a fraction of that. The finance minister is more careful than the president, and the care matters. Ndaba Gaolathe told the Financial Times in September 2026 that the government remained intent on increasing its stake, "but that did not necessarily mean putting more public money into the business". It was looking for a partner with a "modern strategy" to make the company profitable again. In recent years, he said, the market had become "nervous" and uncertain about whether De Beers would "be willing to put as much money as they have been putting [historically] into branding". That is Section XIV's cost, stated in Gaborone: the minister wants the story told again and does not want to be the one who pays for it. Whoever does pay will be paying for the rebound on Botswana's stones. In December 2025 the IMF warned that a recovery in the diamond market was "uncertain" and urged the country not to raise its fifteen per cent, a month after its mission had said the same. A businessman in Gaborone who works in the trade and with the government described the mood in one word, "panic": "Everything rested on diamonds, everything. Now there's just no money any more." The country is being asked to buy more of the thing it lives on, at the bottom, with borrowed money, from a seller who is leaving, into a consortium whose plan it has not seen. Ernest Oppenheimer did that in 1929. He also knew the plan, because it was his. Botswana is not choosing between a royalty and the channel. It already holds a great deal of both. It has held fifteen per cent of De Beers since 2004 and half of Debswana since 1975. Under the 2025 agreement it also sells a rising share of Debswana's run-of-mine through its own company, to half by the end of the term. It has three courses, and the table prices them. It can hold what it has and take its royalties, taxes, Debswana dividends and the fifteen per cent. That costs nothing and leaves the marketing to whoever owns the rest. It can build its own channel through the Okavango Diamond Company. It is already doing so, and the company failed to sell a parcel in September 2025 because the buyers would not pay. Or it can buy more of the group. That puts borrowed money into the equity of a loss-making company whose profits move with the diamond price more than the country's own revenue already does. It also buys exposure to Venetia, Canada and Element Six, none of which is Botswana's problem today. On the arithmetic of a nine per cent deficit the IMF is right. On the model, Boko is right on one condition only: that the consortium he joins intends to rebuild the channel and the story, and funds it. Section XIV costs that at about US$2.5 billion. If the consortium intends to harvest, the quarter he is buying is a quarter of a harvester. He would have done better to keep his money and his run-of-mine. Nothing published by the consortium says which it is. On the capital it has shown, the harvester is the likelier. Ernest Oppenheimer made Boko's choice in 1929, with borrowed money, in a depression. He also controlled the company he was borrowing to buy. For the policymaker: a state that wants the channel should buy control of the channel or build its own. A minority stake in somebody else's harvester is neither. .Botswana is the ground. It already holds a seventh of De Beers, half the mines and a rising share of its own stones. Whether buying a quarter of the rest is Ernest Oppenheimer's choice or a quarter of a harvester depends on a plan nobody has published. THE PEOPLE XVI. The names, the decisions, the pay, the verdict Nicky Oppenheimer stood at Vergelegen on 30 November 2011, at the farewell dinner Anglo gave the family, and read out the sentence his grandfather had written in 1954. The scorecard that follows places each person against the decisions that were his. It gives the pay where it is public in Anglo's remuneration reports, and a verdict in one line. Three things are kept apart in it. What the record shows each person did is in the third column. What each said about it is quoted. What I conclude is in the fourth column, and it is mine. De Beers' chief executives' pay has never been disclosed. Anglo's chief executives' pay is for the whole group, of which De Beers was one division, and is given as that. .On the Anglo side the responsibility runs upward. It is responsibility for oversight, which is what a board is for. This report does not claim to know who originated each decision inside the room, or what any one of them privately understood. Carroll and Parker bought at the top, from the people who set the price, on the brokers' assurance that it was cheap. Carroll's departure was announced within a month of her taking the De Beers chair. Cutifani chaired De Beers for nine years. In those years the shaft approved before he arrived was built and did not deliver. The brand that proved the company's own argument wrong was launched. A train executive was replaced by a lawyer. The business as a whole passed its impairment test in Anglo's accounts every year. In the same years the realised price per carat fell by a third, and the laboratory stone went from a curiosity to a third of surveyed American engagement rings. Wanblad took three impairments. He decided within two years of arriving that the company did not fit. He is trying to sell it in the worst diamond market since 2009, at a reported price his own board's accounts do not yet reflect. He is the only one of the four who has said on the record that the signals were missed. He is also the only one who has had to, because he is the one holding the asset when the music stopped. A board that cannot say what its own division is for cannot test whether it is still worth what it paid. South Africa's listed miners carry assets today that have not been tested with the facts in the room either. Cutifani chaired De Beers for nine years and the business was not impaired while he did. Wanblad took three write-downs and said the signals were missed. The man who is buying it ran it for the family. THE FORTUNES XVII. The money made and the money lost, 2000 to 2026 Kaushik Mehta's company, Eurostar, was one of the largest polished dealers in Antwerp. In November 2018 it told its banks that the stock it had valued at a hundred and thirty-two million euros two months earlier was worth twenty-five million dollars. It was bankrupt by March. The paper so far has scored decisions. This section scores the money. A reader who finishes a report on De Beers should know who got rich and who was ruined. The account has two columns. .The pattern in the two columns is the one the report has been building to. The people who made fortunes read the cycle and held cash, or sold the substitute. The family sold at the top and has said nothing since. The growers sold while the price was high. The retailers sold the stone the growers made, at margins the mined stone never gave them. The people who lost fortunes believed the story: that the price would hold, that the custodian would be there, that the stone kept its value. The dealers of Antwerp and the cutters of Surat carried, on borrowed money, the inventory De Beers no longer held. In 2019, and again in 2024, they found what Mehta found: a stock of diamonds is worth what the next buyer will pay for it and nothing else. Botswana did best of anyone from the story for fifty years, and is the one party that cannot sell. Anglo's shareholders paid for the right to be the custodian and were never told what the word meant. They paid most. The next two sections set them beside the people who paid at retail. For the reader who owns South African producers: the fortunes in this account were made on the way down as well as the way up. They went to the people who understood what the price rested on. The losses were taken by the people who thought the price was the thing. The fortunes were made by the people who read the cycle and held cash, or sold the substitute. The losses were taken by the people who believed the story: that the price would hold, that the custodian would be there, and that the stone kept its value. NOT A REAL DIAMOND XVIII. The first claim: "not a real diamond" In December 2013 De Beers sent its clients a note. It instructed them never to use the word "diamond" for anything that did not meet the definition. Five years later the American regulator changed the definition. The industry has answered the laboratory stone with two claims. This section takes the first. The next takes the second, with the people who believed it. The claim is that the laboratory stone is not a real diamond. The industry does not say those words in a regulated market because it cannot. A laboratory-grown diamond is carbon in the same lattice as a mined one, with the same hardness and the same chemistry. The laboratories need instruments to tell them apart, and De Beers sells the instruments. In July 2018 the United States Federal Trade Commission removed the word "natural" from its definition of a diamond. The definition had read "a natural mineral consisting essentially of pure carbon crystallized in the isometric system". It now reads the same without "natural". The Commission required only that a laboratory stone be disclosed as such. In April 2019 it sent eight marketers warning letters about disclosure. De Beers' 2013 instruction was overtaken by the regulator five years later. The industry's campaigns since then use "real" and "natural" as synonyms. That is the claim made by implication. On 22 April 2021 the National Advertising Division, the American advertising industry's self-regulatory body, examined the Natural Diamond Council's campaign on a challenge from a laboratory grower. It recommended that the council discontinue the implied claim that laboratory stones are not real diamonds. It recommended the same for three specified claims: that laboratory stones have "no resale value", that mined diamonds are "becoming rarer every day", and that laboratory stones carry three times the carbon emissions. The council agreed to comply. The Division did not say that a mined diamond may not be called rare. It said that the specific claims had not been substantiated. In June 2025 the Gemological Institute of America announced that it would stop applying its colour and clarity scales to laboratory stones. It would describe them instead as "premium" or "standard". Its reason, in its own words, was that more than ninety-five per cent of them "fall into a very narrow range of color and clarity". It continues to identify the material as laboratory-grown diamond. The Institute's stated reason is quality distribution. Whether the change also serves the mined stone commercially is my reading, not the Institute's. What is left is a distinction maintained by vocabulary after the chemistry and the law had stopped maintaining it. The Paris jewellers made the same claim against Mikimoto and lost in court in 1924. The accurate version is the one De Beers' own marketing director gave in 2016: a laboratory diamond is a diamond that is not rare. The company has spent the years since avoiding the second half of that sentence. The second half concedes that rarity, not the stone, was always the product. One caution on the numbers that follow in this report. A rough price and a polished price are different products at different stages. The comparison that matters is the one in Section XII's exhibit: polished against polished, at retail against retail. There the laboratory stone sells at about a seventh of the mined one. A South African producer whose product has a chemical identity, as platinum does, should assume that the chemist and the recycler will eventually match it. It should plan for the day the regulator says so. A laboratory diamond is a diamond that is not rare, and since 2018 the law says so. The advertising regulator made the industry's own council withdraw the claim that it is not real in 2021. The product was never the stone. It was the scarcity. THE BELIEVERS XIX. The second claim, and what a store of value that wasn't cost the people who held it In 1970 a London consumer magazine bought four hundred pounds of diamonds. In 1978 it tried to sell them. The best offer it received, after inflation, was worth a hundred and sixty-seven. The second claim is that a mined diamond holds its value. This one can be priced, because people believed it and paid for it. The Natural Diamond Council's advertising said it until the regulator made the council withdraw it in 2021. De Beers' campaigns since 2023 say it in softer words. Cook's phrase is a "store of emotional value", which is careful. The evidence runs the other way on every scale the industry publishes. The table sets it beside the metal. The private holder first. Edward Jay Epstein documented the resale gap in The Atlantic in February 1982, and it has not closed. In his tests a stone resold a week after purchase lost three-fifths of its price. A thief who stole diamonds insured for fifty thousand dollars got two hundred for them. In 2013 a man who tried to sell back a ring was offered at best thirty-five per cent of what he had paid. In 2023 a price aggregator found cash offers for diamonds "forty to fifty per cent below retail". A one-carat laboratory stone is worse, and pawnbrokers will not pay a tenth of its price. The worked example is in the table. A one-carat stone of middle grade bought in November 2011 cost about six and a half thousand dollars on the trend data of the day. That figure is an estimate. The same stone at retail in October 2026 is about four thousand seven hundred. Sold back to the trade, at the gap the studies find, it is worth somewhere between two and three thousand. The same six and a half thousand dollars in gold on 4 November 2011 bought three and seven-tenths ounces. Those ounces were worth about fourteen and a half thousand dollars on 30 October 2026. In rand, the currency the South African buyer earned, the metal multiplied four and a half times while the stone lost two-thirds. .The savers second, because the claim was sold as an investment and not only as a sentiment. Between 2012 and 2017 four of Italy's largest banks sold diamonds to their own customers as an investment. They sold through two firms, Intermarket Diamond Business and Diamond Private Investment. The prices, the competition authority later found, were far above the market's. The diamonds were presented as a safe store of value. The Bank of Italy's survey put the total at €1.9 billion sold to 71,108 customers. The banks took commissions of about fourteen and a half per cent. The authority fined the banks and the two firms fifteen million euros in October 2017. IDB was declared bankrupt in January 2019. Milan prosecutors seized more than seven hundred million euros from the banks and the firms the following month. By September 2021 the banks had refunded €1.16 billion to three-quarters of the customers. The diamonds had been sold as what De Beers' advertising said they were, by people the customers trusted. They were worth about half. The lesson of every such scheme is the one Epstein drew in 1982. The point of "forever" was always that the stone should never come back to the market. The day it does, the holder learns the trade price. The dealers and their banks third, and this is the hidden point of the whole report. When De Beers stopped holding the buffer in 2000, the buffer did not disappear. It moved down the chain. The sightholders, the Antwerp houses and the cutters of Surat carried, on bank credit, the inventory that De Beers no longer held. They carried it without knowing that they had become the custodian, at their own risk, with none of the custodian's margin. In 2014 and 2015 they were made to take rough priced above polished, and refused. In 2019 Eurostar fell with half a billion dollars of debt and an inventory worth a sixth of its mark. In 2024 and 2025 the midstream destocked again and De Beers' own unsold stock rose to two billion dollars. The banks that had financed the chain, Antwerp Diamond Bank, Standard Chartered and ABN AMRO, were already gone or going. The custodian's buffer became the dealer's overdraft. The dealer's overdraft became the bank's loss. The exception proves the rule. The Argyle mine in Western Australia was the source of nearly all the world's pink diamonds. It closed on 3 November 2020. Rio Tinto said in December 2020 that the pinks sold at its tenders had "appreciated over 500 per cent over the past two decades". Its final tender in 2021 set records across the collection. In May 2024 a 1.56-carat Argyle red sold at Phillips in Geneva for US$4.2 million, about US$2.7 million a carat. A stone whose supply has ended, with no reactor able to replace its colour, does what the advertising claims the whole category does. Real, final scarcity is what a stone's value needs. For the rest of the category the scarcity was organised, and it was sold. Put the two claims together and the position De Beers defends in 2026 is visible. It cannot say the laboratory stone is not a diamond, so it says the stone is not special. It cannot say the mined stone holds its value, so it says the value is emotional. Both are true. Neither sells a mined one-carat stone at four thousand seven hundred dollars against a one-carat stone from a reactor at seven hundred. What sold the mined diamond for eighty years was a controlled supply, a buffer and a story, told by a company with the market share to tell it. The company gave up the first in 2000, the second by 2004 and the third in 2010. In 2018 it discovered that the one thing a factory could not copy was the thing it had already dismantled. For the South African saver: the metal was better than the stone, in dollars by a factor of two and in rand by a factor of four. It was better because nobody had to promise anything about it. The metal was better than the stone. A ring bought in 2011 is worth a third of its price to anyone who has to sell it. The same money in gold has more than doubled in dollars and multiplied four and a half times in rand, and nobody had to advertise it. THE CHAIR, OCCUPIED XX. What a custodian would have done, 2012 to 2026 On the morning of 17 August 2012, the day after completion, somebody at Anglo American owned the chair that Ernest Oppenheimer had sat in from 1929. This section is written as if I had been that person. The rule is narrow so that it can be attacked. Only the information available on each date is used. The balance sheet is the one Anglo actually had. The cost of each alternative is stated beside it, from the accounts. Hindsight is excluded by design. What each alternative would have earned was not knowable either: the costs are from the accounts, the protections are the author's estimates. The custodian does not know in 2012 that the laboratory stone will take the engagement ring, any more than Anglo did. He knows only the rule the founder wrote down: a diamond is worth what its scarcity is worth. Scarcity is something a company pays to maintain in the bad years so that it can collect in the good ones. .The scorecard, and the test the thesis has to pass. Add up the custodian's bill. A buffer of about US$1.5 billion. Advertising of about two billion over fourteen years. A billion of stock bought in 2015. A billion of 2022 earnings retained. About a billion of losses carried while holding instead of selling in 2024–26. Less about US$1.7 billion of Venetia and Lightbox capital not spent. The custodian needed about US$4.75 billion more than Anglo put in, over fourteen years. Of that, the stock, about US$3.5 billion, is an asset that is sold again and not a loss. Then add up what it might have protected, on my own arithmetic and without flattery; none of it was knowable in advance. The 2015 collapse softened by perhaps four hundred million of earnings. The stock bought at the bottoms and sold into the recoveries, perhaps six hundred million. The difference between selling at a billion and holding at the carrying value, nine hundred million at least. And the story, whose value I cannot price and will not pretend to. That is about two billion dollars of value protected, against US$4.75 billion of capital committed, of which three and a half comes back as stock. The test is not clearly passed, and I say so. The custodian would not have saved the price of 2022. The factory set a ceiling under it that no stockpile can lift, and Section XIV has priced the attempt. What the custodian would have done is narrower and, I think, sufficient. He would have kept the company solvent through its own cycle. He would not have sold the clients' margin in 2014. He would not have built a shaft into a falling market. He would not have auctioned the whole at the bottom to the people who had run it for the family. The thesis this report began with survives on that narrower claim. A diamond's value was in a company's willingness to lose money to protect it. The custodian would have lost about two billion dollars net to protect it, and would still own the company. Anglo lost about ten billion of cash and value on the whole eighty-five per cent to not protect it, by the account in Section X, and is selling. The platinum producers who read this have the same choice at the same point in their cycle. On the platinum belt the three producers between them are the custodian whether they like it or not. The arithmetic of custody there is the paper this one points to. The custodian would have needed about US$4.75 billion more over fourteen years, most of it stock that comes back, and would have protected perhaps two billion and the company. That does not clearly pass the test. The narrower test it passes: he would still own De Beers. WHAT SOUTH AFRICA STILL HOLDS XXI. Platinum, chrome, manganese: the lesson applied In 1970 the mines of the Witwatersrand hoisted a thousand tonnes of gold, seven-tenths of the Western world's output. The men who ran them had, almost all of them, started in the diamond fields or in houses the diamond fields had built. In 2024 the country produced about a hundred tonnes. No substitute did that. The gold was lost by selling the control. The houses that owned it sold it in stages to companies whose managers sold the mines, and the country kept the royalty. Geology, depth and cost had their part. Nobody should pretend the Witwatersrand at four kilometres is the Witwatersrand at a thousand metres. But the companies that stayed and bought what the sellers left are worth, today, several times what the sellers received. That says what the sellers' judgement was worth. De Beers is the diamond version of the same story. This report ends on what South Africa still dominates, and whether it is run by owners or by managers. .What South Africa still supplies, as a share of world mine output: platinum, rhodium, chromite, manganese, palladium, and gold in 1970 and 2024. Sources: USGS Mineral Commodity Summaries 2025; the International Platinum Group Metals Association; LBMA for the 1970 gold share. .The platinum group metals are the largest case. South Africa supplies about seven-tenths of the world's platinum, four-fifths of its rhodium and two-fifths of its palladium. The supply comes from three companies, Valterra, Impala and Sibanye, with Northam behind them. That position is closer to De Beers' in 1990 than anything else in the world's mineral trade. It faces a substitute, the battery car, and a recycler who returns a growing share of the metal. Against that, the hybrid for now uses more. The rhodium price fell by more than three-quarters between 2021 and 2024. The three companies closed shafts, cut jobs and wrote down assets, which is what De Beers did in 2015 and 2023. None of the three holds stock against the cycle. On the story, the record is better than it was for diamonds after 2010 and worse than it looks. The producers founded the World Platinum Investment Council in November 2014, six of them together, to build investment demand. They have funded the Platinum Guild International's jewellery promotion since 1975. Neither body publishes its budget, and Sibanye walked out of the council in 2019. The producers do spend on the story. What matters is whether the spend is of the scale De Beers' was, two hundred million dollars a year, and it is not. It also matters whether the spend is run for the long term, and a council that loses its second-largest member in a bad year is not. A former De Beers executive, Varda Shine, said in November 2026 that "there is a whole generation of consumers that hasn't been educated about diamonds", in years in which the advertising of handbags and telephones had become "extremely prevalent". Then the sentence that applies to the platinum belt without a word changed: "Once you start treating diamonds as a commodity, you end up where we are today." The three producers treat platinum as a commodity because that is what a hired manager on a three-year plan is paid to do: hoist it, refine it, sell it at the price of the day and report the margin. Nobody has educated a generation of car buyers or jewellery buyers about the metal, on a budget anyone will publish. De Beers had the budget, two hundred and twenty million dollars in 2007, and let it go. The platinum producers have never had it. The lesson of the stone is not that the budget is expensive. It is that the budget is the product. None of the three has proposed the thing Rhodes would have proposed in 1888. That is a channel through which they sell together and hold together, and through which they tell the car makers and the jewellers together what the metal is for. The nearest anyone came was a chief executive in 2017 who said that the way to fix the price was to "take platinum out of the market" by closing shafts. That is the custodian's method without the custodian's stockpile, and the Tin Council's book without the Tin Council's money. Whether three competitors may lawfully sell through one channel is a question for the competition lawyers in every market they sell into. This report does not answer it. What it says is that holding a buffer and telling the story are each lawful on their own, and each has been left to the cycle. And the three are run by hired managers on three-year plans. This country has just watched that record at Anglo American Platinum. It was demerged in 2025 by a chief executive who then sold the rest of Anglo's stake eleven weeks before the price went vertical. Chrome is the second case, and it has already gone the way of the gold. South Africa holds nearly half the world's chromite and once smelted most of it into ferrochrome. China built ferrochrome furnaces on South African ore and cheaper power. South Africa now exports the ore and imports the margin. The smelters at home idle when the power price rises. Manganese is the third: two-fifths of the world's ore, exported as ore. In each case the country has the deposit and has given up the channel, the buffer and the story. The value has moved to whoever holds those. The lesson from De Beers is the four requirements in Section V, not that a dominant producer should run a cartel. The fourth requirement is the one that matters. Supply can be organised, within the law. A buffer can be funded. The state's own holdings and the producers' balance sheets at the top of the cycle could hold a year of rhodium. A story can be told, and is being told, on a budget nobody will publish. What cannot be bought is time. Time is what an owner has and a manager does not. The Oppenheimers held the diamond price for eighty years because they intended to be there in eighty years. The men who ran De Beers after 2011 intended to be there in three. The men who run South Africa's platinum, chrome and manganese intend the same. The record of this report is what that intention produces in a dominant producer when the substitute arrives. It arrived for diamonds in 2018. For platinum it is in the showroom. For the policymaker: the state holds stakes, royalties and the licence in all three. It is the only party in the room with an owner's clock, and it has so far used it to collect, not to hold. South Africa lost its gold by selling the control, and is losing De Beers the same way. It still holds seven-tenths of the world's platinum, half its chrome and two-fifths of its manganese, run by managers on a three-year clock, with the substitute in the showroom. THE BEAR, MET XXII. The case for the managers, and the answer In March 2020 the sightholders could not travel to Gaborone. De Beers let them defer what they had contracted to buy instead of forcing the goods. The price was not cut until August. The case for the managers deserves its full strength before it is answered, and that is where it begins. They would say, first, that the model was dead before they touched it. By 2000 De Beers sold two-thirds of the world's rough and falling. The Americans would not let its executives in and the Europeans would not clear a stockpile. Conceded in part. The single channel could not have been held at forty per cent of the market. But the stockpile and the advertising were not the channel. A company with forty per cent of a market and the only brand in it can hold a buffer and tell the story. Hermès does not control the supply of leather. The family sold the one and shut the other because each cost money in the short run. They would say, second, that nobody saw the laboratory stone. Section XII answers that in the company's own words. They would say, third, that the price of rough is set by China and the Indian cutters, not by De Beers. They would say that the collapse of 2023–25 had causes no custodian controls. Chinese demand for diamond jewellery fell by roughly two-fifths to a half between 2022 and 2024, on the trade's estimates. The midstream destocked. Interest rates rose, and the banks that had financed the chain had left. Russian goods, sanctioned in 2022 and banned by the G7 from 2024, found their way to market through Dubai and India. In 2025 American tariffs hit Surat. Partly true, and it is the strongest point. Those are exactly the cycles the Oppenheimer method was built to absorb, with a buffer and a buyer of last resort. The method had been dismantled. The downturn of 2023–25 was no worse in its causes than 2009. It was far worse in its effect. De Beers met it with no stockpile policy and no marketing budget of the old scale. Its client list had learned in 2015 that it could refuse the goods. But the bear is owed this much. In 2020 the managers did act as custodians. In 2022 they earned the best year since 2014. No custodian could have made China buy. They would say, fourth, that the purchase was fairly priced by the brokers' standards and that no board could foresee a ninety per cent fall. Conceded as to the brokers. But the sellers were the Oppenheimers. They took five billion dollars out at the top of the market a year after putting six hundred million in to keep the company alive. A buyer who does not ask why the people who invented the market are leaving it has not done the one piece of diligence that matters. The brokers priced the earnings. The family priced the cycle. They would say, fifth, that the good mines were kept and the bad sold, which Section XI confirms. They would say that Jwaneng, Venetia and the Namibian ships will be worth a great deal when the market turns. Conceded, and it is the strongest point for the buyer. If the laboratory stone goes to cost, as Cook says, and the mined stone recovers half its lost ground, the consortium will have made the purchase of the decade at Anglo's shareholders' expense. That is possible. It is also the argument every seller at the bottom makes about the buyer. It is why the Oppenheimers were never sellers at the bottom. They would say, sixth, and the Financial Times said it for them on 2 November 2026, that the decline is Anglo's: that De Beers was a business of "former glory" until the professionals bought it, and that the family era is the standard the professionals failed. The trade's retrospectives all start the clock in 2012. This report starts it in 2000, and the article's own dates are the reason. By its account the company abandoned supply management in 2000 and "shifted to advertising its own products and doing less general marketing on behalf of the industry". Its marketing budget was two hundred and twenty million dollars in 2007 and falling. Its share was forty per cent by 2010, on the series the article prints. Every one of those is a family decision and every one predates the cheque. Anglo did not break the custodian. It bought a company that had stopped being one, paid for the glory, and never noticed it had already gone. The bear is conceded the second act in full. He is not conceded the first, because the first is on the record in the family's own words, and because without it the second could not have happened. The bear takes the first point in part, the third in large part, the fourth as to the brokers, and the fifth as to the mines. He is given 2020 and 2022 in full. He does not take the stockpile, the advertising, Lightbox, the carrying value, or the claim that nobody saw the stone. A producer that meets a downturn with no buffer will always be able to blame the downturn, and the blame will always be partly right. That is why the buffer has to be built before the argument starts. THE VERDICT XXIII. Who destroyed De Beers The family dismantled the method and sold the company at the top. The professionals bought the chair without knowing what it was for. They ran the company as a mine. They built things that did not work with the money the good years produced. They carried it at or near the purchase figure until the market had long stopped believing it. They met the substitute with a brand of their own. They are reported to be selling at the bottom to the man who ran it for the family. The laboratory stone did what the factory substitutes in Section XIII did: it became cheap and the market bought it. It walked through a door the owners had opened and the buyers had never noticed was there. Was it one of the worst purchases ever made? The table puts it beside the others, in figures. By the share of the price lost or written off, De Beers stands with Rio Tinto's Alcan, Kinross's Red Back and BHP's shale in the first rank of mining's mistakes. It stands above all of them in one respect: the others bought from a market, and Anglo bought from the people who made it. By dollars it is Anglo's second worst, after Minas-Rio. Was it clever, with inadequate management? The asset was good and still is; Jwaneng is Jwaneng. The purchase would have been defensible had the buyer intended to run the method. Its record shows no sign that it knew there was a method. It bought a mining company from a family that had been running something else, and never noticed the difference. That, and not the laboratory, is the verdict. One condition is attached. The sale is not signed. If the buyer runs the harvester's plan and the market merely stops falling, the loss will be as stated here and the gain will be his. .And the two claims. A laboratory stone is a diamond, by the chemistry and by the American regulator since 2018, and it is a diamond that is not rare. A mined diamond of ordinary commercial quality bought at retail in the last fifteen years is worth less today on the indices the industry publishes, and far less to anyone who has to sell it. The advertising regulator has made the industry's own council stop saying otherwise. Exceptional stones are the exception, and Argyle's pinks show why. The first claim is a distinction held together by vocabulary. The second is an advertisement the regulator has withdrawn. Neither is the thing De Beers sold for eighty years. What it sold was scarcity, and the scarcity was sold in 2000. De Beers failed because the people who owned it sold the scarcity, the people who bought it never knew it was there, and the stone from the factory walked in through the open door. The laboratory that learned to make a diamond did not do it. WHAT RHODES AND ERNEST WOULD SAY XXIV. A note in my own words This section is opinion and I give it as such. Rhodes and the Oppenheimers cannot be asked. I have spent a fortnight with what they wrote and did. I set down what I believe they would say, on my own responsibility. It is shorter than I first wrote it, because the section before last has done the work. Rhodes would say that he amalgamated Kimberley in 1888 because too many sellers of the same stone will always destroy its price. He would say the same of three South African platinum companies. He would not have launched Lightbox. In 2016, when the choice was still open, he would have done one thing fully or nothing, and I think nothing. By 2022 the choice was gone. Ernest would say that the people who sold the stockpile in 2000 did not understand what they were selling. The stockpile was the scarcity, held in a vault, and not an inventory. The day De Beers sold it to working levels was the day a diamond became a commodity with a long delivery time. He would say that the advertising was never a cost. He had spent through the Depression and the war, when nobody was buying, because the story had to be told loudest when the price was weakest. He wrote in 1954 that the aim of his group was "to earn profits, but to earn them in such a way as to make a real and permanent contribution to the well-being of the people and to the development of Southern Africa". His grandson read the sentence out at Vergelegen on the night the family left Anglo. And he would say that his grandson was right to sell in 2011, because the family's job was to read the cycle. He would say that the buyer was careless. A buyer who pays the people who set the price for the right to set it, and then does not set it, has paid for a chair and left it empty. Harry would look at the platinum belt, the chrome and the manganese, run on the same clock as the company that bought his. He would say that the country has lost its diamonds as it lost its gold: to the quarter, not to a laboratory. I will end where I began, with a copywriter's line and a drawer. A diamond is the only thing most families ever buy that is meant to outlive them. It is given on one knee, worn for fifty years and left in a will. Then the will is read and the ring is taken to a jeweller. The family learns what the trade has always known: the stone was worth what someone would pay to give it, and a third of that to anyone who has to sell it. In the same drawer, as often as not, there is a Krugerrand somebody's father bought and forgot. It has no slogan and never had a custodian, and it is worth four and a half times what he paid for it. Both were sold to us as forever. Only one of them was. THE FAMILY SOLD THE SCARCITY IN 2000 AND THE COMPANY IN 2011. THE PROFESSIONALS PAID US$5.2 BILLION AND GOT WELL UNDER HALF OF IT BACK ON ANY ASSUMPTION. THEY CARRIED THE COMPANY AT OR NEAR THE PURCHASE FIGURE UNTIL 2024. THEY ARE REPORTED TO BE SELLING IT FOR A BILLION TO THE MAN WHO RAN IT FOR THE FAMILY. THE SAME CHEQUE IN GOLD WOULD BE WORTH US$12.6 BILLION THIS MORNING. THE CUSTODY THE BUYER WOULD NEED COSTS ABOUT TWO AND A HALF BILLION DOLLARS; HE HAS SHOWN HALF A BILLION. SOUTH AFRICA STILL HOLDS SEVEN-TENTHS OF THE WORLD'S PLATINUM, HALF ITS CHROME AND TWO-FIFTHS OF ITS MANGANESE, RUN BY MANAGERS ON A THREE-YEAR CLOCK, WITH THE SUBSTITUTE IN THE SHOWROOM. THE METAL WAS BETTER THAN THE STONE. .REFERENCES ANGLO AMERICAN AND DE BEERS: RESULTS, REPORTS AND ANNOUNCEMENTS Anglo American plc. Anglo American to acquire the Oppenheimer family's 40 per cent interest in De Beers. Press release, London, 4 November 2011 (US$5.1bn; the shareholder loans; interest, top-up and break fee; Botswana's pre-emption right; the quotations of Cynthia Carroll and Sir John Parker). Anglo American plc. De Beers transaction: Government of the Republic of Botswana decides not to exercise its pre-emption right. 31 July 2012. Anglo American plc. Completion of the acquisition of the Oppenheimer family's interest in De Beers. 16 August 2012 (US$5.2bn; 85 per cent). Anglo American plc. Annual Report 2012 (net segment assets, Diamonds, US$12,944m; Venetia underground; Jwaneng Cut-8); Annual Report 2013; Annual Report 2016. Anglo American plc. Preliminary and annual results for 2013, 2014, 2015 (Snap Lake impairment US$0.6bn), 2016, 2017, 2018 (Voorspoed; Peregrine/Chidliak US$87m), 2019, 2020, 2021, 2022, 2023 (22 February 2024: impairment US$1.6bn "principally relating to goodwill"; rough inventory write-down US$0.2bn), 2024 (20 February 2025: impairment US$2.9bn; carrying value US$4.1bn) and 2025 (20 February 2026: impairment US$2.3bn pre-tax; carrying value US$2.3bn, attributable US$1.9bn; underlying EBITDA US$(511)m; trading EBITDA US$(424)m; inventory reduced by US$0.9bn; Venetia steady state "mid-2030s"); the 2025 results script; half-year results 2015, 2020, 2025 and 2026 (30 July 2026: revenue US$1,583m; EBITDA US$(113)m; realised price US$105/ct; Venetia pause; sightholder list), and the third-quarter production report of 20 October 2026 (realised price US$84/ct; "continue to work towards a transaction in a challenging market"). Anglo American plc. Portfolio simplification: De Beers to be divested or demerged. 14 May 2024. Anglo American plc. Preliminary results presentation, 20 February 2026, and interim results call, 30 July 2026 (Duncan Wanblad on the sale process: "all of these parties are strategic"; "we are not exclusive with anybody"). Anglo American plc. De Beers rights issue of US$1bn, 11 February 2010. Anglo American plc. Results for the half-year to 30 June 2011, 26 July 2011 (De Beers' record first half). De Beers. Results for 2011, 10 February 2012 (sales US$7,378m; EBITDA US$1,721m; production 31.3 Mct). De Beers Group. Appointment of Philippe Mellier, May 2011; Philippe Mellier to step down and Bruce Cleaver appointed, 27 May 2016; Al Cook appointed chief executive, 10 October 2022; Bruce Cleaver to leave the board, January 2024. De Beers Group. Lightbox Jewelry, 29 May 2018; Lightbox price changes, Q4 2023 and 12 May 2024; a new strategy ("Origins"), 31 May 2024; closure of Lightbox, 8 May 2025. De Beers Group. Agreement with the Government of the Republic of Botswana: heads of terms, 30 September 2023; signing, 25 February 2025 (the sales agreement; licences to 2054; the development fund). De Beers Group. Gahcho Kué: enforcement of security over Mountain Province Diamonds' interest, 16 October 2026. De Beers Group. Venetia: production pause and capital rephasing, June 2026. De Beers and Anglo American. Annual reviews and releases on diamond stocks, 2000 to 2004 (US$3.065bn end-2000; US$2.655bn end-2001; US$2.355bn end-2002; US$1.745bn end-2003; "working levels" mid-2004). European Commission. Decision clearing Supplier of Choice, 16 January 2003, and Anglo American's release of the same day (Gary Ralfe: "truly transformational"). United States Department of Justice. De Beers pleads guilty to price fixing; US$10m fine. 13 July 2004. Commission Decision 2006/520/EC (the Alrosa supply contract), 22 February 2006. PETRA, MOUNTAIN PROVINCE AND OTHER COMPANIES Petra Diamonds. Acquisition of Cullinan (R1bn, 2008) and Finsch (R1,425m, 2011); results 2014 to 2025; restructuring of March 2021; sale of Koffiefontein (2024) and Williamson (2025). Petra Diamonds. Sale of the Blue Moon rough (29.6 ct, US$25.6m to Cora International, February 2014); Sotheby's Geneva, the Blue Moon of Josephine (12.03 ct polished, about US$48.5m, 11 November 2015); other Cullinan blue diamonds, 2010–2021; De Beers' purchases, in partnership with Diacore, of Cullinan blue parcels, 2020 and 2021 (US$80.6m in all). Mountain Province Diamonds. Announcements of October 2026. Winspear Resources. Acquisition by De Beers, 2000 (C$306m). Ekapa Minerals. Acquisition of Kimberley Mines, completed 21 January 2016 (R102m). Trans Hex. Acquisition of the Namaqualand mines, 2011 (R225m). Pandora. Lab-grown diamonds, 5 May 2021, and results for 2024. Signet Jewelers. Results and conference calls 2024–26 (lab-created diamonds as a share of revenue). Richemont and Signet: Bloomberg, 15 October 2026, on talks to join the Global Diamond Consortium. GOVERNMENTS, REGULATORS AND OFFICIAL STATISTICS Parliament of Botswana. Statement by the Minister for State President, Moeti Mohwasa, 17 July 2026 (the competitive process; three shortlisted bidders; the Global Diamond Consortium as preferred bidder). Government of Botswana. Statements of President Duma Boko, 14 November 2024, 25 February 2025 and 15 December 2025 ("the IMF are not Batswana, they are not voters"); Minister Bogolo Kenewendo, July and August 2025 ("full control over this strategic national asset"). International Monetary Fund. Botswana: 2025 Article IV consultation, November 2025 (the caution on increasing the stake in De Beers). Bank of Botswana and Ministry of Finance. Budget statements 2025 and 2026; mineral revenue as a share of GDP; the revised deficit of 22 September 2026. S&P Global Ratings (12 September 2025; 13 March 2026) and Moody's (October 2025; 25 September 2026) on Botswana. Debswana. Sales for 2022, 2023 and 2024. United States Federal Trade Commission. Revised Guides for the Jewelry, Precious Metals and Pewter Industries, July 2018; warning letters to eight marketers of simulated and laboratory-created diamonds, 2 April 2019. BBB National Programs, National Advertising Division. Decision on the Natural Diamond Council's advertising claims (value retention, scarcity and emissions), and the council's agreement to comply. Gemological Institute of America. New descriptive terminology for laboratory-grown diamonds, 2 June 2025. Kimberley Process. Annual statistics, 2024. RESEARCH, SURVEYS AND BOOKS Epstein, E. J. 'Have You Ever Tried to Sell a Diamond?' The Atlantic, February 1982, and The Rise and Fall of Diamonds (1982): the 1910 Oppenheimer memorandum on scarcity. Shor, R. 'A review of the political and economic forces shaping today's diamond industry.' Gems & Gemology, Fall 2005 (market shares; the stockpile; Argyle; Russia; Canada; Supplier of Choice). Zook, C. Unstoppable (Bain & Company) on the 1999 De Beers review. Bain & Company and the Antwerp World Diamond Centre. The Global Diamond Industry, 2019 ("5% to 15% of natural diamonds in value terms through 2030"), 2020–21 and later editions. The Knot. Jewelry and Engagement Study, 2019 to 2025 (laboratory-grown centre stones 12, 18, 36, 46, 52 and 61 per cent; average spend). Golan, E. Lab-grown wholesale price index, 2018 to 2026, as published in JCK (−86.5 per cent to March 2024; −96 per cent to the first quarter of 2026). Zimnisky, P. Global Rough Diamond Price Index, 2007 to 2025. Rapaport. RapNet Diamond Index, one-carat, 2023 to 2025. Gem Council of India (GJEPC). Export statistics for polished laboratory-grown and natural diamonds, financial years 2019 to 2026 and calendar 2022 to 2026. CareEdge Ratings. Lab Grown Diamonds, May 2024. MIT Sloan School of Management. De Beers' Diamond Dilemma (case study). Ad Age. Top advertising slogans of the century, 29 March 1999. PRESS: THE MODEL AND THE 2011 SALE Rapaport. Interview with Nicky Oppenheimer, London, 14 July 2000 (published 27 July 2000): "We are moving from being the seller of last resort to being the seller of choice." Rapaport. Oppenheimer bids for De Beers, 15 February 2001, and the Ralfe interview of the same month (the stockpile at US$3.065bn). CNN Money. De Beers goes private, 15 February 2001 (US$17.6bn; the diamond business valued at US$8.3bn). JCK. De Beers after Bain, 1 October 1999; Supplier of Choice begins, 1 July 2001 ("we do not ever see us returning to the full custodian role"); Gareth Penny, Person of the Year, 1 February 2004 ("We absolutely destroyed shareholder value"; "We are not a cartel"); De Beers and Forevermark, 1 November 2008 (the end of generic advertising; "past campaigns cost an estimated $200 million per year"); Supplier of Choice, RIP, 11 April 2014. IDEX Online. De Beers to close its US generic marketing offices, 23 December 2010 ("the crisis changed the timing, not the strategy"); Gary Ralfe's sightholder address, 18 January 2005; Ralfe's retirement, 1 March 2006. Christian Science Monitor. De Beers and Argyle, 28 December 1982. The Northern Miner. Argyle and the CSO go separate ways, 1996. Mail & Guardian. De Beers sees another year of growth, 5 February 2004 (stocks US$1.745bn); Oppenheimers to keep De Beers stake, 17 January 2011 ("it seems to me an unlikely event"); De Beers no longer jewel in crown, 11 November 2011 (the Anglo board's doubts; "instructed to end the arrangement"); De Beers 2011 total sales up 26%, 10 February 2012. Miningmx (B. Ryan). The Oppenheimers eye new projects in Africa, 4 November 2011 ("I was extremely surprised when John Parker came to see me"; "reasons that I am not prepared to go into"); Why the Oppenheimers sold out of De Beers, 7 November 2011 (the loans; the rights issue; Mary Slack; Jonathan; Venetia; "Do I have an ambition that Jonathan should succeed me as chairman of De Beers? Absolutely!"); Botswana may pass on De Beers pre-empt, 11 July 2012; De Beers' Penny quits firm, 23 July 2010; Oppenheimer to retire from Anglo's board, 25 February 2011; the appointment of Philippe Mellier, 6 June 2011 ("no mining experience, no experience of diamonds and no experience of southern Africa"). Business Report (IOL). Diamonds are not forever, 4 November 2011 (James Teeger: "We didn't approach Anglo, Anglo approached us"); 24 November 2011. Reuters. Anglo American to buy Oppenheimers out of De Beers, 4 and 7 November 2011 (Liberum; Cadiz; the share price); Insight: Oppenheimers consign diamonds to the past, 22 November 2011 (the earlier approaches; Parker's September visit). Fin24. Diamond dynasty, 27 November 2011. Mmegi. Of De Beers, politics and strategy, 3 August 2010. JCK. Gareth Penny interview, 29 September 2010; Oppenheimer family gives up control of De Beers, 4 November 2011. Daily Investor. The Oppenheimer family saved R111 billion through one decision, 4 August 2026. PRESS: THE RECORD, THE ASSETS AND THE MANAGERS JCK. Philippe Mellier interview, 21 January 2015 ("I have a business. I am not a banker. I am not a supporter of customers"; "you need to own your margin"); the July 2015 sight, 16 July 2015; Mellier's departure, 27 May 2016 ("now is the right time for me to pass the baton"); Bruce Cleaver on Lightbox, 7 February 2023 ("one of the scariest things I've ever done"); De Beers' new strategy, 31 May 2024; How Lightbox became a bust, 13 May 2025; Paul Zimnisky on the diamond market, 28 June 2024; Edahn Golan on lab-grown prices, 27 March 2024; lab-grown wholesale in freefall, 2026; the engagement market, 2026; Al Cook at JCK Las Vegas, 15 June 2026. Rapaport. How De Beers sees its new synthetics business, 30 May 2018 (Cleaver: "no uniqueness and no rarity factor"; "you can lose in the sea"); the August 2015 price cut; Rapaport's call for Mellier's resignation, November 2015; De Beers reduces prices of Lightbox, 12 May 2024; De Beers to close Lightbox, 8 May 2025; De Beers introduces a 2.0 version of "A Diamond Is Forever", November 2024; Anglo slashes De Beers value, 20 February 2025 and 22 February 2026; Anglo picks Gareth Penny-led consortium as top De Beers bidder, reports, 17 July 2026; Lab-grown taking over engagement-ring spending, The Knot, 2026; Martin Rapaport on De Beers' synthetics, 4 July 2018. National Jeweler. Lightbox significantly lowered its prices, 26 February 2024; Lightbox to go dark by summer, 12 May 2025 (Cook: "Kind of"); Q&A with Al Cook on the lab-grown bubble and dwindling supply, 9 June 2026 ("$30 and $100 a carat"; "1,000 percent margins"; "the bubble will burst"); De Beers sale in final, most challenging phase, 30 July 2026. CNBC and TechCrunch. De Beers launches Lightbox, 29–30 May 2018 ("may not be forever, but is perfect for right now"; the US$94m plant). Mining Weekly. De Beers lifts 2014 earnings, 13 February 2015; De Beers' underlying EBITDA improves, 21 February 2017; Lightbox launch, 29 May 2018. Miningmx. De Beers: don't call synthetics a diamond (the December 2013 client note); Anglo boss admits diamond industry misread lab-grown threat, 4 August 2026 (Wanblad: "Hindsight would probably show that we might have wanted to have been a little bit more aggressive on the signals we were getting"; "Looking in the rear-view mirror is a much more perfect science"); Can De Beers make diamonds forever again?, 31 August 2026 (Avi Krawitz: "In hindsight it was probably their biggest mistake"); Anglo favours De Beers suitors with diamond experience, 15 August 2025; Zijin's bid and the De Beers field, 30 January 2026. IDEX Online. Wanblad on lab-grown, 5 August 2026. Financial Times. Interview with Duncan Wanblad, 4 August 2026, as reported by Miningmx and IDEX; FT Mining Summit, October 2024 ("There is no way De Beers is a fit in the portfolio"). Gulf News. De Beers creates man-made diamonds it will never sell (Simon Lawson), 2015–16. JCK. Forevermark CEO sounds off on synthetics, labs and A Diamond Is Forever (Stephen Lussier), 3 June 2016 ("I don't think so today"; "it is not inherently rare, and as a result, it won't have enduring value"). Fortune. Al Cook on Lightbox, 4 June 2024 ("We know how to do it and we're coming back"). Rough & Polished. James Allan on De Beers, April 2025. PRESS: THE SALE AND BOTSWANA Bloomberg. Anglo has chosen a preferred bidder for De Beers, Botswana says, 17 July 2026; the terms of the Global Diamond Consortium's offer (about US$1bn; US$750m upfront; US$250m deferred; earn-outs; US$500m injection), 29 July 2026; Botswana likely to take about 25 per cent, African states below half, 13 October 2026; Richemont and Signet in talks to join, 15 October 2026; Botswana's president on De Beers, 15 December 2025 ("We want to take a bigger share and become the real owners"). Semafor. Anglo names preferred bidder for De Beers stake, 20 July 2026. Weekend Post (Gaborone), 21 September 2026. bne IntelliNews. Botswana nears key decision point on Anglo's proposed US$1bn sale, October 2026. Reuters. Al Cook at Reuters NEXT, 16 June 2026 ("We've never been closer than we are to a sale"); Botswana's budget and ratings, 2025–26; Bruce Cleaver's consortium withdraws, 13 July 2026 ("difficult to see an appropriate return on investment over the short term"). Miningmx. Botswana and De Beers sign, 3 February and 25 February 2025. VOA. Botswana and De Beers sign agreement, 26 February 2025. Botswana Daily News. Boko's remarks of 14 November 2024. JCK. Boko on De Beers, 2025 ("De Beers is not doing its job"); the Global Diamond Consortium, 30 July and 14 October 2026; Gareth Penny on the duties of a chairman (Enter the Boardroom podcast), June 2026 ("The chair should be the last to speak"). Engineering News. Botswana's mineral revenue and the 2026 budget revision, 22 September 2026. Financial Times. De Beers' inventory of about US$2bn, February 2025. Financial Times, 2 November 2026, on De Beers and Venetia (Venetia: production suspended for two years, union consultations concluded mid-September 2026, 3,500 direct employees informed, "South Africa's last major diamond mine"; De Beers' marketing spend about US$220m in 2007 and US$166m in 2018, undisclosed this decade, the "largest category marketing investment in more than 10 years" announced in 2025 without a figure; Paul Zimnisky's series of De Beers' share of world supply by value, 1990–2026, and his rough price index on a 31 December 2007 base, both read from the charts; Bain's advice c. 2000–02 against legitimising laboratory-grown stones, on two people with knowledge of the work; Lightbox "cataclysmic", "catastrophic", De Beers "surprised" by retailers' appetite; "Nobody on the Anglo board really understood it"; "Anglo has destroyed a very good business"; "caught asleep at the wheel"; "didn't automatically have a right to exist"; "you can only market something if you know what you're selling"; Sir John Parker, diamonds "an exception in the gloom"; Ndaba Gaolathe interview, September 2026, on the stake, public money, a "modern strategy" and a market "nervous" about branding spend; the IMF warning of December 2025; Gareth Penny's consortium valuing Anglo's stake at about US$1bn and Richemont "part of the consortium", on a person familiar with the matter; Zimnisky on mine closures and "a swift rebound in prices"; Varda Shine on a generation "that hasn't been educated about diamonds" and "Once you start treating diamonds as a commodity, you end up where we are today"; De Beers Group statement on "a shared approach to natural diamond marketing through collective investment"; Anglo American declined to comment). Photographs by Waldo Swiegers. THE DIAMOND FIELDS, RHODES AND THE PRECEDENTS South African History Online. De Beers Consolidated Mines formed, 13 March 1888; Sir Ernest Oppenheimer. Kimberley.org.za and the Barnato histories (the cheque for £5,338,650, 18 July 1889). Encyclopedia.com. The London Diamond Syndicate, February 1890; De Beers' share of world output. Rotberg, R. The Founder: Cecil Rhodes and the Pursuit of Power (1988). Wheatcroft, G. The Randlords (1985): Rhodes, Rudd, Barnato, Beit, Wernher, Robinson, the Joels, Phillips, Eckstein at Kimberley; Beit's £25,000 loan to Robinson, 1886. Phimister, I. Rhodes, Rhodesia and the Rand (Gold Fields of South Africa, 9 February 1887). Wikipedia and the company histories: Barnato's JCI (1889) from his De Beers holding; Anglo American (1917); CDM (1920); Oppenheimer's De Beers chairmanship (December 1929). Johnson Matthey. The history of Rustenburg Platinum Mines (the 1932 amalgamation under JCI). Impala Platinum. Company history (Union Corporation, 1966; first production 1969). Anglo American Platinum. Formation from JCI's platinum interests, 1995. Bahrain and the Gulf pearl trade: Bahrain government trade returns 1929–31 (exports £141,000 to £43,000); the pearling fleet (over 3,000 boats to about 300 by 1946), as cited in the Gulf economic histories; Mikimoto's 1916 patent and the Paris litigation of 1924. Chilean nitrate: BASF, Oppau, 9 September 1913; Library of Congress country study (nitrate about half of government revenue, 1880–1920); exports US$279m (1929) to US$35m (1932); COSACH, 1931–January 1933. Indigo: BASF's synthetic indigo, July 1897; world natural output about 19,000 t in the mid-1890s to about 1,000 t by 1913; Calcutta export returns. Ostrich feathers: South African trade statistics 1913 (feathers the fourth export after gold, diamonds and wool); the 1914 collapse. Tin: the International Tin Council's default of 24 October 1985 and the London Metal Exchange's suspension of tin trading to 1989. Iron ore, potash, oil, aluminium and nickel: company and industry statistics as cited (the four largest seaborne iron-ore exporters about 70% of 2024 exports; Uralkali's exit from BPC, 30 July 2013; OPEC's 1986 and November 2014 decisions; United States v. Alcoa, 12 March 1945; Inco's share of nickel in 1950). United States Geological Survey. Mineral Commodity Summaries 2025 (South Africa: platinum about 70% of world mine output; palladium 38%; chromite 46%; manganese 40%). International Platinum Group Metals Association (rhodium above 80%). LBMA and the Chamber of Mines (South African gold output 1,000 tonnes in 1970, about 70% of Western output); World Gold Council (about 100 tonnes in 2024). Sibanye-Stillwater (Neal Froneman, 2017, on taking platinum out of the market). Anglo American plc. Demerger of Valterra Platinum, May 2025; placing of the residual 19.9%, 4 September 2025. SOURCES FOR THE PAY, THE RETURN, THE CHRONOLOGY AND THE ASSETS Anglo American plc. Directors' Remuneration Report 2011 (Cynthia Carroll: salary £1,170,000, bonus £962,000, benefits £42,000, total £2,174,000; 33,492 notional LTIP shares vested, £1,090,834; pension contributions £351,000; Sir John Parker, chairman: £650,000 in 2010 and 2011). Anglo American plc. Annual Report 2018 and 2021 remuneration reports as reported (Mark Cutifani single-figure totals £14.2m for 2018 and £9.8m for 2021: mining-technology.com, 5 March 2019; interactive investor, 8 April 2022). Duncan Wanblad's reported totals 2022–25 (Simply Wall St, from Anglo's reports, about US$30m in all). Cumulative totals for Carroll (2007–13) and Cutifani (2013–22) are not given in this report because the single-figure tables for every year could not be read. Anglo American plc. Appointment of Cynthia Carroll as chairman of De Beers, 28 September 2012 (De Beers results for 2012); announcement that Cynthia Carroll will step down as chief executive, 26 October 2012. De Beers Group. Mark Cutifani appointed chairman of De Beers Group, 18 July 2013. Alstom. Philippe Mellier leaves Alstom (appointment at De Beers effective 1 July 2011), May 2011. De Beers Group. De Beers commits major capital investment to extend the life of Venetia, 19 March 2013 (the underground project approved by the De Beers and Anglo American boards in 2012); Anglo American Annual Report 2012, approved projects (Venetia underground, capex "under $3 billion", first production 2021). De Beers Group. Partnering with host countries (De Beers Consolidated Mines: 74 per cent De Beers, 26 per cent Ponahalo). De Beers Group and Mining Technology. Venetia production pause, 13–14 July 2026 (two years; about 4,400 employees). Anglo American plc. Results 2023, 2024 and 2025, project tables (Venetia underground: US$2.3bn total with US$0.8bn remaining at end-2023; about US$2.6bn with about US$0.8bn remaining at end-2025; first underground production June 2023; steady state "mid-2030s"). Anglo American plc. Half-year financial report 2026, 30 July 2026 (De Beers: revenue US$1,583m against US$1,952m; underlying EBITDA US$(113)m against US$(189)m; realised price US$105/ct against US$155/ct; unit cost US$64/ct against US$87/ct; production 14.9 Mct against 10.2 Mct; consolidated sales 12.4 Mct against 11.0 Mct). Anglo American plc. Production report for the third quarter of 2026, 20 October 2026 (production 6.8 Mct; year-to-date realised price US$97/ct; "continue to work towards a transaction in a challenging market"; 2026 guidance 25–28 Mct), and Rapaport's reading of it (third-quarter realised price US$84/ct; consolidated revenue US$731m). Anglo American plc. 2025 annual results script and question-and-answer transcript, 20 February 2026 (John Heasley: the US$2.3bn carrying value "is on an enterprise value basis"; attributable US$1.9bn; "reduced inventory by $0.9 billion in the year"; "midstream inventory at broadly normalised levels"; Duncan Wanblad on the shape of the sale). Anglo American plc. Half-year 2025 question-and-answer transcript, 31 July 2025 (Heasley: De Beers' funding "really all intercompany financing"; inventory about US$2.3bn at end-2024 and about US$1.8bn at mid-2025). Financial Times, as reported by Mining.com, 31 December 2024 (Al Cook: inventory "roughly $2 billion", the largest since 2008). De Beers Group. De Beers Group announces intention to close Lightbox business, 9 May 2025 (Element Six "will focus only on industrial synthetic diamonds" and "is centralising CVD production at its facility in Oregon"). National Jeweler. Take a look at Lightbox's new $94m manufacturing lab, 30 October 2020 (Gresham, Oregon; 200,000 carats a year). Chimia, vol. 15 (1961), Synthetische Diamanten (De Beers' first synthesis, September 1958, Adamant Research Laboratory, Johannesburg, under J.F.H. Custers; announced by Harry Oppenheimer, 17 November 1959). Canada NewsWire, De Beers changes its name to Element Six, 30 September 2002. Element Six. About us ("a global leader in synthetic diamond"). SOURCES FOR THE REGULATORS, THE COUNCIL AND THE PRODUCERS' BODIES BBB National Programs, National Advertising Division. National Advertising Division recommends Natural Diamond Council discontinue certain claims, 22 April 2021 (challenger Diamond Foundry; the claims on emissions, scarcity, resale value and "real"; the council "agrees to comply with the NAD's recommendations"). JCK (R. Bates). Watchdog chides Natural Diamond Council over ad claims, 22 April 2021. Gemological Institute of America. GIA to introduce new descriptive terminology for laboratory-grown diamonds, 2 June 2025 (Tom Moses: "More than 95% of laboratory-grown diamonds entering the market fall into a very narrow range of color and clarity"). Centurion Jewelry News. Revised FTC guidelines remove "natural" from definition of diamond, 25 July 2018. United States Federal Trade Commission. FTC sends warning letters to companies regarding diamond ad disclosures, 2 April 2019. Mining Weekly. De Beers, six others form Diamond Producers Association, 28 May 2015 ("at least $6 million a year"). Rapaport. DPA annual budget rises to $70m, 24 January 2018 (about US$50m in 2017; about US$70m in 2018). Rapaport. Alrosa to stop funding the Natural Diamond Council, 6 March 2022, and Russia crisis puts diamond marketing in jeopardy, 27 June 2022 (Alrosa 45 per cent of a US$70m budget). JCK. Natural Diamond Council gets a boost, 20 June 2025 (budgets: 2021 US$84m; 2022 US$95m set, about US$62m realised; 2023 US$54m; 2024 US$36m; 2025 US$22–38m). National Jeweler. "A Diamond Is Forever" is back again, backed by $20m, 15 September 2023. World Platinum Investment Council. Launch release, London, 18 November 2014 (founded by Anglo American Platinum, Aquarius Platinum, Impala Platinum, Lonmin, Northam Platinum and Royal Bafokeng Platinum). Miningmx. Sibanye-Stillwater bails from WPIC, 20 November 2019. Platinum Guild International. What we do (founded 1975; funded by the producers). Centurion Jewelry News. PGI is now solely funded by Anglo American Platinum, 9 May 2023. Neither body publishes its budget. SOURCES FOR THE METAL AND THE STONE LBMA gold PM fixes: 4 November 2011 US$1,749.00 (SD Bullion daily table); 16 August 2012 AM fix US$1,603.50 (London fixing archive; the PM fix for that day was not retrieved and lies within US$1,603–1,617); 2022 annual average US$1,800.09 (Kitco/LBMA, 27 January 2023); 30 October 2026 US$3,874.40. Nedbank Economics Unit, monthly average exchange rates (November 2011 R8.1723; August 2012 R8.2822); R16.71 at 30 October 2026. Anglo American plc. Completion of the acquisition, 16 August 2012 (US$5.2bn). World Gold Council. Gold Demand Trends, full years 2022 to 2025 (central-bank purchases 1,136 t, 1,037 t, 1,045 t and 863 t against total demand of 4,741 t, 4,899 t, 4,974 t and 5,002 t). Author's arithmetic: US$5.2bn at US$1,603.50 is 3,242,906 oz, worth US$12.56bn at US$3,874.40 (2.42×); in rand, R43.07bn in August 2012 and R209.95bn at 30 October 2026 (4.87×); US$6,500 at US$1,749 is 3.72 oz, worth US$14,400 (2.22×), and R53,000 against R241,000 in rand (4.5×). Rapaport. RapNet Diamond Index releases: 3 January 2023 (2022 −10.7%); 5 September 2023 (one-carat 6,403; −24.8% year on year, from which the 2022 level is derived); 6 January 2026 (2025 −9.9%); 4 August 2026 (one-carat 3,880); 6 October 2026. IDEX Online. Polished price index, 5 January 2026 (peak March 2022; 2023 −17.9%; 2024 −13.7%; 2025 −10.9%). Zimnisky, P. Global Rough Diamond Price Index (all-time high 207.3, February 2022; 123.1 in November 2025, −40.6%), via Stockwatch, 7 November 2025. Anglo American plc. Production reports for the fourth quarter of 2025 (5 February 2026) and the first quarter of 2026 (De Beers' rough price index, December 2006 = 100: 107 for 2024; 94 for 2025, 80 including stock rebalancing; 68 in the first quarter of 2026). StoneAlgo. Diamond prices, one-carat round, 30 October 2026 (US$4,737, blended grades). TheDiamondPrice.com. Natural v laboratory-grown one-carat medians, October 2026 (laboratory US$716). PriceScope. Diamond price roundups, 2011 (trend data; the one-carat 2011 retail figure in this report is an estimate from them), May 2021 and November 2022 (G VS1 one-carat US$7,909 and US$8,714 a carat). JCK. Lightbox lowering prices, 10 May 2024 (US$500–900 a carat). Edahn Golan, AWDC State of the Market, May 2026. Bain & Company and AWDC. The Global Diamond Industry: Portrait of Growth, December 2012 (2011 pipeline: rough US$15bn, polished US$24bn, retail US$71bn). De Beers. Diamond Insight Report 2023, as reported by Rapaport (2022 pipeline: US$17.2bn, US$27.8bn, US$87bn). GJEPC and IDEX Online, 26 May 2026 (India's natural polished exports US$12.16bn on 16.0 Mct; laboratory polished 18.8 Mct; laboratory volume above natural from April 2026). Angel One (Commerce Ministry data), 14 August 2026 (laboratory export price US$388.77/ct in 2022 and US$71.13/ct in January–May 2026). National Jeweler. Q&A: Al Cook on the lab-grown bubble, 9 June 2026 ("$30 and $100 a carat"). Epstein, E. J. Have you ever tried to sell a diamond? The Atlantic, February 1982 (the resale tests, as summarised; the original text was not retrieved for this report); interview, Frontline, The Diamond Empire, 1994. Christian Science Monitor. Review of The Rise and Fall of Diamonds, 7 July 1982 (resellers receive about sixty per cent of wholesale). Marketplace (American Public Media). Diamonds are not a jewel investment, 22 February 2013 (a best offer of thirty-five per cent of the price paid). StoneAlgo. Where to sell a diamond online, 19 January 2023 (cash offers forty to fifty per cent below retail). Rio Tinto. Historic 2021 Argyle pink diamonds tender delivers record-breaking results, 27 October 2021 ("double digit price growth"; the mine closed 3 November 2020). Australian Resources & Investment, 27 October 2021 (the Australian Diamond Portfolio pink index +30 per cent in twelve months; a dealer's estimate, and marked as one). Autorità Garante della Concorrenza e del Mercato. Press release, 30 October 2017 (fines of €15.35m on Intermarket Diamond Business, Diamond Private Investment, UniCredit, Banco BPM, Intesa Sanpaolo and Monte dei Paschi). ANSA, 19 February 2019 (the seizure of more than €700m). Panorama, 31 January 2019 (IDB declared bankrupt by the Milan court, 15 January 2019). Banca d'Italia, survey of diamond sales through banks, reported to Parliament 8 February 2022, via Milano Finanza (€1.878bn sold; 71,108 customers; commissions €272.7m; €1.160bn refunded to 52,440 customers by 30 September 2021). Rio Tinto. 2020 Argyle Pink Diamonds Tender results, 17 December 2020 ("appreciated over 500 per cent over the past two decades"). National Jeweler, 15 May 2024 (the Argyle Phoenix, 1.56 ct, US$4.2m at Phillips Geneva). Sullivan, J. C. How diamonds became forever. The New York Times, 3 May 2013, and interview, NPR Here & Now, 2013 (Frances Gerety's own account of writing the line late at night; the agency's initial reaction; the line in every De Beers advertisement from 1948). WARC. De Beers Consolidated Mines: A Diamond Is Forever campaign, 1947–1990s. Ad Age. Top advertising slogans of the century, 29 March 1999. Oppenheimer, N. Farewell address to Anglo American, Vergelegen, 30 November 2011 (Brenthurst Discussion Paper 2011/11), quoting Sir Ernest Oppenheimer's chairman's statement of 1954: "to earn profits, but to earn them in such a way as to make a real and permanent contribution to the well-being of the people and to the development of Southern Africa". Boyajian, W. E. An economic review of the past decade in diamonds. Gems & Gemology, Fall 1988 (CSO sales US$2,723m in 1980, US$1,472m in 1981, US$1,257m in 1982). Christian Science Monitor. How diamond miners manipulate market to profit in slump, 18 August 1982; De Beers cuts dividend, 25 August 1982 (the interim halved); Losses aren't forever, De Beers shows, 29 January 1987 (buying surplus stones to maintain prices; the stockpile doubled to US$1.95bn by 1984). Acemoglu, D., Johnson, S. and Robinson, J. An African success story: Botswana (2001; in Rodrik, ed., In Search of Prosperity, 2003): per-capita growth of 7.7 per cent a year, 1965–98, the highest in the world. World Bank, constant GDP per capita for Botswana (FRED mirror): US$508 in 1966, US$7,160 in 2022, constant 2010 dollars. Wikipedia and Debswana histories. Orapa discovered 1 March 1967; about US$70 a head at independence (nominal). SOURCES FOR THE FORTUNES Northern Miner. Consortium offers US$17.6 billion for De Beers (February 2001); De Beers offer sweetened; De Beers buyout okayed (May 2001: US$18.7bn; DB Investments owned 45 per cent by Anglo, 45 per cent by the family, 10 per cent by Debswana). Anglo American plc. Press release, 20 December 2004 (Botswana's direct 15 per cent from 1 August 2004); results releases 6 February 2003, 25 July 2003, 5 February 2004 and the Mail & Guardian, 3 February 2005 (De Beers' ordinary dividends US$400m for 2003 and US$450m for 2004). How We Made It In Africa. South Africa's Oppenheimer family sets up Africa fund, 2011 (Tana Africa Capital, US$300m with Temasek). Billionaires.Africa, 5 October 2026 (Fireblade Aviation, 2014; Oppenheimer Generations, 2018; the Future Trust, 2020). Bloomberg via Mining.com, 20 October 2021 (the family's wealth about US$8.2bn). Kitco/Reuters, 31 January 2023 (Debswana sales US$4.588bn in 2022, a record; diamonds about 30 per cent of revenue and 70 per cent of foreign exchange); Kitco/Reuters, 1 March 2024 (2023: US$3.44bn); Mmegi, February 2025 (2024: P24.6bn, −46 per cent). Mining.com/Reuters, 20 January 2026 (the state's stockpile of 12 million carats against an allowable 6.5 million; mineral revenue P10.3bn for 2025–26 against an average of P25.3bn). Bank of Botswana. Moody's downgrade to Baa1, 17–18 October 2025; S&P downgrade to BBB−, 13 March 2026. International Monetary Fund. Botswana: staff completes 2025 Article IV mission, 26 September 2025 (deficit 7.1 per cent of GDP for 2024–25; GDP −3.0 per cent in 2024; about −1 per cent in 2025). S&P Global Ratings via AllAfrica, 22 March 2026 (deficits 9.3 and 8.9 per cent; reserves US$3.8bn at end-2025 from US$7.5bn in 2017). Antwerp World Diamond Centre. Antwerp diamond industry trade figures 2025, 12 January 2026 (US$19.07bn, −22.4 per cent; about US$41bn in 2022). Rapaport. Eurostar Diamond Traders close to bankruptcy, 12 March 2019, and Antwerp's Eurostar enters bankruptcy, 28 March 2019 (about US$500m of debt; inventory marked at €132m in September 2018 and admitted at US$25m in November). KBC. KBC takes decision on Antwerp Diamond Bank, 19 September 2014. National Jeweler. Why Standard Chartered shut down its diamond unit, 15 June 2016; ABN AMRO closing New York and Dubai diamond lending offices, 21 May 2018. Rapaport. ABN AMRO to cease Hong Kong diamond lending, 13 August 2020. Wikipedia, Surat Diamond Bourse (opened 17 December 2023; Rs 3,200 crore; 4,200 offices). DeshGujarat, 18 December 2025 (about 40–50 offices operating). Rapaport Magazine. The quiet tragedy of Surat's diamond industry, 3 December 2025 (wages and hours; the union's count of at least a hundred suicides in two years). ThePrint via the Business & Human Rights Resource Centre, 4 November 2024 (the Diamond Workers Union Gujarat's count of 71 suicides in eighteen months). News Arena India, 24 May 2025 (the Gujarat relief package). Outlook India, 28 August 2025 (the tariff layoffs). StockAnalysis. Henan Liliang Diamond, income statements 2021–24 (revenue CNY 498m, 906m, 752m, 686m; net profit CNY 240m, 460m, 364m, 201m). CRISIL. Rating rationale, Greenlab Diamonds LLP, 1 July 2022 (operating margin 60–66 per cent). Bloomberg via Mining.com, 22 August 2022, and GJEPC via Rough & Polished, March 2023 (India's laboratory polished exports about US$1.3bn in 2021–22 and about US$1.7bn in 2022–23); Angel One, 14 August 2026 (US$1.19bn in 2025 on 14.95 Mct). National Jeweler. Pandora's moving from natural to lab-grown diamonds, 4 May 2021. JCK, 6 February 2025 (Pandora's laboratory-grown revenue DKK 315m in 2024). Rapaport. Signet's focus on lab-grown fashion jewelry is paying off, 4 September 2025 (14 per cent of fashion sales). Brilliant Earth. Fourth-quarter and full-year 2024 results, 12 March 2025 (gross margin 60.3 per cent). JCK, 27 March 2024 (Golan: retail gross margin about 64 per cent on a one-carat laboratory stone against 40 per cent on a mined one). IDEX Online, 22 March 2026, and Engineering News/Bloomberg, 20 March 2026 (the sightholder list cut from 69 to about 45 for the contract from 1 July 2026). National Jeweler. Petra Diamonds completes restructuring, 11 March 2021 (noteholders 91 per cent). Petra Diamonds. Annual report 2025 (net loss US$116m). Mining News North, 6–9 October 2026, and Anglo American's third-quarter report (Mountain Province's 49 per cent of Gahcho Kué transferred to De Beers; full ownership from 16 October 2026). Rapaport, 17 July 2026, and Engineering News, 15 October 2026 (the Global Diamond Consortium; Pluczenik; Richemont and Signet in talks). Fortune, 20 February 2025, and IDEX Online, February 2025 (Chinese diamond demand down about 40 per cent in two years; Paul Zimnisky: by as much as 50 per cent in 2024). SOURCES FOR THE COMPARISON TABLE Rio Tinto. Results for 2008 (aluminium impairment US$7.9bn after tax), 2011 (US$8.9bn, announced 9 February 2012), 2012 (US$11.0bn, announced 17 January 2013) and 2013 (US$1.3bn); the Alcan acquisition at US$38.1bn enterprise value, July 2007. Kinross Gold. Results of 15 February 2012 (Tasiast goodwill impairment US$2,490m), 13 February 2013 (US$3.09bn after tax) and 31 July 2013 (US$1,335m after tax); the Red Back acquisition at US$7.1bn, completed 17 September 2010. BHP. Acquisitions of Fayetteville (US$4.75bn) and Petrohawk (US$12.1bn plus about US$3bn of debt), 2011; onshore US asset review, 15 January 2016 (US$7.2bn pre-tax); sale to BP for US$10.5bn and Merit Energy for US$0.3bn, 27 July 2018; cumulative impairments assembled from BHP's annual results and marked approximate. Anglo American plc. Acquisition of 49 per cent of Minas-Rio, 23 April 2007 (US$1.15bn); acquisition of the remainder with 70 per cent of Amapá, 17 January 2008 (about US$5.5bn); capital cost raised to US$8.8bn and impairment of US$4.0bn, 29 January 2013; impairment of US$2.9bn, 24 July 2015. THE WORKING FOR THE EXHIBITS The metal and the stone (Section II) and the believers (Section XIX): gold ounces are the dollar sum divided by the LBMA PM fix of the date (the 16 August 2012 AM fix where the PM fix was not retrieved), valued at US$3,874.40; rand values use the Nedbank monthly average rate of the month of purchase and R16.71 at 30 October 2026. Anglo's return (Section X): the forty per cent's share of any dividend paid to all shareholders is 40 per cent; sale proceeds 40/85 of Anglo's reported US$1bn in 2027; money multiple is cash back divided by US$5.2bn; the internal rate of return is on annual cash flows with dividends spread evenly over 2013–22; De Beers' underlying EBITDA 2013–22 sums to US$11,837m from the table in Section X. The cost of custody (Section XIV): about 24 Mct of De Beers' own saleable volume at US$105/ct; a third of sales withheld for two years with the mines at full output; interest at 8 per cent on the new stock as it builds; operating losses at full output of US$0.2bn a year on the H1 2026 run-rate and US$0.5bn on the 2025 record; the existing inventory of about US$1.4bn at end-2025, by author's arithmetic from Anglo's statements, is counted as already financed. The margin chain (Section XII): the pipeline ratios are Bain's 2011 and De Beers' 2022 totals; the one-carat figures are the best dated figure at each stage and are not a matched basket; the 2011 retail figure and the 2022 RapNet level are estimates and are marked. The custodian's scorecard (Section XX) is built from the figures in the table and the accounts cited, and every assumption is stated in the text. All of these are the author's calculations, not the companies'. SOURCES FOR THE FAMILY'S TRADE (SECTION VIII) Anglo American plc. AA plc and De Beers, RNS of 15 February 2001 (US$17.6bn; 0.43 Anglo shares and US$14.40 cash per linked unit; the unwinding of the 35.4% cross-holding; Anglo's US$943m cash and US$701m of preference shares; Central Holdings' contribution of about 10.4 million Anglo shares and US$37.5m cash; the US$3.55bn senior facility). Northern Miner. Consortium offers US$17.6 billion for De Beers; De Beers offer sweetened after institutions demur; De Beers buyout okayed (2001: Central Holdings' 45% "costs US$1.2 billion"). CNN Money, 15 February 2001, and Rapaport, Nicky's Big Deal (the diamond business at US$8.3bn; Rapaport's own US$10.7bn). Anglo American plc. De Beers results releases of 6 February 2003, 5 February 2004, 23 July 2004, 10 February 2006, 9 February 2007, 8 February 2008, 23 July 2008, 11 February 2010 and 11 February 2011 (net debt US$3,152m at end-2001, US$906m at end-2003, US$4,057m at end-2007, US$3.2bn at end-2009 excluding shareholder loans, US$1.76bn at end-2010; the dividends and preference redemptions Anglo received each year; the US$500m shareholder loans of 2009 and the US$1bn rights issue of February 2010, Anglo US$450m); De Beers results for 2011, 10 February 2012 (sales US$7,378m; EBITDA US$1,721m; DTC prices +29%); half-year results, 26 July 2011 (the record first half; rough prices up about 35%). Anglo American plc. Press release of 4 November 2011 (US$5.1bn; interest at 3.5%; the US$50m and US$75m protections; the listing uplift; Botswana's pre-emption; CHL's shareholder loans of US$265m; EOSIL's 25.2 million Anglo shares) and of 16 August 2012 (US$5.2bn). Anglo American plc and De Beers. Supplier of Choice: the European Commission's favourable position, 16 January 2003 (Gary Ralfe: "truly transformational"). United States Department of Justice. De Beers pleads guilty to price fixing, 13 July 2004 (US$10m). United States District Court for the District of New Jersey, the De Beers class action settlement of US$295m, approved May 2008. JCK. De Beers' Time of Reckoning, 2000 (Ralfe at the GIA symposium: "almost no value" on the stockpile; "no sacred cows"); Gareth Penny, Person of the Year, 1 February 2004. Mail & Guardian. De Beers to review its diamond inventories, 18 April 2000 ("the burden of swollen inventories"); Oppenheimers to keep De Beers stake, 17 January 2011; De Beers no longer jewel in crown, 11 November 2011 (RBC: 25–30% below its valuation; "prices were insane, at a peak"; "stuck with this asset anyway"). Miningmx. New De Beers CEO fits the bill, 6 June 2011; The Oppenheimers eye new projects in Africa, 4 November 2011; Why the Oppenheimers sold out of De Beers, November 2011; Oppenheimers sell part of Anglo stake, 15 December 2010. Fin24. Diamond dynasty, 27 November 2011. Family Business Magazine. Oppenheimer family sells its stake in De Beers (the Bloomberg-cited analyst: "on the low side, but for the Oppenheimers there was only one buyer"). Northern Miner. Anglo takes the reins at De Beers, November 2011 (Peter Major: "I don't think Anglo overpaid"). City AM, 19 December 2010 (the Sunday Times report that Anglo was preparing to buy the family out). The family's return is the author's arithmetic on these inputs: about US$1.2bn of cost in 2001, mostly in Anglo shares, against US$5.2bn received in August 2012 and about US$0.9–1.1bn of gross dividends less about US$0.6bn reinjected: about 14 per cent a year, unadjusted for the 2004 step from 45 to 40 per cent. SOURCES FOR THE ALL-IN ACCOUNT (SECTION X) Anglo American plc. Preliminary results 2014, 13 February 2015 (De Beers capital expenditure US$476m for 2013 and US$689m for 2014; De Beers' 51% share of Gahcho Kué capex about US$0.5bn; 2014 net finance costs and income tax expense US$264m); Preliminary results 2015, 16 February 2016 (capex US$697m; net finance costs and tax US$274m; Snap Lake impairment); Preliminary results 2016, 21 February 2017 (capex US$526m; net finance costs and tax US$242m); Preliminary results 2017, 22 February 2018 (capex reduced by 48% to US$273m); Preliminary results 2018, 21 February 2019 (US$417m); Preliminary results 2019, 20 February 2020 (US$567m); Preliminary results 2020, 25 February 2021 (US$381m; Venetia underground US$2.1bn with US$1.3bn remaining); Preliminary results 2021, 24 February 2022 (US$565m); Preliminary results 2022, 23 February 2023 (US$593m; Benguela Gem about US$0.2bn at 50%); Preliminary results 2023, 22 February 2024 (US$623m; rough stock written down US$0.2bn); Preliminary results 2025, 20 February 2026 (US$353m against US$536m for 2024; Venetia underground about US$2.6bn spent with about US$0.8bn remaining; Jwaneng Cut-9 about US$0.4bn at Anglo's 19.2%; trading losses of US$424m on stock rebalancing; De Beers losses carrying no or limited tax benefit); Half year financial report 2026, 30 July 2026 (capex US$115m; EBITDA loss US$113m; the formal sales process). Anglo American plc, press release of 16 August 2012 (US$5.2bn). In none of these documents is a De Beers dividend, net debt or intercompany funding figure published; the distribution bracket, the 2024-26 funding line and the carry on the stock are the author's arithmetic on the segment figures above. SOURCES FOR THE FAMILY'S OTHER EXIT (SECTION IX) Anglo American plc. Notice of Annual General Meeting and Shareholder Information 2005 (Substantial Shareholdings at 22 February 2005: E. Oppenheimer & Son Holdings Limited, 52,250,206 shares, 3.49%; notified interests of E. Oppenheimer & Son International Limited, E.O. & Son Holdings Limited, Maitland Trustees Limited and Respite Holdings Limited; 1,493,839,387 ordinary shares in issue at 31 December 2004). Anglo American plc. AA plc and De Beers, RNS of 15 February 2001 (the 35.4% cross-holding; Central Holdings' contribution of about 10.4 million Anglo shares). Anglo American plc. Press release of 4 November 2011 (all-cash consideration; EOSIL's 25.2 million Anglo shares not to be voted). Anglo American plc. Press release of 25 February 2011 and Miningmx, Oppenheimer to retire from Anglo's board, 24 February 2011; The Jewish Chronicle, End of era for Anglo American, 3 March 2011 (retirement at the AGM of 21 April 2011 after 37 years). Mail & Guardian. Tycoon buys stake in Anglo American, 11 November 2006 (17 million shares to Larry Yung; about US$800m). The Namibian. Oppenheimers sell third of family stake, November 2006 (17 million shares, 33% of the holding; "sectors that are traditionally counter cyclical"; Anglo at R343.98 in Johannesburg). JCK (R. Bates). Oppenheimer Family Sells Anglo Shares, 1 February 2007. IDEX Online. Oppenheimer Sells Anglo American Shares, 15 December 2010 (2.1 million at £30.44, £63.91m; the 2006 sale of 17 million shares, a 1.01% holding, for about £425m, leaving 2.29%). Miningmx. Oppenheimers sell part of Anglo stake, 15 December 2010 (25.2 million remaining, just under 2%; James Teeger: "We have no intention of selling further shares"). Mining Journal. Oppenheimers sell shares in Anglo American, December 2010 (£63.9m; 8% of the holding). Family Business Magazine, citing Bloomberg and the Financial Times, November 2011 (the stake reduced "to 1.9% from 8% in 2000"; Jonathan Oppenheimer "blocked from taking his place for corporate governance reasons"). Business Day. Oppenheimers halve Anglo stake, 22 March 2017 (the holding cut from 1.8% to 0.9%, to 12.6 million shares, in the December 2016 quarter; the prior holding of 27.3 million before December 2010; Anglo's range of R149 to R218 in the quarter; Teeger on valuation; Volcan's 13%). Sunday Times. Anglo exit from SA the 'right thing to do', 26 March 2017 ("Anglo was becoming less strange and more normal"; "if we thought the price was high we would sell them"; the ethos ended with the 1999 move to London; the stake "of only 0.9%"). Billionaires.Africa. Nicky Oppenheimer's family stock portfolio returned nearly 25% in 2025, 18 October 2026 (Stockdale Street Ltd accounts; no listed holdings disclosed). The price record. City AM. The best performing FTSE 100 stock of 2016, December 2016 (the intraday low of about 215p in January 2016; the lowest daily average of 221.1p; 277.9p on 1 January and 1,165.5p on 28 December 2016). Mining Weekly, 4 October 2016 (the record low of 215.55p). Motley Fool, 31 December 2012 (1,856p at the year end). Digrin, Anglo American monthly price and dividend series (US dollar prints of about US$46.90 for November 2006, US$27.55 for August 2012, US$4.05 for January 2016 and US$13.80 to US$15.21 for the December 2016 quarter; the ex-dividend record from 2012 to 2026, summing to about US$15.3 a share paid from September 2012 to September 2026, including the 80 cent special of 2021). Hargreaves Lansdown, Anglo American dividend table 2022 to 2026. Trading Economics and Global Banking and Finance, 30 October 2026 (3,704p; market value about £39.7bn); The Daily Dispatch, Anglo American share price, 6 October 2026 (the record of 4,348p on 8 September 2026). The 170 to 190 million share counterfactual, the drawdown and the return are the author's arithmetic on these inputs at sterling rates of about US$1.57 in 2012, US$1.42 in January 2016 and US$1.32 in October 2026; Anglo's share count of about 1.4 billion in 2012 is from its Annual Report 2012. Reference for Business, Anglo American PLC company history (Anglo the largest single shareholder in De Beers in 1926; the bid of May 1927; the chair in December 1929 with Rothschild support through Morgan Grenfell; the Diamond Corporation of February 1930). The sequence is also in Section V. SOURCES FOR THE SELLER'S RECORD Anglo American plc, announcements of the disposals: the Free State and Elandsrand mines to Harmony (2001–02); 20% of Gold Fields to Norilsk Nickel (29 March 2004, R7.63bn); AngloGold Ashanti share sales of 20 April 2006 (19.7m at US$51.25), 2 October 2007 (67.1m at US$44) and 17 March 2009 (39.9m to Paulson & Co at US$32); Rustenburg to Sibanye (2016) and Kroondal (2023); the Thungela demerger (June 2021, with R2.5bn of cash); the placing of 19.9% of Valterra Platinum at R845 (September 2025); the De Beers 40% purchase (November 2011, completed 16 August 2012). Market values at the closes of 22 October 2026 (AngloGold Ashanti US$85.19; Gold Fields US$35.14; Valterra R1,165.16; Thungela R19.7bn; Harmony and Sibanye-Stillwater at the ex-Anglo mines' share of earnings applied to market value) and dividends per share since each sale from the companies' records, as compiled in The Destruction of Anglo American, Northbound Research Exposé, October 2026, the one exhibit in this report drawn from another Northbound paper, at the author's instruction. Duncan Wanblad at the FT Mining Summit, October 2024 ("There is no way De Beers is a fit in the portfolio"); Anglo American, Portfolio simplification announcement, 14 May 2024.PLATES · FROM THE PERSONAL COLLECTION OF THE AUTHOR Two certificates, 1914 and 1946 The two documents reproduced here are from my personal collection of share certificates, photographed for this report. They bracket the event this report turns on. The first was issued three years before Anglo American existed and twelve before Ernest Oppenheimer took a seat on the De Beers board; the second was issued seventeen years into his chairmanship, and carries the stamp of his company as the keeper of the De Beers share register. Both are cancelled. The paper outlasted the promise printed on it. .The warrant is bilingual because it was made for Paris. De Beers was one of the most widely held shares on the Bourse before 1914, and a bearer warrant, which passed from hand to hand without a register, was the continental investor's instrument. The medallion at the top left is the French Treasury's stamp for the 2 per cent duty on foreign securities, struck in Paris in May 1914, three months before the war closed the diamond market and De Beers stopped its mines. The capital printed on the face is the structure Rhodes left behind in 1888: 800,000 preference shares of £2 10s carrying a fixed cumulative dividend of 40 per cent, which is £1 a share a year, ranking ahead of 1,000,000 deferred shares that took whatever was left. The preference shares were the safe paper; the deferred were the bet on the stone. The director's line is signed in his own hand by H. P. Rudd, a son of Charles Dunell Rudd, the partner who had stood beside Rhodes since the first De Beers Mining Company of 1880 and whose name is on the Rudd Concession of 1888; the son sat on the De Beers board in the second generation of the founding families. Anglo American did not yet exist. Ernest Oppenheimer was a Kimberley diamond buyer and the town's mayor, with no seat on the board that sealed this warrant. .Thirty-two years separate the two documents and the difference is in the small print. The share capital has risen to £5,500,000, the deferred shares to 1,400,000, and the 40 per cent preference dividend on £2 10s is unchanged. The certificate is a registered one, not a bearer warrant: the holder's name is on the face, the transfer is recorded in the books of the company, and the distinctive numbers of the twenty-nine shares run down the left-hand margin in six parcels, which is how a small holding was built up by purchase over the years. The telling mark is the one added after the holder's death. The stamp of 1951 is that of the Anglo American Corporation of South Africa, Limited, as registrar: by then the register of De Beers was kept in the offices of the company Ernest Oppenheimer had founded in 1917, had used to take control of the diamond fields between 1926 and 1929, and had chaired alongside De Beers ever since. The certificate was issued in the first full year of peace, with the stockpile intact, the Diamond Producers' Association in control of supply and the single-channel system at full strength. A year later, in 1947, a copywriter in Philadelphia would write the line that this report opens with. The company that paid for it was, by then, Anglo's.

Dr Duarte da Silva: A diamond is forever. De Beers was not.

Related Stories

Dr Duarte da Silva on Anglo (pt 2): The squandered house

BizNews monthly webinar: What's in, what's out, and why

Duarte Da Silva: The destructIon of Anglo American

Dr Duarte da Silva: The richest beggar on earth