Dr Duarte da Silva: The Platinum corollary

South Africa holds 83% of the world’s platinum-group metal reserves, yet platinum sells for less than half the price of gold, even though the world mines about 21 times more gold. In the latest Gold Papers instalment, Dr Duarte da Silva argues the reason is the buyer: central banks and investors price gold as money, while carmakers and chemical plants price platinum as an industrial input. That is why Bushveld miners, written off at $895 an ounce in 2025, multiplied their earnings within a year, then watched platinum give back more than two-fifths of its record. His lesson for SA gold producers is to stop planning gold as if it were platinum..By Dr Duarte F da Silva*.The argument for South African gold was built on the Witwatersrand: an endowment mispriced against its geology, a burial by price mistaken for a burial by geology, a jurisdiction discount that punishes the ground and not the operator, a second life that recovers metal instead of digging for it. A fair reader may ask whether it works anywhere else. The Bushveld Complex holds eighty-three per cent of the world's known platinum-group reserves and produces seven of every ten ounces of platinum mined. Its producers cut thousands of jobs in 2023 and 2024 on a basket price that had halved, and in the eighteen months to this week reported earnings many times those of a year earlier, with Valterra's realised basket price up eighty-five per cent in the first half of 2026. The market ran three consecutive years of platinum deficit to 2025, the last of 1.44 million ounces, and on 9 September 2026 the World Platinum Investment Council turned its 2026 forecast into a small surplus as investors sold. Hydrogen, the buyer the industry promised, took 71,000 ounces in 2025, far below the path to the 875,000 forecast for 2030, while demand from data storage and semiconductors has taken ruthenium to two and a half times its 2025 price. A quarter of platinum supply is already recycled, and a company that treats chrome tailings produced a record 95,885 ounces in the year to June 2026 without sinking a shaft. On most tests the Bushveld behaves like the Witwatersrand. On price it does not, and the difference explains why gold's re-rating has lasted while platinum's has already given back more than two-fifths..THE DEPARTURE1. Why test a gold argument on platinumAn argument tested on one case may be peculiar to that case. The argument for gold has been made with one geology, one price history, one set of companies and one basin, and if each of its steps depends on something particular to the Witwatersrand it does not travel. The test is to apply it to the nearest thing the country has to a second Witwatersrand: a mineral province of the first rank, in the same jurisdiction, mined by the same houses, with a price that has just done in eighteen months what gold did over several years. The Bushveld Complex is that province. It is the largest layered igneous intrusion on earth, about two billion years old, and its two mineralised horizons, the Merensky Reef, found by Hans Merensky in 1924 and thirty to ninety centimetres thick, and the UG2 chromitite beneath it, hold about three-quarters of the world's platinum resources and half its palladium. It is mined by Valterra, Impala, Sibanye-Stillwater and Northam, whose ancestors were the Witwatersrand's mining houses, with the same narrow-reef, hand-held methods, on the same power grid and under the same regulator..The gold argument has six steps, and each can be checked against the platinum record: the endowment set against what the market prices; the burial, where a verdict about price was written up as a verdict about geology; the discount the market charges for the address; the new buyer that re-rates a metal from outside its old cycle; the loop of recovered metal that competes with the mine; and the second life that pays without a shaft. The test has a limit. Showing that the method works on platinum does not prove the gold thesis right, only that the method is not peculiar to gold. It can also show where the two metals part company, and on the evidence below they part on a question the gold argument has tended to take for granted: what kind of buyer sets the price.Tested on the Bushveld, the gold argument mostly holds. It fails on price: the rarer metal sells for less than half as much as gold, and the reason is the buyer.EXHIBIT A · THE SAME GROUNDII. Eighty-three per cent of the world, on shallower groundThe endowment comes first. The United States Geological Survey's 2026 summary puts the world's platinum-group metal reserves at 76,000 tonnes and South Africa's at 63,000, eighty-three per cent, against Russia's 11,000, Zimbabwe's 1,300, the United States' 590 and Canada's 310. In 2025 the world mined an estimated 170 tonnes of platinum, of which South Africa produced 120, seventy-one per cent or seven ounces in ten, with Russia at 20, Zimbabwe at 18, Canada at 5 and the United States at 1.8. Palladium is the one metal where the Bushveld is second: Russia's 84 tonnes against South Africa's 70 of a 190-tonne world. Rhodium, iridium and ruthenium are, in effect, South African metals. Few mineral provinces hold a comparable share of any traded metal; the Witwatersrand at its peak in the early 1970s produced about seventy per cent of the world's gold.The Bushveld's ground is also easier than the Witwatersrand's. The platinum mines work mostly at one to two kilometres, against the three to four of the deepest gold mines, and the rock is cooler and the seismicity lower. The Minerals Council put the fall in the country's platinum-group metal output in 2025 at 4.1 per cent, and Statistics South Africa's latest reading, for July 2026, had PGM production 13.5 per cent below a year earlier, the largest drag on a mining sector down 7.5 per cent. Absa attributes part of that fall to a safety stoppage at Impala's Rustenburg mines and expects some of it back in the August figures, which were due on 13 October. South African platinum output is about 3.9 million ounces, 26 per cent below its 2006 peak of 5.3 million, as Kitco noted in reporting Northam's results on 28 August 2026, and the World Platinum Investment Council expects a rise of only about one per cent in 2026. Reserves have moved the other way, with the price: PwC's annual review of the country's mining companies, published on 6 October 2026, counted the platinum producers' reserves at 298 million 4E ounces, up from 261 million in 2023, and their average mine life at 43 years, up from 38. The falling output is the kind of sunset the gold producers wrote at Kloof: a supply curve drawn on a price deck, in a province holding eighty-three per cent of what the world will need for as long as it uses the metal. The ground is first-rank and the production trend is written as terminal, and the gap between the two is where the gold argument starts.Eighty-three per cent of the world's reserves and seven ounces in ten of its platinum, from shallower mines than the Witwatersrand's, with a supply curve the producers draw down from 5.3 million ounces in 2006 to 3.9 million now.EXHIBIT B · THE SAME HOUSESIII. The same lineage, grid and labour modelThe corporate lineage matters because half the gold argument concerns the custodians and not the rock. The names on the platinum belt are the gold houses one generation on, and the boards that took the platinum decisions of 2024 were schooled in the gold decisions of 2000 to 2020. Anglo American, founded in 1917, became the world's largest platinum producer through Anglo American Platinum and demerged it with effect from 31 May 2025 as Valterra, then sold its remaining 19.9 per cent on 4 September 2025 for about R44 billion and left the platinum business altogether; it is the house that created AngloGold, which has since moved its primary listing to New York. Union Corporation, which became part of Gencor, the other great gold house, built Impala, now the second-largest producer. Gold Fields' Kloof, Driefontein and Beatrix mines became Sibanye in 2013, which bought Aquarius and Anglo's Rustenburg mines in 2016, Stillwater in 2017 and Lonmin, owner of the Marikana mine, in June 2019, and is now a PGM company with a gold division. Northam is the exception, an independent that has grown by development and acquisition, and this year it opened a formal sale process after an approach the press attributed to the largest of the others. The custodial history of the platinum belt is the gold belt's, with the names moved one column across.The operating model is the same too. The Merensky and UG2 reefs are mined the way the Carbon Leader was mined: narrow, tabular, hand-held drilling on a stope face, labour-intensive to a degree no other mineral province of the first rank tolerates, on a grid whose tariff has risen more than 900 per cent in two decades. The belt's labour history is the country's. The killings at Marikana on 16 August 2012, when thirty-four striking miners died at Lonmin's mine, now Sibanye-Stillwater's, and the five-month strike of January to June 2014 that stopped the three largest producers and took more than a million ounces out of the market, put the same social contract under the same strain as the gold-mine strike of 1987. The custodians' response to the 2024 price was the gold response: shafts on care and maintenance, retrenchments in the thousands, capital withheld and a demerger. Retrenchment has continued into the recovery. On 8 September 2026, a week after reporting group adjusted earnings before interest, tax, depreciation and amortisation more than doubled, Sibanye-Stillwater opened consultations on restructuring its Kwezi shaft, with 1,114 jobs at risk, citing losses in 2024 and 2025 and the limits of the shaft's licensed mining area. These are the same companies, grid, unions and boards, taking the same decision on a different metal.Anglo demerged its platinum business and then sold out of it; Union Corporation's Impala passed to Gencor; Gold Fields' mines became Sibanye. The platinum belt's custodians are the gold houses one generation on, and in 2024 they took the gold decision on a shallower reef.EXHIBIT C · THE SAME BURIALIV. The obituary of 2024 and the recovery of 2025The platinum belt offers the clearest recent case in South African mining of a verdict about price written up as a verdict about geology. Between 2022 and early 2025 the PGM basket fell by roughly half: palladium from a record above 3,400 dollars in March 2022 to below a thousand, rhodium from above 20,000 to a fraction of it, platinum to a closing low of 895.03 dollars on 8 April 2025. The producers cut. Restructurings in 2023 and 2024 cut thousands of jobs across the Rustenburg and eastern-limb complexes; shafts were placed on care and maintenance; Anglo American, whose platinum subsidiary had been the largest producer in the world for a generation, announced it would demerge the business, and did so within the year. The trade press wrote the obituary of the South African PGM industry in the same register the gold obituaries were written in: structural decline, electric vehicles ending the autocatalyst, a jurisdiction too costly to mine. The cost pressure was real, electricity tariffs above all.Then the price moved. Platinum rose from about 895 dollars in April 2025 to a then-record above 2,450 dollars on 26 December, ended the year at about 2,030, more than double, and set an all-time high of 2,923.70 dollars on the New York futures market on 26 January 2026. It has since retreated: 1,694 dollars at the close on 12 October 2026, near a ten-week low, about seventeen per cent down on the year and forty-two per cent below the record, with gold at 4,172 dollars 2.46 times its price. Palladium is about 1,150 dollars, near its lowest since September 2025, and rhodium 9,200, up about a quarter in twelve months. The producers' results for the periods that caught the move are set out in the table below. PwC's count for the year to June 2026 has the platinum companies' revenue up fifty-nine per cent, to R429 billion, on a platinum price that averaged eighty per cent higher, with production still below its 2019 level. The common feature is that output barely changed and earnings multiplied by ten to thirty; the capital that followed went to dividends first, and then, at Bokoni, back into ground that had been written off eighteen months earlier..The results that caught the move. Company results releases of 29 July (Valterra), 28 August (Northam), 1 September (Sibanye-Stillwater), 3 September (Impala) and 4 September 2026 (African Rainbow Minerals). Moz and koz are million and thousand ounces; 3E, 4E and 6E are the platinum-group metal baskets each company reports; AISC is all-in sustaining cost; HEPS is headline earnings per share.The sequence matches gold's: an industry declared terminal at the bottom of a price cycle, shafts closed on a price deck, and a recovery within a single reporting year on a price the deck had not contemplated. The difference is speed. Gold's price multiplied over a quarter of a century from 1999; platinum doubled between April and December of one year, and its obituary and its recovery ran in the same volume of the same trade journals. A price that doubles in nine months can halve in nine, as the basket did between 2022 and 2025, and platinum stood forty-two per cent below its January record on 12 October 2026.The industry cut thousands of jobs and demerged its largest producer at the bottom; within a year Impala returned 17.1 billion rand and Valterra's headline earnings rose 1,633 per cent.EXHIBIT D · THE CROSSOVERV. Why the rarer metal sells for less than half the price of goldThe price history the two metals share is the clearest test of what the gold argument rests on. Platinum was the dearer metal for most of the twentieth century. On annual averages it slipped below gold only briefly in the early 1980s, and on the London benchmark it did not trade below gold on any day from 1996 until the collapse of Lehman Brothers in late 2008. Platinum's London fix peaked at 2,276 dollars an ounce on 4 March 2008, when gold was below a thousand, and on the annual averages the United States Geological Survey publishes, platinum averaged 1,578 dollars that year against gold's 874, one and eight-tenths times the price. The premium had a physical basis. The world mined about 173 tonnes of platinum in 2025 on the World Platinum Investment Council's count and 3,672 tonnes of gold on the World Gold Council's, about twenty-one times as much (about nineteen times on the Geological Survey's rounder figures); and because little gold is consumed, its above-ground stock, which the gold council put at 222,600 tonnes at the end of June 2026, grows every year.The premium did not survive. Gold overtook platinum again in August 2011, and platinum stayed below it into late 2012; on annual averages platinum was again the dearer metal in 2013 and 2014. It fell below gold in early 2015 and has stayed there: the London benchmark has fixed beneath gold on every day since mid-January 2015, a run BullionVault had already called the longest since 1900 by April 2017. The discount widened after the diesel-emissions scandal of September 2015 turned European buyers away from the diesel cars whose catalysts carried platinum. It reached a then-record 1,150 dollars an ounce in February 2024, the ratio of gold to platinum peaked at about 3.6 on 21 and 22 April 2025, and on 12 October 2026 gold at 4,172 dollars was 2.46 times platinum at 1,694, a discount of about 2,480 dollars. Platinum's rally narrowed the ratio without coming close to closing it. On annual averages platinum rose from 961 dollars in 2024 to 1,275 in 2025, and gold from 2,388 to 3,432; on the month-end London fixes of January to September 2026 the averages are 1,923 and 4,538, a ratio of 2.4..Gold and platinum: annual average prices 1991–2024, monthly averages from January 2025, and the close on 12 October 2026. Platinum was the dearer metal for most of the period and on every day from 1996 to late 2008; it has traded below gold every day since mid-January 2015. Sources: USGS Minerals Yearbook and Mineral Commodity Summaries (Engelhard quotations), 1991–2024; monthly averages 2025–2026 from YCharts (platinum) and xauticker.com (gold); 12 October 2026 close: Trading Economics (gold), NYMEX (platinum). My reading of the crossover is that gold has stopped being half money and half commodity. It is now priced almost entirely as money, while platinum is still priced as an industrial input. The World Gold Council's figures for 2025 show the change. Of 5,002 tonnes of demand, investment took 2,175, up eighty-four per cent, and central banks 863, together about three-fifths of the market. Jewellery, the commodity buyer, took 1,542 tonnes, eighteen per cent less than the year before, and the price rose anyway. When the largest fabrication use of a metal shrinks by nearly a fifth in a year of record prices, that use has stopped setting the price, and in the second quarter of 2026 jewellery demand fell again, to 278 tonnes, its lowest since the pandemic. Official buying points the same way: central banks bought more than 1,000 tonnes in each of 2022, 2023 and 2024 and 863 tonnes in 2025, against an average of 473 tonnes a year from 2010 to 2021. The pace has slowed but not stopped: the Council's Gold Demand Trends put first-half 2026 buying at 345 tonnes, and its monthly count of reported purchases, published on 6 October, reached 170 tonnes by the end of August, 39 of them in August, led by China's 20, Poland's 8 and Uzbekistan's 8, with Russia and Jordan the only sellers. China then reported a further 23 tonnes for September, its twenty-third consecutive month of buying and its largest since October 2023, in a month when the price fell more than six per cent, and gold-backed exchange-traded funds took in a record 31 billion dollars in the third quarter, which lifted their holdings to a record 4,256 tonnes. No central bank holds platinum as a reserve asset. Imperial Russia minted platinum roubles from 1828 to 1845, and no state has used the metal as money since. Platinum's demand is the reverse. Of 8,671,000 ounces in 2025, on the platinum council's figures, automotive catalysts took 3,039,000, industry 2,266,000 and jewellery 2,214,000; investment of every kind took 1,152,000, thirteen per cent. Its price is set by car sales, emission rules and chemical plants, and at the margin by investment flows that reverse quickly, as the 389,000 ounces the council expects exchange-traded funds to sell in 2026 show. It explains why platinum doubled in nine months and gave back more than two-fifths in the next eight, while gold's re-rating has held through a year of falling jewellery demand. Scarcity does not set a price. The buyer does. The crossover matters to this comparison in two ways. It limits the platinum case. South Africa's eighty-three per cent of the world's platinum-group metal reserves earns it no monetary premium: the metal is valued as an input, exposed to substitution, to the electric car and to the next cycle, and a producer that plans on a cautious platinum price deck is being prudent about a commodity. It also points back at gold. The gold producers' price decks were set in the same boardrooms by people trained on the same cycles, and in my view they treat gold as if it were still half commodity, a price to be averaged and discounted back toward its long-run mean. For a metal that central banks bought at more than a thousand tonnes a year from 2022 to 2024, that is the wrong model, and a gold shaft closed on a commodity deck at a record price is the error the platinum comparison makes easiest to see. Platinum was the dearer metal on every day from 1996 to 2008; it has been the cheaper one every day since mid-January 2015, and on 12 October 2026 gold cost 2.46 times as much. The world mines about twenty-one times more gold. Scarcity does not set the price. The buyer does. EXHIBIT E · THE SAME DISCOUNT, TESTED AND NOT ASSERTED VI. Who the low-cost producer is The jurisdiction discount is where the gold argument is most often accused of special pleading, so the platinum test has to be strict. No published comparison of enterprise value to earnings or price to net asset value between South African and North American PGM producers was found for 2026, and none is asserted here. What can be tested is the claim underneath the discount, that the South African address is the expensive place to mine, and this year the evidence runs against it. Valterra's all-in sustaining cost in the half-year was 996 dollars per 3E ounce; Impala guides 25,250 to 26,250 rand per 6E ounce for the year ahead. Sibanye-Stillwater's American operation, Stillwater in Montana, the one PGM mine in the jurisdiction the market prices at a premium, mined at an all-in sustaining cost of about 1,350 dollars an ounce in the half-year after a United States production tax credit and about 1,530 before it, ran negative free cash flow of 28 million dollars, produced a third less in 2025 than in 2024, and drew from its chief executive, on 2 September, the sentence that "there may ultimately be no viable basis for the continued operation". About 420 unionised workers at its Stillwater East mine and Columbus metallurgical complex struck the next day, 3 September, and stayed out for five weeks, until a three-year wage agreement announced on 7 October sent them back to work on 9 October. The Bushveld producers multiplied their earnings while the premium jurisdiction lost money on the same metal in the same half-year. The market has re-rated, though not evenly. Over the twelve months to the start of September, Impala's shares gained twenty-six per cent, Northam's thirty-six, Valterra's forty-six, against a basket that gained eighty-five per cent and earnings that multiplied by ten to thirty. Whether that constitutes a discount to a global peer cannot be stated without a multiple. What can be stated is that on 25 August 2026 Northam announced an unsolicited, non-binding approach from a major producer, which the press named as Valterra and which Valterra has not confirmed, and that on 16 September it said no offer had been received and invited credible parties into a formal process, with responses due in December. By 9 October Valterra and Sibanye-Stillwater had confirmed receiving the sale documents. Sibanye's chief executive said it had no interest, Impala had already distanced itself, and Valterra's said its capital was going into its own mines. A low-cost producer is for sale and its neighbours are in no hurry, which is how the gold majors treated the Witwatersrand's dumps before others bought them. The capital that is arriving on the belt is Australian. Southern Palladium, listed in Sydney and Johannesburg, was granted a thirty-year mining right on 7 August 2026 for Bengwenyama on the eastern limb, a 40-million-ounce UG2 and Merensky resource with a maiden reserve of 6.3 million ounces, in which the Bengwenyama community holds thirty per cent. On 13 October 2026 it published an updated pre-feasibility study that puts the project at a post-tax net present value of 1.96 billion dollars and a return of thirty-nine per cent, against 857 million and twenty-six per cent in the study of July 2025, on peak funding of 329 million dollars, for about 200,000 ounces a year in a first stage rising to about 420,000, with about a million tonnes of chrome concentrate a year at full scale. The mine plan is unchanged. What moved was the price deck, a basket of 2,210 dollars an ounce against 1,406, and the chrome, 17 million tonnes over the mine's life against six. A definitive feasibility study is due at the end of the first quarter of 2027, and the box-cut waits on waste and water permits. The majors will not buy from each other at this price; a junior funded from Australia is preparing to build. The evidence supports the discount claim in one form only: the premium jurisdiction is the high-cost one. Montana, the premium jurisdiction, is cash-negative at more than 1,500 dollars an ounce and lost five weeks to a strike. The Bushveld makes the same metal at 996. EXHIBIT F · THE NEW BUYER, SCORED VII. Hydrogen behind its forecast path, and the buyer nobody forecast The fourth step looks for the demand that re-rates a metal from outside its old cycle; for gold it was the central bank. For platinum the industry's answer has been hydrogen: proton-exchange-membrane electrolysers that need iridium at the anode and platinum at the cathode, and fuel cells that need platinum at the cathode. The World Platinum Investment Council's own study of April 2024 put hydrogen demand at about 875,000 ounces by 2030, eleven per cent of the market, up from 40,000 in 2023: 645,000 in fuel cells, 372,000 of it in heavy trucks, and 229,000 in electrolysers. On the same council's figures of 9 September 2026, hydrogen took 71,000 ounces in 2025 and is forecast at 77,000 for 2026, growing "at a more measured pace"; to reach the 2030 figure it would have to grow more than elevenfold in four years. The International Energy Agency's review of June 2026 explains why: electrolyser capacity doubled in 2025 to more than four gigawatts, but nearly three-quarters of the new installations were in China; low-emission hydrogen is almost a million tonnes a year, under one per cent of supply; and of the 27 million tonnes in the 2030 project pipeline, little more than four million are committed. The new buyer exists, but it is slow and mostly Chinese. Valterra's marketing head put the bull case on the record on 1 September: six million ounces of platinum demand if fuel-cell trucks take a fifth of the world's fleet at current loadings, with "thousands" already on Chinese roads. It is a producer's forecast, to be read beside the council's actuals. The buyer that did arrive was in nobody's forecast. Ruthenium, a metal the industry treats as a by-product, averaged 690 dollars an ounce in 2025 and trades at 1,850 in the second week of October 2026, more than two and a half times as much, on demand from hard-disk data storage; iridium, used in the crucibles that grow single crystals for electronics, stood at 8,400 from a 2025 average of 4,400; Johnson Matthey's May 2026 report put ruthenium's 2025 deficit at 293,000 ounces and described the metal as in "deep structural deficit", with iridium edging into deficit. The artificial-intelligence build-out, which nobody on the platinum belt planned for, is buying the two metals of which South Africa is, in effect, the only producer. Investment is the buyer that has turned. Bars and coins took 402,000 ounces in 2025 and are forecast at 313,000 in 2026; exchange-traded funds, which bought 200,000 ounces in 2025, are forecast to sell 389,000; and in China, where a value-added-tax change took effect last year, jewellery fabrication fell seventy-six per cent and bar demand seventy-four per cent in the second quarter of 2026. China's Guangzhou exchange launched platinum and palladium futures in November 2025 and opened them to foreign investors on 28 September 2026, and its stocks now enter the council's balance. Both new buyers are industrial. Neither is the kind of buyer that re-rated gold, which is why platinum's gains have proved easier to reverse. Hydrogen took 71,000 ounces in 2025 against a council forecast of 875,000 by 2030. Ruthenium, a by-product nobody forecast, went from a 2025 average of 690 dollars to 1,850 on the data-centre build-out. EXHIBIT G · THE METALS NOBODY ELSE HAS VIII. Ruthenium, iridium and a monopoly the country has never priced The endowment claim is strongest, and least discussed, in the minor platinum-group metals. Platinum has Russia and Zimbabwe as second and third producers; palladium has Russia first. Rhodium, iridium and ruthenium have, to a first approximation, only the Bushveld: they are recovered as by-products of the platinum and palladium circuit in proportions the reefs dictate, and the UG2 reef in particular carries them in tenors no other deposit on earth matches. The world's supply of ruthenium is a few tens of tonnes a year and of iridium less than ten, and almost all of it leaves through the smelters and refineries of the Bushveld's four producers. The country has never priced that monopoly, because the metals were too small to matter, and the buyers have now changed. Ruthenium's use in the magnetic recording layers of hard drives, and iridium's in the crucibles that grow the single crystals the semiconductor and lighting industries need, have put both in the path of the data-centre build-out, which is what the prices above reflect. Rhodium, the autocatalyst's metal, is 9,200. A basket that the market prices as "platinum" is, in its minor metals, a set of near-monopolies with a new industrial buyer, and Valterra's realised basket of about 2,800 dollars in the first half of 2026 is where that shows. The Witwatersrand's main by-product, uranium, is still contested; the Bushveld is paid for four. Rhodium, iridium and ruthenium are, to a first approximation, Bushveld metals, and a basket the market calls “platinum” is in part a set of near-monopolies with a new buyer. EXHIBIT H · THE LOOP THAT ALREADY RUNS IX. Recycling, chrome tailings and a second life already in the accounts On recovered metal the platinum belt is ahead of the gold belt. Autocatalyst recycling returned 1,232,000 ounces of platinum in 2025 and is forecast at 1,422,000 in 2026; with jewellery and industrial scrap, secondary supply is forecast at 1,802,000 ounces, a quarter of total supply, as older, higher-loaded catalysts reach the scrapyards. The tailings loop the gold argument proposes for the Witwatersrand is also running here, and it is listed. Sylvania Platinum treats chrome tailings, the residue of chrome mines on the UG2 reef, which carry platinum-group metals the chrome operators never wanted. On 15 September 2026 it reported a record 95,885 4E ounces for the year to June, above its own guidance, on revenue that more than doubled to 226 million dollars, with no debt and 67 million dollars in cash, and guided 85,000 to 95,000 ounces for the year ahead. Its Thaba joint venture was commissioned in August 2025, and a new geological model has since cut its expected grades, putting the mine plan under review. Twelve chrome-recovery plants work UG2 residue across Sibanye-Stillwater's footprint, six of them its own and five Glencore's. In the six months to June 2026 the company's South African operations produced 951,000 tonnes of chrome for an operating profit of R1.1 billion, and the chrome credit cut their all-in sustaining cost by R7,480 a 4E ounce. Output was 18 per cent lower because the Marikana bulk tailings chrome plant closed in the second half of 2025, when the re-mining of its tailings dam was complete: a tailings loop run to the end. South Africa is forecast to produce about 26 million tonnes of chrome this year, sixty-one per cent of the world, and the company expects its own volumes, one million tonnes in 2016, to run at 2.3 million a year until 2033. A co-product that pays alongside the primary metal, a tailings retreater without a shaft and a recycled quarter of the market are all operating on the platinum belt, in audited accounts, and have been for years. Two lessons carry over to gold. The co-product can be the larger business: on the UG2, the chrome the platinum miner discards is a commodity in its own right, and South Africa supplies sixty-one per cent of the world's, and the tailings retreater's economics are carried by the chrome as much as the platinum. It is the uranium case at the gold dumps, made in a metal with a functioning market. And the loop needs a fiscal signal: no South African statute rewards the recovered ounce, of platinum or of gold, and a country that holds eighty-three per cent of the primary reserves without a recycling incentive leaves the second life of its own metal to whoever offers one. The platinum belt has also gone further than the gold belt on the power that makes a second life cheap. Northam is commissioning 80 megawatts of solar at Zondereinde and is building 140 megawatts of wind and 255 of solar, 475 megawatts in all, to save about a billion rand a year against a tariff that rose 12.74 per cent in April 2025 and 8.76 per cent in April 2026, with 8.83 per cent to follow in 2027 after the regulator corrected a 54.7-billion-rand error in its own decision; the largest single input to a narrow-reef mine after labour is electricity, and in my view the producers best placed after the 2024 trough are those that started generating their own. A retreatment plant on chrome tailings run on owned power is a different business from one run on the grid, as it would be at the gold dumps, and the platinum belt has the accounts to show it. .The platinum ledger. World Platinum Investment Council, Platinum Quarterly Q2 2026, published 9 September 2026. The 2026 forecast turned from a 297,000-ounce deficit in May to a 265,000-ounce surplus, almost entirely on a 601,000-ounce cut in expected investment demand; the next edition is due in November 2026. Sylvania made a record 95,885 ounces from chrome tailings without a shaft; twelve chrome plants work UG2 residue on Sibanye-Stillwater's footprint; a quarter of platinum supply is recycled. THE BEAR, MET X. What the sceptic rightly takes The bear's first blow is the autocatalyst. More than a third of platinum demand and most of palladium's rides in exhaust systems, battery-electric vehicles have none, and the council's own forecast has automotive platinum falling from 3,039,000 ounces to 2,904,000 in a year; Johnson Matthey has palladium and rhodium moving to small surpluses in 2026, and the substitution of platinum for palladium, a support for the platinum price, has a sell-by date. Conceded, and it is why hydrogen is scored here on actuals: the old buyer is fading, the promised buyer is late, and the market between them was carried by industrial demand, investment and a deficit. The deficit has now gone. The first half of 2026 ran a surplus of 548,000 ounces, and on 9 September the council turned its whole-year forecast from a 297,000-ounce deficit to a 265,000-ounce surplus, on a 601,000-ounce cut in expected investment. The deficit was an investment-flow story as much as a mine story, and the flow has reversed. The second blow is that the discount cannot be proved. No multiple was found and none is printed, and the bear is entitled to say that "the premium jurisdiction was on strike" is an anecdote about one Montana mine. Conceded; the claim is the cost curve, not the valuation, and the cost curve is audited. The third is the price: down about seventeen per cent in 2026, forty-two per cent off its record, in a market now forecast in surplus, with nothing to stop it falling back toward 895. Conceded as the cycle, and the reason the 2024 obituary reads as a warning about 2027 as much as a proof about 2025. The fourth is policy. The Waterberg project, a large shallow deposit on the northern limb, was stalled in September 2026 at what the largest shareholder in its majority owner, Platinum Group Metals, called "the mountain pass … the mountain pass being smelting", because the state will not permit concentrate to be smelted abroad and the integrated producers have not offered commercially viable terms to treat it at home; an updated feasibility study is due in November, and Impala's chief executive said on 15 September that a smaller Waterberg has a "very likely future". Beneficiation policy is a supply constraint the geology did not impose, and the lock is a clause. The fifth is that supply is falling for reasons that are not price (tariffs up more than 900 per cent, a labour-intensive method, ageing shafts) and the falling curve may be geology after all. Partly conceded: the Bushveld's shallow reefs and eighty-three per cent of reserves make geology the least likely of the explanations, and the supply curve faces the same test the Witwatersrand's does: a price the deck did not contemplate, which arrived in 2025 and has since given back more than three-fifths of its gain. The sixth is the crossover itself: a metal that has sold below gold for eleven years may have been re-rated for good, and the eighty-three per cent may be worth less than the geology implies. Conceded; it limits the platinum comparison and, for the reasons given above, strengthens the gold case. The bear takes the autocatalyst, the multiple, the cycle, this year's deficit and platinum's old premium. The endowment, the burial, the cost curve and the loop remain. The bear takes the autocatalyst, the multiple, the cycle, this year's deficit and platinum's old premium over gold. The endowment, the burial, the cost curve and the loop remain. THE RETURN TO GOLD XI. What platinum teaches the gold argument Five of the six steps hold on platinum: a first-rank endowment written as a sunset; an industry declared terminal at 895 dollars that then multiplied its earnings; a premium jurisdiction losing money on the metal the Bushveld makes at a profit; a new buyer late from one direction and early from another; and a loop that already recycles a quarter of supply. The method is not peculiar to one basin. What differs is the price, and the difference favours gold. A metal priced by central banks and investors has held a re-rating that a metal priced by carmakers and chemical plants gave back within months. Platinum's other lessons for gold are practical. The gold argument calls for a co-product circuit at the dumps against an operator's public refusal; the platinum belt runs twelve on chrome, a commodity of which the country supplies sixty-one per cent of the world. It calls for a certified, recycled ounce without a fiscal incentive; the platinum loop shows that recycled metal follows the fiscal incentive, and South Africa offers none. And platinum shows how fast sentiment can turn on an industrial metal: a recovery that took gold a generation took platinum a year, and more than two-fifths of the price went in the eight months that followed. The Witwatersrand's latest obituary was written on a price deck, by serious people, in the same years and the same journals. The platinum record says the reversal can come quickly, and for gold, with a monetary buyer beneath it, a reversal is more likely to hold. The platinum belt already runs the co-product circuit the gold belt is still arguing about. Its price shows what gold's does not: an industrial metal's re-rating can be undone in months. THE VERDICT XII. Two provinces, one country, one method South Africa holds the two most concentrated precious-metal endowments on earth within two hours' drive of each other, mined by the descendants of the same houses on the same grid, and it has spent the last decade writing obituaries for both. September 2026 Impala returned 17.1 billion rand to its shareholders; on 29 July Valterra reported headline earnings up 1,633 per cent; on 28 August Northam declared a record; on 15 September a chrome-tailings retreater reported a record year; and on 2 September the owner of the American mine that was supposed to be the safe jurisdiction warned it might have to close, and its workers struck the next day. Every one of those companies is mining a metal the market said was finished eighteen months ago, from reefs shallower than the Witwatersrand's, in a province holding eighty-three per cent of the world's platinum-group reserves. The country did not run out of platinum. It ran out of patience with a price, and the price came back before the patience did. The two residues carry the same lesson: the metal is where the geology put it, and the market prices it where the last cycle left it. The Bushveld is the Witwatersrand with a shallower reef, a faster clock and a working co-product, and an argument built for one basin works on the second. It works with one amendment, which the crossover supplies. The Bushveld's metals are priced as industrial inputs and will be written off again the next time the basket falls. Gold is priced as money, and my view is that the gold industry's error has been to plan it as though it were still platinum. The comparison also carries a caution. The platinum belt's loop has no South African fiscal incentive behind it; its largest new project is stalled at a smelter by a beneficiation rule; its labour model has changed little since Marikana; and its recovery rests on a price that halved in three years, doubled in one and has given back more than two-fifths in eight months, in a market now forecast in surplus. Sylvania and the chrome plants give the province an income that does not depend on the stope, but they are small. The Witwatersrand has a listed retreater with a fifty-eight per cent margin, six million ounces of surface reserves and an eight-hundred-million-tonne lined facility under construction, so the two provinces have things to teach each other in both directions. South Africa holds both, and has rarely read them together. Read together, they leave one question open. If the buyer sets the price, why does the country that owns eighty-three per cent of the ground accept the price it is shown? .THE BUSHVELD HOLDS EIGHTY-THREE PER CENT OF THE WORLD'S PLATINUM-GROUP RESERVES, WAS DECLARED FINISHED AT 895 DOLLARS AND WITHIN EIGHTEEN MONTHS MULTIPLIED ITS EARNINGS. PLATINUM WAS DEARER THAN GOLD ON EVERY DAY FROM 1996 TO 2008. SINCE JANUARY 2015 IT HAS BEEN CHEAPER, AND ON 12 OCTOBER 2026 GOLD COST 2.46 TIMES AS MUCH. SCARCITY DOES NOT SET A PRICE; THE BUYER DOES. .*Dr Duarte F da Silva • Capital Markets Strategist • Managing Director, Northbound Processing • Germiston, September 2026.A NOTE ON THE FIGURES Every figure in this paper was checked against its source in the week of 7 September 2026, re-checked on 23 September 2026, and checked again line by line on 25 September 2026, when prices were moved to 24 September 2026 and the corrections listed in the library index were made; Section V, its chart and the third banner figure were added on 25 September 2026. The paper was checked again on 8 October 2026, when prices were moved to 7 October 2026 and the company and market statuses were brought to that date, and on 10 October 2026, when prices were moved to the 9 October close, the Section V chart was redrawn at monthly resolution from January 2025 and carried to the 9 October price, the company results in Section IV were set out in a table, and the central-bank and Statistics South Africa figures were brought to their latest releases of 6 October and 10 September 2026. It was checked again on 13 October 2026, the day before publication, when prices were moved to the 12 October close, the chart's labels were carried to that date, and four reports published since the last check were added: Southern Palladium's updated study of 13 October, PwC's SA Mine 2026 of 6 October, China's September gold purchase and the World Gold Council's third-quarter count of exchange-traded fund flows, both of 7 October. The sources follow. Reserves and production: PGM reserves 76,000,000 kg world, South Africa 63,000,000 kg (82.9%), Russia 11,000,000, Zimbabwe 1,300,000, United States 590,000, Canada 310,000; platinum mine production 2025 estimated 170,000 kg world, South Africa 120,000 (70.6%), Russia 20,000, Zimbabwe 18,000, Canada 5,000, United States 1,800; palladium Russia 84,000, South Africa 70,000, world 190,000 (US Geological Survey, Mineral Commodity Summaries 2026, January 2026). Bushveld ~2.05 billion years, Merensky Reef 30–90 cm (other sources give up to 150 cm), found by Hans Merensky in September 1924 (standard references; Johnson Matthey Technology Review 43(4)); ~75% of world platinum and ~50% of palladium resources. South African PGM output −4.1% in 2025 (Minerals Council media statement, 9 February 2026); electricity tariffs up more than 900% since 2007 (Minister of Electricity, 2026). Witwatersrand about 70% of world gold output at the 1970 peak (LBMA). Eskom tariffs: +12.74% (April 2025); NERSA's MYPD6 path of 5.36% and 6.19% was revised after a R54.7bn regulatory-asset-base error to +8.76% (April 2026) and +8.83% (2027/28) (Eskom, 16 March 2026; Engineering News, 10 March 2026). Prices, 12 October 2026 close: platinum US$1,693.50 (NYMEX; Trading Economics; bullion-rates.com gives US$1,693.47), palladium US$1,146, rhodium US$9,200 (Trading Economics); iridium US$8,400 and ruthenium US$1,850 (Umicore prices, 9 October 2026); gold US$4,171.88 (Trading Economics; bullion-rates.com gives US$4,171.71); gold/platinum ratio 2.46 and discount ~US$2,480 (author's arithmetic); platinum about −17% in 2026 against a year-end of US$2,033 (Fortune, 31 December 2025), about +4% over twelve months, palladium about −27% and rhodium about +26% over twelve months, platinum near a ten-week low and palladium near its lowest since September 2025 (Trading Economics, 13 October 2026); 42% below the record, and (2,924 − 1,694) ÷ (2,924 − 895) = 61% of the gain given back (author's arithmetic); on 9 October 2026 the figures were US$1,695 and 2.47, on 7 October US$1,640 and 2.51, on 24 September US$1,753 and 2.44; 2025 closing low US$895.03 on 8 April 2025 (bullion-rates.com; intraday lows were lower); then-record above US$2,450 on 26 December 2025 (Kitco, US$2,454 intraday); record US$2,923.70 on NYMEX futures, 26 January 2026 (metalcharts); palladium record US$3,440.76, March 2022; 2025 averages iridium US$4,400 and ruthenium US$690 (USGS). Market balance: World Platinum Investment Council, Platinum Quarterly Q2 2026, 9 September 2026 (2025 deficit revised to 1,440 koz; 2026 forecast turned from a 297 koz deficit (Q1 edition, 18 May 2026) to a 265 koz surplus on a 601 koz cut in expected investment; H1 2026 surplus 548 koz (Q1 +304, Q2 +244), H2 deficit −283 koz expected; total demand 8,671 → 7,089 koz; automotive 3,039 → 2,904; jewellery 2,214 → 1,883; industrial 2,266 → 2,385 incl. hydrogen 71 → 77 koz; bar and coin 402 → 313; ETFs +200 → −389; exchange stocks +384 → −112 (GFEX stocks now included); recycling 1,669 → 1,802 koz (autocatalyst 1,232 → 1,422); above-ground stocks 1,745 koz (2.4 months) → 2,010 koz (3.4 months); Chinese jewellery fabrication −76% and bar demand −74% year on year in Q2 2026; 2027 forecasts due in November 2026); World Gold Council, Central Bank Gold Statistics, 6 October 2026 (August net purchases 39 t; 170 t reported year to date; China 20 t, Poland 8 t, Uzbekistan 8 t, Kazakhstan 7 t, Turkey 3 t, Czech National Bank 2 t; Russia −6 t, Jordan −3 t) and Gold Demand Trends Q2 2026 (H1 central-bank buying 345 t; Q2 jewellery 278 t); People's Bank of China +740,000 oz, about 23 t, in September 2026, its twenty-third consecutive month and its largest addition since October 2023, in a month when gold fell more than 6% (Yahoo Finance, 7 October 2026); gold exchange-traded funds: record inflows of US$31bn in Q3 2026, holdings +67 t in September to a record 4,256 t (World Gold Council, Gold ETF holdings and flows, 7 October 2026); Statistics South Africa, Mining: production and sales, July 2026 (P2041, 10 September 2026: total production −7.5% y/y, PGMs −13.5% y/y; August release due 13 October 2026 and not seen when this paper was closed; Absa on the Rustenburg safety stoppage, BusinessDay, 11 October 2026); PwC, SA Mine 2026, 6 October 2026 (PGM reserves 261 → 298 million 4E oz and average mine life 38 → 43 years, 2023 to 2026; PGM companies' revenue +59% to R429bn; average dollar platinum price +80% in the twelve months to June 2026; PGM production below 2019 levels; Miningmx and IT-Online, 6 October 2026); WPIC, Platinum Essentials, 30 April 2024 (~875 koz by 2030 ≈ 11% of demand, up from 40 koz in 2023; fuel cells 645 koz incl. 372 koz heavy-duty; electrolysers 229 koz); GFEX platinum and palladium futures launched 27 November 2025 (Reuters), open to foreign investors from 28 September 2026 (Mining Weekly, 25 September 2026); Johnson Matthey PGM Market Report, 14 May 2026 (platinum deficits 951 koz in 2025 and 371 koz in 2026 on JM's own count; palladium +214 koz and rhodium +15 koz surpluses in 2026; ruthenium 2025 deficit 293 koz and in "deep structural deficit"; iridium edging into deficit; South African supply 3,957 → 3,873 koz). IEA Global Hydrogen Review 2026, 18 June 2026 (electrolyser capacity doubled to >4 GW in 2025, nearly three-quarters of new installations in China; almost 1 Mt low-emission hydrogen; 2030 pipeline 27 Mt, of which about 4.2 Mt committed on the executive summary's figures). Valterra: demerger effective 31 May 2025; Anglo American sold its entire remaining c.19.9% (c.52.2 million shares at R845, R44.1bn, about US$2.5bn) by accelerated bookbuild on 4 September 2025, settled 9 September 2025 (Anglo American press release, 4 September 2025); H1 2026 (29 July 2026) — refined 1.7 Moz (+25%), basket ~US$2,800/oz (+85%) on the results call (Miningmx reports US$2,710, +80%), AISC US$996/3E oz, revenue R82bn, adjusted EBITDA R33.4bn, HEPS R82.02 (+1,633%), interim dividend R15bn, net cash R23.7bn, 2026 AISC guidance ~US$1,050; Hilton Ingram on 6 Moz of truck-fleet demand (Mining Weekly, 1 September 2026). Impala FY2026 (3 September 2026): 6E 3.50 Moz, revenue R135.1bn, EBITDA R43.6bn, headline earnings R22.9bn (31×), free cash flow R22bn, total dividend 1,855c/share including a 955c additional ordinary dividend, R17.1bn returned to shareholders, adjusted net cash R22bn, FY2027 guidance 3.30–3.50 Moz at R25,250–26,250/6E oz (Implats results; Modern Mining). Sibanye-Stillwater H1 2026 (1 September 2026): SA PGM 831,307 4E oz including purchased concentrate (789,647 oz own production) and adjusted EBITDA R19.2bn; group adjusted EBITDA R31.8bn (+111%); US PGM 137,930 2E oz, AISC US$1,347/2E oz including a US$181/oz Section 45X credit, US$1,528 before it, notional free cash flow −US$28m; Stillwater 284,069 oz in 2025 vs 425,842 in 2024; "there may ultimately be no viable basis for the continued operation" (Richard Stewart, strike-notice SENS, 2 September 2026; Mining Weekly, 2 September 2026); USW strike of ~420 workers at Stillwater East and the Columbus metallurgical complex (East Boulder not on strike) from 07:00 on 3 September 2026, ended by three-year agreements, East Boulder ratified 30 September and Stillwater East and Columbus announced 7 October, work resuming 9 October (Mining Weekly, 7 October 2026); Sibanye acquired Lonmin, including the Marikana operations, by all-share scheme completed 7 June 2019 (Sibanye-Stillwater transactions page). Kwezi shaft restructuring, 1,114 jobs at risk (781 employees, 333 contractors), losses of R208m (2024) and R91m (2025), licence-area limits (Mining Weekly, 8 September 2026); group headline earnings per share +216% (BusinessDay, 2 September 2026). Northam FY2026 (28 August 2026): 939 koz 4E (938,754 oz own operations), revenue ~R54bn, EBITDA R16.6bn, HEPS 3,044c (+699%), dividends R17/share, net cash R2.7bn; 475 MW renewables: 80 MW Zondereinde solar in commissioning, 140 MW Karreebosch wind (2027) and 255 MW Thakadu solar (mid-2027) under construction (Mining Weekly, 28 August 2026); South African output ~3.9 Moz, 26% below the 2006 peak of 5.3 Moz (Kitco, 28 August 2026); SA mine supply +1% forecast for 2026 (WPIC); approach announced 25 August 2026, named by Bloomberg (28 August) as Valterra and unconfirmed; formal process, invitation letters sent, no offer received, 16 September 2026; expressions of interest due 1 December 2026 (Miningmx, 25 August 2026); Valterra and Sibanye-Stillwater confirmed receiving the information memorandum, Richard Stewart "No interest", Craig Miller "our focus is really around extracting the value from our portfolio", Impala having distanced itself earlier (Miningmx, 7 October 2026). Southern Palladium (ASX and JSE: SPD): Bengwenyama mining right granted by the DMPR on 7 August 2026 for 30 years; 70% held through Miracle Upon Miracle Investments, 30% by the Bengwenyama community; resource ~40.25 Moz PGM plus gold, maiden reserve 6.29 Moz; updated PFS of 13 October 2026: post-tax NPV US$1,961m at 8% real (optimised PFS of 10 July 2025: US$857m), IRR 39.1% (26.4%), peak funding US$329m (US$279m; R5.47bn against R5.46bn, the dollar rise reflecting a stronger rand forecast), 6E basket US$2,210/oz (US$1,406), AISC US$943/6E oz, about 200 koz 6E a year rising to about 420 koz, chrome concentrate about 450 kt a year rising to about 1 Mt at 85.6% chromite recovery, 17.2 Mt over the life of mine (6.0 Mt), life-of-mine plan of 41.9 Mt at 6.54 g/t 6E and ore reserve unchanged; DFS due end of Q1 2027; box-cut and decline subject to waste-management and water-use permits (company announcement, 13 October 2026, as reported by Kalkine and Discovery Alert; Rio Times, August 2026). African Rainbow Minerals: Bokoni approved 23 July 2026 (capex R15.2bn, IRR 28%; IOL, 23 July 2026), NPV R5.9bn restated in FY2026 results, 4 September 2026. Twelve-month share performance Implats +26%, Northam +36%, Valterra +46% (Miningmx, 3 September 2026). Sylvania Platinum FY2026 (15 September 2026): record 95,885 4E oz (+18%), net revenue US$226.3m (+117%), EBITDA US$114.2m, net profit US$66.4m, cash US$67.2m, no debt, FY2027 guidance 85–95 koz; Thaba JV commissioned August 2025 with first products dispatched after September 2025, grades revised down and mine plan under review (Sylvania 2026 Annual Report, 15 September 2026). Sibanye chrome: twelve chrome-recovery plants on its footprint, six its own and five Glencore's, H1 2026 (six months to 30 June, reported 1 September 2026): chrome production 951 kt (−18%), sales 888 kt (−17%), chrome operating profit R1.1bn (US$65m, −18%), by-product credit reducing AISC by R7,480/4E oz, decline mainly from the closure of the Marikana BTT chrome plant in H2 2025 on completion of the ETD1 tailings re-mining (Sibanye-Stillwater H1 2026 results, Form 6-K, 1 September 2026); chrome revenue is not disclosed separately for H1 2026, so no revenue share is given; volumes 1.0 Mt (2016) to 2.3 Mt a year until 2033, South Africa ~26 Mt forecast for 2026 = 61% of world chrome (Mining Weekly, 16 July 2026). Impala developed by Union Corporation, production from 1969; Gencor formed in 1980 by the merger of Union Corporation and General Mining (Implats history). Jubilee Metals' sale of its South African chrome and PGM business to One Chrome completed 2 January 2026. Waterberg stalled on smelting (Johnny Copelyn of HCI, the largest shareholder in Platinum Group Metals Ltd, which owns ~50% of the Waterberg JV; Miningmx, 2 September 2026; updated feasibility study due November 2026; Impala's chief executive on a smaller project with a "very likely future", Miningmx, 15 September 2026). Gold and platinum (Section V and chart): annual averages 1991–2025 from USGS Minerals Yearbook (Gold 1994, 1996, 2001; Platinum-Group Metals 1995, 1996, 2001, 2021) and Mineral Commodity Summaries (1996, 2006, 2010, 2014, 2018, 2022, 2026), Engelhard quotations, 1991–2024, and LBMA annual averages for 2025 (gold US$3,431.54, platinum US$1,274.73; the USGS estimates were US$3,300 and US$1,200); monthly averages January 2025 to September 2026 for the chart: platinum from YCharts (2025: 948, 978, 980, 958, 1,021, 1,240, 1,391, 1,334, 1,430, 1,618, 1,564, 1,898; 2026: 2,414, 2,141, 2,054, 2,017, 1,991, 1,738, 1,616, 1,782, 1,786); gold from xauticker.com (2025: 2,710, 2,895, 2,983, 3,218, 3,309, 3,353, 3,340, 3,368, 3,668, 4,058, 4,087, 4,309; 2026: 4,753, 5,020, 4,856, 4,721, 4,587, 4,228, 4,073, with August and September 2026 taken as about 4,260 from the month opens and closes on goldtickers.com); gold record US$5,589 on 28 January 2026; platinum London fix record US$2,276 on 4 March 2008; ~US$2,250 with gold ~US$986 on 3 March 2008 (CNBC, 3 March 2008); platinum at a premium through the twentieth century with two early-1980s annual-average exceptions, no day below gold from 1996 until late 2008, benchmark below gold every day since mid-January 2015, 'longest run since 1900' (BullionVault, 10 April 2017), then-record discount US$1,150 in February 2024 and the diesel-scandal explanation (BullionVault, 9 February 2024); gold first above platinum on 24 August 2011 (Numismatic News); USGS annual averages show platinum above gold in 2013 and 2014; USGS's Engelhard series records platinum below gold from 6 February 2015 (Mineral Industry Surveys, PGM, March 2015); gold/platinum peak about 3.60 on 21 and 22 April 2025 on daily closes (bullion-rates.com; GuruFocus gives 3.45 on 16 April); gold traded above platinum for some days in December 2008 (BullionVault). Gold 2025: mine production 3,671.6 t; demand 5,002.3 t, investment 2,175.3 t (+84%), central banks 863.3 t, jewellery 1,542.3 t (−18%), technology 322.8 t (World Gold Council, Gold Demand Trends Full Year 2025, 29 January 2026); central-bank purchases above 1,000 t in each of 2022, 2023 and 2024 (2024 now 1,092.4 t), 863.3 t in 2025, 2010–21 average 473 t; H1 2026 345 t; Q2 2026 jewellery 278 t, lowest since the pandemic; H1 2026 total demand 2,522 t (+2%) (WGC, Gold Demand Trends Q2 2026); above-ground stock 222,600 t at end-June 2026 (WGC). On USGS figures (gold 3,300 t, platinum 170 t) the mining ratio is about 19×. Platinum mine supply 2025 5,561 koz ≈ 173 t and demand shares (WPIC, Platinum Quarterly Q2 2026); platinum not held in official reserves (ABN AMRO's Georgette Boele, FXStreet, 30 August 2019); Russian platinum coinage 1828–1845 (standard numismatic references). The reading of gold as priced 'almost entirely as money' and of the gold producers' price decks as commodity decks is the author's view. Restructuring job numbers for 2023–24 are described qualitatively; sector employment estimates were not verified for this paper. The six-step method is the author's summary of the gold argument's approach; no valuation multiple is asserted. Structure: Paper 31 Part I; it stands alone and relies on no other paper. Northbound Processing • Germiston, October 2026. The Gold Papers — Paper 31, Part I. This article does not constitute investment advice. Platinum market figures are the World Platinum Investment Council's and Johnson Matthey's as cited and use the Council's Q2 2026 report of 9 September 2026, which the Council will next update in November 2026; company figures are from the results cited; no valuation discount is asserted beyond the audited cost comparison. The framing of the Bushveld as the "corollary" of the gold thesis and of the six-step method is the author's own.

Dr Duarte da Silva: The Platinum corollary

Related Stories

Heystek: The tide’s gone out on the JSE – and SA Inc forgot its swimming trunks

Duarte da Silva: Gold loses its anonymity, and SA's coal-fired ounce should worry

Duarte da Silva: SA's richest goldmine is the bin — and almost nobody's mining it

Duarte da Silva: The country we declared explored