Investing in Gold Miners and Royalty Companies

In the three months to June 30, 2026, gold fell 16%, its worst quarter since 2013, and the VanEck Gold Miners ETF fell 21%. In those same three months Agnico Eagle, the best-run of the large producers, mined 855,816 ounces at an all-in sustaining cost of $1,459, sold them at an average of $4,483, and generated $1.34 billion of free cash flow, the most in its history. It returned a record $625 million to shareholders in a quarter in which the sector ETF lost a fifth of its value. That is the asset class in one quarter: the business made more money than it ever had, and the equity behaved like a leveraged bet on the metal’s next move, because that is what it is.

Two and a half years earlier, on November 28, 2023, Panama’s Supreme Court ruled the contract for Cobre Panamá, one of the largest copper mines in the world and the source of a gold-and-silver stream Franco-Nevada had paid roughly $1.36 billion to own, unconstitutional. The mine stopped within days. Franco-Nevada, a company with no trucks, no pits and no unions, the model built to have no operating risk, wrote off $1.17 billion on a single asset in a country it did not operate in. By mid-2026 the mine was processing stockpiles under a government permit and Panama had yet to decide its future.

Those two facts are the guide. Gold-mining equities are operating leverage on the gold price, and the leverage runs both ways on a schedule the gold price does not keep; the royalty model moves the risk from the pit to the parliament rather than removing it. What follows is how each business makes money, how to read the cost number that decides whether it does, what the long record shows, what can end you, and what the tax code does for you when you own the mine instead of the metal. The metals themselves, their demand, supply and wrappers, belong to the hub’s flagship, Investing in Precious Metals, and its companions Investing in Gold, Investing in Silver and Investing in Platinum and Palladium; this guide does not repeat that ground.

Everything about how these two businesses behave follows from what each one sells. A miner sells ounces it has dug up, at whatever the market pays on the day, after spending a fixed and rising amount to dig them. A royalty company sells a contractual slice of someone else’s ounces, having spent its money years earlier and once.

Start with the miner. Revenue is ounces times the gold price. Cost is set by rock, not by price: grade, waste moved, recovery, diesel, power, labour, government royalties, and the capital needed to keep the mine its current size. Because the cost is fixed in the short run and the price is not, the margin moves far more than the price.

On September 8, 2026, with gold at $4,443.90 on our tape and Agnico Eagle’s second-quarter all-in sustaining cost of $1,459, the margin was $2,985 an ounce. A 10% rise in gold takes that margin up 14.9%; a 10% fall takes it down the same. At Barrick’s $1,866 cost the same move is 17.2% each way. That is the leverage in a good year. In 2015, with gold near $1,100 and costs near $1,000, a 10% move in gold was a move of more than 100% in the margin, which is why the sector fell 80% while the metal fell 45%.

Now the royalty company. A royalty is a right to a percentage of a mine’s revenue, usually a net smelter return of 1% to 5%, for the life of the mine including any expansion or discovery on the ground it covers. A stream is a right to buy a fixed share of a mine’s gold or silver at a fixed, low price, often a few hundred dollars an ounce, in exchange for a large upfront payment the miner uses to build or expand. Neither holder pays for the trucks, the strip, the cost inflation or the sustaining capital; both hold the price upside and the exploration upside on land they did not have to buy.

Wheaton Precious Metals’ cash cost in the second quarter of 2026 was $568 per gold-equivalent ounce against a cash operating margin of $3,875, and OR Royalties reported a cash margin of 96.8% of revenue. No miner has ever printed that number, and no royalty company will ever print a cost overrun.

What the two share is that they are companies, not metal: a balance sheet, a management that can dilute you, a dividend it can cut, a share price set by equity investors’ appetite for gold exposure rather than by gold demand, and a claim on ounces still in the ground in a country with a government. And each is taxed by the United States as an equity, the one respect in which Washington treats the derivative of gold more kindly than gold itself.

A reader who cannot read a reserve statement cannot read a miner, so the vocabulary comes first. The asset is an orebody. Its size is stated as resources, the tonnes and grade geologists believe are there at three levels of confidence, and reserves, the subset a mine plan says can be extracted at a profit at an assumed gold price. That assumed price is the first number to look for, because reserves are a function of the price the company plugs in.

Raise it and low-grade rock at the edge of the pit becomes ore, so reserves grow with no drilling at all. Cut it and the same rock returns to being waste, which is what happened across the industry in 2013, when the majors re-ran their plans at a lower price and wrote down tens of billions of dollars of assets bought at a higher one; Barrick’s Pascua-Lama charge alone was $5.1 billion, inside an $8.7 billion second-quarter impairment.

Grade is the gold content of the ore, in grams per tonne; strip ratio is the waste moved per tonne of ore, and it rises as pits deepen; recovery is the share of the gold the plant captures, typically 85% to 95%. Tonnes times grade times recovery is ounces, and most large mines carry ten to twenty years of them, replaced each year by drilling, by acquisition, or not at all.

The capital comes in two kinds. Sustaining capital keeps the mine producing at its current rate: replacing trucks, raising the tailings dam, developing the next underground level. Growth capital builds something new. The distinction matters because the industry’s headline cost measure includes the first and excludes the second, and because the second is where the money goes when the price is high. Barrick’s 2026 attributable capital guidance was $3.8 to $4.2 billion, cut from $4.0 to $4.45 billion in August 2026 mostly because spending at Reko Diq, its copper-gold development in Pakistan, was pushed out; Kinross’s was $1.5 billion; Newmont called 2026 its production low point before growth resumes in 2027. Each of those sentences is a promise to spend the margin.

Gold is rarely alone in the rock. Most large mines produce copper, silver, lead or zinc alongside it, and companies report gold costs either co-product, allocating costs across all metals, or by-product, subtracting the other metals’ revenue from the gold cost. Newmont’s $1,621 for the second quarter of 2026 is a by-product number, so its 17,000 tonnes of copper and 7 million ounces of silver in the quarter are already inside it. Gold Fields’ Salares Norte in Chile reported an all-in sustaining cost of $269 per ounce for the first half of 2026, net of by-products, on 337,000 ounces; that is a convention, not a typo, and the same mine on a co-product basis would show a number several times higher. Check the convention before you compare two companies.

One figure decides whether a miner is a business or a liquidation, and it is the cost number. Before 2013 the industry reported cash costs, the direct cost of mining and processing an ounce, and a generation of investors learned in 2011 to 2013 that a company reporting $600 cash costs could lose money at $1,600 gold. The World Gold Council published its guidance note on all-in sustaining costs in June 2013, updated in 2018, and nearly every listed producer has reported on it since.

All-in sustaining cost, or AISC, adds to cash costs what a mine must spend to stay the size it is: sustaining capital, sustaining exploration, site and attributable corporate overhead, reclamation accruals, and royalties and production taxes. It excludes growth capital, income taxes, interest, and the cost of buying the mine. All-in cost adds growth capital back. The gap between gold and AISC is therefore the cash the mine throws off before taxes, debt service and new projects, and it should show up, discounted, in free cash flow.

Three things to know. The first is that the measure is not audited or defined by accounting standards, so companies have latitude in what they call sustaining versus growth, and a company whose AISC looks low next to its free cash flow is often classifying spending as growth. Newmont’s second-quarter 2026 AISC of $1,621 sat next to $2.2 billion of free cash flow on 1.3 million ounces, which reconciles.

The second is that AISC rises with the gold price, and not only through inflation: higher prices let mines process lower-grade ore profitably, which raises the unit cost while raising total profit, and they raise the royalties governments take as a share of revenue. Agnico’s AISC went from $1,339 for 2025 to a 2026 range of $1,400 to $1,550 and a second-quarter print of $1,459, on a realized price that went from $3,453 to $4,483.

The third is that energy is the swing input: diesel and power run to a fifth or more of a large open pit’s cost, and the mid-2026 oil price is why several majors’ second-half cost commentary turned cautious.

The industry average, on the World Gold Council and Metals Focus series, was $1,605 an ounce in the third quarter of 2025, up 9% on the year, so the five here bracket it. Use the number as a strike price: the margin as a share of the gold price tells you how much of a correction the company can absorb before it stops making money. At the September 8, 2026 close on our tape, Agnico’s margin was 67% of the price and Gold Fields’ 57%; a 45% fall in gold, the size of the 2011–2015 decline, would leave Agnico with $985 an ounce and Gold Fields with $551. In 2015 the equivalent figures for most of the industry were near zero.

The five large producers behave differently enough that “gold miners” is not one bet, so each gets its latest published numbers here. The figures are from each company’s results releases for the second quarter of 2026 unless dated otherwise; they are the fast-moving part of this guide.

Newmont is the largest producer by ounces and the only one in the S&P 500. Its 2025 produced the largest free cash flow a gold miner has reported, a record $7.3 billion, on 5.9 million attributable ounces at a by-product AISC of $1,358. It spent the year clearing the decks after the Newcrest deal: $4.5 billion of after-tax proceeds from selling six non-core mines, $3.4 billion of debt retired, $3.4 billion returned to shareholders, and net cash of $2.1 billion at year-end on 118.2 million ounces of reserves.

In the second quarter of 2026 Newmont produced 1.3 million ounces at a by-product AISC of $1,621, generated a second-quarter record $2.2 billion of free cash flow and returned $1.9 billion in dividends and buybacks since its first-quarter report. Its 2026 guidance is about 5.3 million attributable ounces at about $1,680, with 2026 the low point for production before growth resumes in 2027. Newmont is the index; whatever the sector does, it does.

Barrick, renamed Barrick Mining in 2025 to reflect its copper ambitions, is the cautionary major. Its 2025 was 3.26 million ounces, 17% fewer than 2024, at an AISC of $1,637, on 85 million ounces of reserves. Its second quarter of 2026 was good on its own terms: 796,000 ounces against guidance of 730,000 to 770,000, an AISC of $1,866 on a cost of sales of $1,993, and record underground tonnes at Cortez in Nevada. The cash did not follow the ounces: attributable free cash flow was $141 million after a one-time $200 million payment, and $1.4 billion for the half.

The reason it beat guidance is the reason to be careful. Loulo-Gounkoto in Mali, which produced 723,000 ounces in 2024, spent the second half of 2025 under a court-ordered provisional administration and came back into the plan only after a November 2025 settlement that cost Barrick about $430 million; Section 10 has the story. Barrick’s 2026 AISC guidance of $1,760 to $1,950, set at a $4,500 gold price, is the highest of the four 2026 ranges we verified, and its growth capital is going to Reko Diq in Pakistan and Fourmile in Nevada. It is the major with the most leverage to gold and the most exposure to governments, which is the same fact stated twice.

Agnico Eagle is the quality name, and the market has priced it that way for years. It produced 3.45 million ounces in 2025 at an AISC of $1,339, the lowest of the majors, generated a record $4.4 billion of free cash flow, returned $1.4 billion and raised its dividend by 12.5%; its second quarter of 2026 was described above. Nearly all of its production is in Canada, Finland, Australia and Mexico, which is what its cost advantage and its multiple are made of; 2026 guidance is 3.3 to 3.5 million ounces at $1,400 to $1,550.

Kinross is the mid-cost operator whose numbers show what a mine plan does to a quarter. It produced 492,000 gold-equivalent ounces in the second quarter of 2026 at an AISC of $1,821, above its 2026 guidance of $1,730, because Tasiast in Mauritania and Paracatu in Brazil carried the quarter while its American mines were in transition and waste stripping ran ahead of ore. It still generated $727 million of attributable free cash flow after $406 million of capital and $327 million of tax, on guidance of 2.0 million ounces and $1.5 billion of capital for the year. A high AISC in a quarter of stripping is often the cheapest ounces of the next two years, if the plan says so.

Gold Fields, the South African-domiciled major with mines in Australia, Ghana, Chile, Peru and South Africa, shows what a single asset can do to a company. Its first-half 2026 production of 1.267 million ounces was up 12% and its AISC of $1,893 up 13%, both because of Salares Norte, which produced 337,000 gold-equivalent ounces in the half at $269 and threw off $1.19 billion of adjusted free cash flow on its own.

Group adjusted free cash flow of $2.2 billion was more than double the prior period and net debt fell to $437 million. It bought Gold Road Resources in Australia in October 2025 for about A$3.7 billion on the headline count, A$3.3 billion net of Gold Road’s stake in Northern Star, a deal struck near the top of the run whose value depends on where gold is in 2028.

Read the five together and the pattern is the one the cycle always produces. Costs are up 10% to 25% on 2025 guidance, because higher prices raise royalties, lower cut-off grades and energy bills at once. Cash is going out as buybacks, dividends and debt repayment, which is new; in 2011 it went into acquisitions and new mines. And at each company there is a project or a purchase that will look prescient or expensive depending on a gold price nobody there controls.

The margins above need a fixed reference, and ours is our own tape. It holds daily front-month gold futures closes from Yahoo Finance (GC=F), 500 sessions from September 12, 2024 to September 8, 2026, in Invest Alternative’s own collection engine; futures closes differ from the London benchmark by tens of dollars, and every figure here is ours, not a market-wide one.

The window opened at $2,551.20 and closed at $4,443.90, a gain of 74.2%; the one-year change was +22.2%. The average close was $3,447 in 2025, $4,863 in the first quarter of 2026, $4,514 in the second and $4,258 in the third to September 8. The highest close was $5,318.40 on January 29, 2026; the low after it was $3,985.60 on July 16, a drawdown of 25.1%; the second quarter was −13.4% close to close, against the −16% of the London benchmark. Annualised daily volatility was 24.3% across the window and 31.3% in 2026 to date. Our Precious Metals sub-index, an equal composite of the four metals rebased to 100 on September 2, 2025, stood at 139.05 on September 8, up 39.8% on its published one-year change.

Set the miners’ costs against that series and the boom explains itself. Agnico’s 2025 AISC of $1,339 against our 2025 average close of $3,447 was a margin of $2,108, 61% of the price; a year earlier, on 2024 prices near $2,400 and costs near $1,250, the margin had been about half that in dollars. By the second quarter of 2026 it was $3,055 at Agnico, $2,893 at Newmont, $2,693 at Kinross, $2,648 at Barrick and $2,621 at Gold Fields on our $4,514 average, between 58% and 68% of revenue. Margins like these, held for two consecutive years, have no precedent in the AISC era, and the stock prices, up about 50% over the year to June 2026 on GDX’s trailing return against 22% for the metal, say the market does not expect a third.

The sales pitch for miners is leverage; the record, over the longest clean window available, is a lag. The VanEck Gold Miners ETF, GDX, was launched on May 16, 2006, holds the large and mid-cap producers plus the major royalty companies, and is the vehicle most retail investors use. Its own performance pages give its since-inception return as 5.3% a year (Schwab, as of May 31, 2026); other performance services quote 5.9% and 6.8%, depending on end date and method. The range is the honest number.

Gold over the same window did far better. The London price in the week GDX launched was about $700; to our tape’s $4,443.90 on September 8, 2026 that is 6.35 times, about 9.5% a year over 20.3 years. A twenty-year holder of the sector ETF has therefore made a little more than half the annual return of the metal it was supposed to lever, with roughly twice the volatility; at 5.3% a year the dollar became $2.82 while gold’s became more than six. The S&P 500 returned about 9.5% a year with dividends over the two decades to August 2026, roughly what gold did, so the miners lost to the stock market by the same margin, and with a far worse drawdown.

The gap has four causes, each a permanent feature of the business.

Cost inflation

The price of an ounce rose sixfold from 2006 to 2026, and the cost of producing one rose several-fold too: from cash costs near $300 an ounce in 2006 (from the record) to an industry-average all-in sustaining cost of $1,605 in the third quarter of 2025, a broader measure than existed in 2006 but the same direction of travel. Grades fell, pits deepened, energy and labour repriced and governments raised their take, so the margin that leverage is supposed to multiply spent most of the period being eaten.

Capital destruction

At the top of every cycle the industry buys mines and builds projects at prices that only make sense at the top, and writes them off at the bottom. The 2013 wave, when Barrick took $5.1 billion on Pascua-Lama and the majors together impaired tens of billions of dollars of assets bought in 2010 and 2011, is the textbook case.

Dilution

Equity is the industry’s currency for acquisitions and for surviving the bust, so ounces per share fall even when ounces rise.

Hedging

Through the 2000s, Barrick and others had sold forward years of production below $400 and spent billions unwinding the book as gold ran. The practice is largely gone, which is one reason the 2024–2026 margins reached the shareholder.

The bust is the part of the record to memorise. From the September 2011 top, gold fell 45%; the sector ETF fell 80.3% to January 19, 2016, lost more than half its value in 2013 alone, and did not regain its 2011 high for 2,402 trading sessions, nine and a half years, until the run of 2025. A 45% fall in the price of the product became an 80% fall in the price of the producer because the margin went to zero and the industry, having spent the boom, had to issue equity and sell assets at the bottom to survive. The margin is always thin at the bottom.

The lag is not constant, and the periods in which miners lead are exactly the ones a reader is tempted to buy after. The mechanism is the margin, and the margin has a cycle with three phases.

In the expansion phase, the gold price rises faster than costs, which are set by mine plans, labour contracts and energy prices that adjust with a lag; the margin widens and the equities rise by more than the metal. 2001 to 2007, the first eight months of 2016, 2019 to mid-2020 and 2024 to early 2026 were expansion phases, and in each the sector beat the metal by a wide margin.

In the catch-up phase, costs rise into the higher price as grades are lowered, royalties step up, wages reset and the projects sanctioned in the expansion absorb cash; the metal can keep rising while the equities go sideways or fall, as in 2010 to 2011, 2021 to 2023 and the second quarter of 2026, with GDX down 21% against gold’s 16%.

In the contraction phase, the price falls into a cost base raised to meet it, the margin collapses, and the equities fall far more than the metal: 2008, when the sector lost roughly 70% in seven months against the metal’s 30%, and 2011 to 2015.

Two things decide which phase you are in, and neither is the gold price. The first is the direction of cost guidance relative to the price: an industry whose AISC guidance rises 10% to 25% into a year, as the majors’ did for 2026, is in catch-up however good the trailing quarter looked. The second is what the companies do with the cash. In an expansion it goes to the balance sheet and the shareholder; late in one it goes to acquisitions at premiums and to projects justified at the spot price.

Newmont’s purchase of Newcrest in late 2023 for about $19 billion of enterprise value, Gold Fields’ purchase of Gold Road and Royal Gold’s $3.5 billion combination with Sandstorm and Horizon in 2025, and the industry’s 2026 capital guidance are the tell of a cycle moving from the first phase into the second. The dividends and buybacks of 2025 and 2026 are the strongest argument that the miners have learned something since 2011; the capital guidance is the argument that they have not learned everything.

The royalty model is the cleanest form of gold-price leverage there is, and its weakness is not the one most readers expect. The business was invented in 1986, when the original Franco-Nevada, run by Pierre Lassonde and Seymour Schulich, paid $2 million for a 4% royalty on a Nevada property that became Barrick’s Goldstrike, the richest gold mine in North America; the royalty paid for itself many times a year for decades and never sent an invoice for a truck.

Franco-Nevada was relaunched by public offering in December 2007. Silver Wheaton, the first company built on streams, was created in 2004 by Wheaton River Minerals, which merged into Goldcorp the following year, and became Wheaton Precious Metals in 2017. Royal Gold in Denver and OR Royalties in Montreal, Osisko Gold Royalties until 2025, complete the big four.

The contracts exist because mines are expensive to build and miners are expensive to finance. A producer that needs $500 million for a plant can issue equity, borrow, or sell a stream: a promise to deliver, say, 10% of the mine’s gold for its life at $400 an ounce, in exchange for the $500 million today. To the miner, the stream is cheaper than equity at the bottom of the cycle and ruinously expensive at the top, because it gave away the upside on those ounces for ever. To the buyer, it is a perpetual call on the gold price with a strike far below spot and a premium paid once. Both royalties and streams attach to the ground, so a discovery or an expansion on the covered claims accrues to the holder without a further dollar.

The economics follow. Wheaton’s cash cost in the second quarter of 2026 was $568 per gold-equivalent ounce, the fixed delivery prices written into its contracts, against a margin of $3,875; OR Royalties, holding mostly royalties, kept 96.8% of its $97.8 million of revenue as cash margin. Franco-Nevada’s revenue for the quarter was $580.9 million, up 57%, on 132,405 gold-equivalent ounces sold; Wheaton’s was a record $929 million, with $543 million of net earnings and a record $1.8 billion of revenue in the half.

Royal Gold reported revenue of $451 million, up 115%, and record operating cash flow of $335 million. Gold was 76% of that revenue, silver 12% and copper 8%; its new Kansanshi stream and the Sandstorm and Horizon assets it bought in 2025 arrived as gold rose 37%, silver 117% and copper 40% on the year. Each company runs with a few dozen employees.

What the model gives up is control. The holder cannot fix a mine it does not run, and its revenue concentrates wherever its biggest contract sits: Canadian Malartic for OR Royalties, Salobo in Brazil for Wheaton, Mount Milligan and Pueblo Viejo for Royal Gold, though it says no single asset exceeds 13% of its revenue after the 2025 deals, Cobre Panamá for Franco-Nevada. The growth also comes from other people’s decisions: Wheaton guides to 860,000 to 940,000 gold-equivalent ounces for 2026 and about 1.2 million by 2030, OR Royalties to 80,000 to 90,000 and roughly 50% growth by 2030, and both depend on mines being built on schedule by companies the royalty holders do not manage.

The royalty companies trade at a large premium to the producers, the premium is the reason most readers who look at them decide not to buy, and it is mostly justified. The sector values itself on price to net asset value: the present value of the contracted ounces at a consensus gold price, discounted at 5%, set against the market capitalisation. Producers have historically traded near or below 1.0 times that figure; the royalty companies have traded at two to three times it for most of the last decade. We could not verify the multiples as of September 2026 on this desk, and they move with the gold price; the structure of the premium does not.

The premium pays for four things.

No cost inflation

The delivery price in a stream contract is fixed, usually with a small annual escalator, so an ounce bought under a 2016 contract costs roughly what it did in 2016 while a producer’s cost rose by half. Wheaton’s average moved from $406 to $568 in a year because new contracts entered the mix, not because old ones repriced.

Diversification without overhead

Franco-Nevada holds interests in more than 400 assets on its own count, with no operating staff at any of them.

Optionality on the ground

Exploration success on covered ground belongs to the holder without a dollar of drilling.

The dividend record

Franco-Nevada has raised its dividend every year since its 2007 listing, nineteen consecutive annual increases by its own count as of 2026, including through the 2013–2015 bust when most producers cut theirs. That is the reason the equity is held through corrections by investors who sell miners at the first cost overrun.

What the premium does not buy is immunity from the equity market or from governments. A royalty company at 2.5 times net asset value has already priced in a gold price well above spot, so its shares move less than a producer’s for a given move in gold, in both directions; its leverage is lower precisely because its quality is higher. It does not remove jurisdiction, as the next section shows. And a stream bought at the top of the cycle is capital destroyed as surely as a mine bought there; the 2025 consolidation of the royalty sector, led by Royal Gold’s $3.5 billion combination with Sandstorm Gold and Horizon Copper, was struck at prices that will be judged in 2028.

Two of the largest losses in the sector in the 2020s were decisions by governments, and both hit companies that had done everything else right. Neither the miner’s cost curve nor the royalty holder’s contract diversifies that risk away: every ounce sits under a sovereign, and the sovereign’s share of the ounce rises with the price of it.

Cobre Panamá produced 350,000 tonnes of copper in 2022 and 331,000 in 2023, roughly 1% to 1.5% of world supply, with gold and silver by-products sold in advance to Franco-Nevada under a stream for which the company had paid about $1.36 billion since 2012. On November 28, 2023, after weeks of national protests over a renewed contract the legislature had approved a month earlier, Panama’s Supreme Court found the contract unconstitutional. First Quantum, the operator, stopped the mine within days; Franco-Nevada wrote the stream down to nil in the fourth quarter, an impairment of $1,169 million, and filed for arbitration claiming at least $5 billion.

The mine sat idle through 2024 and 2025. In April 2026 the government allowed First Quantum to process the stockpiles on site, about 38 million tonnes of ore holding some 70,000 tonnes of recoverable copper, and Franco-Nevada expects 23,100 ounces of gold and 265,000 ounces of silver from them in 2026. An interministerial commission is reviewing the mine’s future; the government expects to choose among its options, including a state-owned operator, by the end of 2026, and formal talks are expected late in 2026 or early in 2027 on terms carrying a higher tax and royalty burden. A company built to have no operating risk lost its largest single asset, for three years and counting, to a court.

Loulo-Gounkoto is the producer’s version. Mali adopted a new mining code in 2023 that raised the state’s share and its royalties; Barrick, whose complex there produced 723,000 ounces in 2024, disputed its application to existing operations. The state detained Barrick employees in late 2024 and blocked exports, and on June 16, 2025 a Malian commercial court placed the mine under a six-month provisional administration, after which the state ran it without the company. The November 2025 settlement cost Barrick about $430 million for the dropping of charges, the release of its four detained employees, the return of about three tonnes of seized gold and the return of the mine in December 2025.

Kyrgyzstan’s seizure of Centerra’s Kumtor mine in May 2021 followed the same script, and the sovereign’s argument in each case, that terms agreed at $1,200 gold are unfair at $3,000, is one every host government makes somewhere in every boom.

The practical reading is a country weighting. Agnico’s multiple has been higher than Barrick’s for years because a buyer of Agnico is not buying Mali, Pakistan, Papua New Guinea or the Dominican Republic; a mine in a stable jurisdiction can be a bad mine but it cannot be an expropriated one. When you read a producer or a royalty company, tabulate its revenue by country from the annual report, note the share in any country that has changed its mining code in the previous five years, and discount that share, because the market will the next time a parliament meets.

The losses in this sector that cannot be recovered are not the 80% drawdowns but the ones that end the company, and they come in three kinds. The first is the tailings dam. Every mine stores its processed waste as slurry behind an earthen dam, among the largest structures humans build and the least inspected. On January 25, 2019, the dam at Vale’s mine at Brumadinho in Brazil collapsed and killed 270 people; the Fundão dam at Samarco, owned by Vale and BHP, had failed on November 5, 2015, killing 19; Imperial Metals’ Mount Polley dam in British Columbia had failed on August 4, 2014. The industry’s response was the Global Industry Standard on Tailings Management, published in August 2020.

A tailings failure is uninsurable, unhedgeable and priced at zero in the equity until it happens; the only protection is to read the tailings disclosure every major has published since the 2020 standard and to prefer companies with few upstream-construction dams and none above populated valleys.

The second is fraud, and the sector’s own case is the one every analyst is trained on. Bre-X Minerals, a Calgary junior, claimed through 1996 and early 1997 to have found the largest gold deposit in history at Busang in Indonesia; the drill samples had been salted, the chief geologist fell from a helicopter on March 19, 1997, a week before Freeport-McMoRan’s resample results were released, and a company that had reached a market value of about C$6 billion was worthless by May 1997. Canada’s National Instrument 43-101 for resource disclosure is the direct result; the lesson holds that a resource statement is a claim about rock nobody has seen, and the smaller the company and the larger the number, the more the claim needs checking.

The third is the junior, which fails without fraud. Most exploration companies find nothing; most that find something cannot finance it; and in a contraction the equity market that funds them closes at once, so the cash runs out with the drill program half done. The 2008 crisis and the 2013–2015 bust each wiped out hundreds of listed juniors. This is not a case against juniors for a reader who can read a 43-101 report and size a position that can go to zero; it is a case against buying the sector’s leverage through its most fragile names.

Owning the sector costs less than owning physical gold and more than nothing: no premium over spot, no vault and no assay, but a fund fee or a spread, a withholding tax on foreign dividends, and for some products a structural decay that dwarfs all three.

The ETFs

GDX charges 0.51% a year and holds the large and mid-cap producers plus the major royalty companies, of which Wheaton alone was 5.8% of the fund in 2026 and the royalty names together are usually put at 10% to 15%, a share we did not verify; GDXJ, the junior fund, charges 0.52%. iShares’ MSCI Global Gold Miners fund (RING) charges 0.39% and Sprott’s factor fund (SGDM) 0.46% and US Global’s (GOAU) 0.60%, as published on their fund pages. Against GLD’s 0.40% and GLDM’s 0.10%, the sector ETFs cost about what the metal ETFs do; the difference over a decade is a few hundred dollars per $100,000, not a decision.

Single stocks

Commissions are zero at every large US broker and the spreads on Newmont, Agnico, Barrick, Franco-Nevada and Wheaton are a cent or two. The cost that is not zero is withholding: Canada withholds 15% of dividends paid to US taxable accounts under the treaty, recoverable as a foreign tax credit and waived entirely for dividends paid into an IRA or 401(k); South Africa withholds 20% on Gold Fields’ dividends, reduced to 15% with treaty paperwork through the depositary, which also charges a few cents a share a year on the ADR. On a 1% yield these are rounding errors.

Leverage products

The leveraged daily funds on the miners (Direxion’s NUGT and JNUG, the largest, have been 2× since they were cut from 3× on March 31, 2020) reset their leverage every day, so over any longer period their return is the leveraged daily return compounded, which in a volatile sideways market is a steady loss; at the sector’s 30% to 50% volatility the decay can run to tens of percent a year with no net move in the index. They are trading instruments, not a way to own the sector.

Owning the mine beats owning the metal in one place by statute, and it is the reason a taxable, long-horizon holder of gold exposure should look at the equities at all. Physical gold and the grantor-trust ETFs such as GLD are collectibles under IRC §1(h)(4) and §408(m), and a long-term gain is taxed at a maximum federal rate of 28%, plus the 3.8% net investment income tax of §1411 above $200,000 of modified adjusted gross income for a single filer or $250,000 joint, a combined 31.8%. Futures are Section 1256 contracts at a blended 26.8%. Investing in Gold covers all of that.

A miner, a royalty company, or an ETF that holds them is a corporation or a regulated investment company, and its shares are ordinary capital assets. A gain on shares held more than a year is taxed at 0%, 15% or 20% depending on income, plus the same 3.8% surtax, a top combined rate of 23.8%; a gain on shares held a year or less is ordinary income at up to 37%, as for the metal. The dividends are qualified dividends under §1(h)(11), taxed at the same 20% ceiling, including dividends from Canadian companies listed on a US exchange and from any treaty-country company; Gold Fields’ dividends through its ADR also qualify.

The Canadian and South African withholding is a credit against that tax on Form 1116, or without the form up to $300 for a single filer and $600 joint. Losses on miners are ordinary capital losses, subject to the wash-sale rule of §1091, which does not apply to physical metal; a reader who wants to harvest a loss in a gold position and stay exposed can sell GLD and buy a miner without triggering it. A miner is a stock, so it can be held in any IRA at any custodian without the depository rules that apply to bullion, and a Roth wipes out the difference between 23.8% and 31.8% entirely.

The point of the difference is size. On a $30,000 long-term gain, the metal in GLD costs $8,400 at 28% and $9,540 with the surtax; the same gain in Newmont or Franco-Nevada costs $6,000 or $7,140. That is an 8-point spread on the gain, thousands of dollars per $100,000 over a decade like 2016 to 2026, and it is the only argument for the equities that does not depend on the cycle.

The sector’s marketing is about leverage; the durable edge is in the tax line, and in dollars it looks like this. The assumptions: $100,000 invested on September 8, 2026 and sold ten years later; gold rises 5% a year in nominal terms; GLD at its 0.40% fee, taxed at 31.8% on exit; GDX at 0.51%, paying an assumed 1% dividend taxed at 23.8% each year and reinvested, with 23.8% on the gain at exit; top federal rates, no state tax. The miners are run three ways: tracking the gold price, lagging it by about three points a year, the middle of the 2006–2026 gap, and beating it at 1.5 times, roughly what the margin arithmetic gives at 2026 costs. The yield and the multipliers are assumptions, not forecasts, and the arithmetic is ours.

GLD, the metal in a trust

$100,000 compounds at 5% less the 0.40% fee to $156,490. The gain of $56,490 is taxed at 31.8%, $17,964, leaving $138,526, a 3.31% annual return on the cash. At a flat 28% without the surtax, $140,673.

Miners tracking gold

The shares compound at 5% less 0.51%, the 1% dividend pays $3,165 of tax over the decade and adds $10,135 to the cost basis as it is reinvested, and the position ends at $166,976. The gain over the raised basis is $56,842, taxed at 23.8% for $13,528, leaving $153,448, a 4.37% annual return. The miners made the same 5% as the metal and finished $14,922 ahead, more than half of it from the tax rate and the rest from the dividend the metal cannot pay.

Miners lagging by three points

If the sector repeats its 2006–2026 record and compounds at 2% while gold does 5%, the position ends at $124,958 before tax and $121,057 after, 1.93% a year and $17,469 behind GLD. The tax advantage is worth about eight points of the gain; a three-point-a-year lag is worth thirty points of the principal. That is the case against the sector as a substitute for the metal, in one line.

Miners at 1.5 times gold

If the leverage works, 7.5% a year becomes $211,274 before tax and $187,570 after, 6.49% a year, $49,044 ahead of the metal. If gold is flat for the decade instead, GLD returns $96,071 with no tax due, and the same leverage runs the other way: at a 5% annual decline in the shares the miners return $61,376.

A single royalty stock with no fund fee, tracking gold and paying the same 1%, ends at $160,197; in a Roth, GLD ends at $156,490 and the miners at $170,963. The order of the results is the guide’s argument: the tax code hands the equities an eight-point head start, the dividend adds a little, and the cycle decides whether the leverage gives it back.

There are five ways into the sector, ranked here honestly, and then an order in which to do things. The sector ETF is the default and for most readers the right answer: GDX at 0.51% holds the majors and the royalty companies in one line, trades a cent wide, and is the instrument the margin rule in Section 7 is written for; its weakness is that it holds the higher-cost majors at full weight. A basket of single names costs nothing in fees and lets you tilt: equal positions in Agnico, Newmont, Franco-Nevada and Wheaton are a lower-cost, lower-jurisdiction-risk version of the index with a larger royalty weight, at the price of four annual reports a year.

The royalty companies alone suit a reader who wants gold-price exposure with equity tax and can accept a lower beta and a premium multiple. The junior fund and single juniors are for readers who have read a 43-101 report and can hold a position that goes to zero.

The sequence, as we would run it:

1. Decide what the position is for. If it is a substitute for holding gold, stop, and buy the metal in the wrapper Investing in Gold recommends; the record says the miners are not that. If it is an equity position that happens to be leveraged to gold, size it inside the equity allocation, not the gold sleeve.
2. Read the margin: gold less the latest quarterly AISC of Agnico and Newmont, as a share of price. Above 60% with capital guidance rising, buy slowly or wait; below 30% with capital guidance falling, buy the sector.
3. Choose the split between producers and royalty companies by the rule in Section 9, and the instrument by fee: GDX for one line, four single names for control.
4. Put the position in a Roth or traditional IRA if you have the room, where the Canadian withholding is waived and the 23.8% becomes zero or deferred.
5. Review on the reporting calendar: February for full-year results and guidance, when the cost rule in Section 3 is applied, and the quarter-ends for the margin. Sell a company, not the sector, when its AISC rises more than 10% in a year its realized price rose less than 10%.
6. Rebalance to the target weight at every review and never add on strength; the sector’s entire excess return over the metal has come from being bought in contractions.

Seven readings would change the view. Each carries the level at which it would, as of September 8, 2026, so a reader in 2027 can check it against the date.

• The margin on our tape. Gold at $4,443.90 against Agnico’s $1,459 and Newmont’s $1,621 is a margin of 64% to 67% of price. A fall in gold to $3,300 with costs unchanged takes the margin to 51% to 56% and moves the sector rule from trim to hold; at $2,400 it is 32% to 39% and near the buy zone. A rise of more than 10% in the majors’ 2027 AISC guidance, due in February 2027, does the same work from the cost side.
• Capital guidance. Barrick’s 2026 attributable capex of $3.8 to $4.2 billion and Kinross’s $1.5 billion are the marks. Aggregate 2027 guidance up more than 15% with the gold price flat is the late-cycle signal; guidance cut with the price flat is the early sign of the contraction that precedes the buy.
• Deals. Two or more acquisitions by the majors or the royalty companies at premiums above 30% within a year is the tell that the cycle has moved into its spending phase. Royal Gold’s $3.5 billion Sandstorm combination and Gold Fields’ A$3.7 billion Gold Road purchase, both in 2025, were the first two of this cycle.
• Cobre Panamá. Panama’s decision on the mine’s future is due by the end of 2026, with talks expected to open late in 2026 or early in 2027. A restart under First Quantum with the stream intact is worth several times Franco-Nevada’s expected 2026 stockpile deliveries of 23,100 ounces of gold; a state-owned operator that does not honour the stream is a second impairment.
• Mali and the code. Loulo-Gounkoto is back in Barrick’s 2026 guidance; any new export block or administration, or the adoption of Mali’s approach by another West African government, is a sell for the producers most exposed.
• Energy. An oil price sustained above the level the majors used for 2026 guidance shows up in AISC within two quarters; the full-year results in February 2027 will say whether the $1,459 to $1,893 range held.
• GDX against the metal. The ETF rose about 140% on its net asset value in calendar 2025, or 155% on one total-return measure; the two published figures disagree and the fund’s own page could not be opened, so treat the range as the number. It was up about 50% over the year to June 2026 against 22% for gold, fell 21% against 16% in the second quarter, and closed at about $98.56 on September 8, 2026 on published price data, above its end-March level. A twelve-month window in which GDX beats the metal from a margin reading above 60% falsifies the sector rule in Section 7, and we will say so.