Gold — A Complete Market Guide (2026)
Data as of 7 October 2026. Market structure is shown for full-year 2025 from the USGS Mineral Commodity Summaries 2026, the World Gold Council’s supply-and-demand statistics (data to 30 June 2026), UN Comtrade and the miners’ 2025 filings.
Gold is the oldest monetary asset still trading every business day, and in 2025 the gold market did something unusual: the price averaged a record $3,442 an ounce — 44% above 2024 and, after inflation, 45% above its 1980 average — while US real interest rates averaged their highest since 2007, a combination that normally works against it. This report is the free, big-picture primer on how the gold market works: where gold is mined, who buys it and why, how the price is set, and which economic regimes favour it. For the company-level data behind the story — every producer screened by production, reserves and cost — go to Metal Pilot’s gold mining stock screener.
TL;DR & Key Takeaways
- What it is: a near-indestructible precious metal that is at once a reserve asset, a store of value, jewellery and a minor industrial input. Almost all the gold ever mined still exists — about 222,600 t above ground, some 58 years of mine output on the WGC’s count — so gold’s defining feature is that stock, not its annual flow.
- Who supplies it: mines produced an estimated 3,300 t in 2025, led by China, Russia and Australia; no country mines more than about 12%, so there is no supply cartel. Recycled jewellery and scrap add about 1,400 t, 27% of total supply on the WGC’s count.
- Who buys it: 2025 demand reached a record 5,173 t. Investment — bars, coins and ETFs, 2,209 t — overtook jewellery, and central banks bought 848 t after more than 1,000 t a year in 2022–2024. China and India take half of all consumer jewellery, bar and coin demand.
- How it is priced: the gold market sets its price in London (the LBMA Gold Price auction) and on COMEX futures, not by producers. It averaged $3,442/oz in 2025 and $4,452/oz over the 12 months to September 2026.
- The regime that favours it: gold pays no yield and is priced in dollars, so it does best when real interest rates fall, the dollar weakens or confidence in financial assets breaks. Since 2022 official-sector buying has overridden high real rates — the swing factor to watch alongside the real rate itself.
Numbers to remember: mine production ≈ 3,300 t (2025e) · market size ≈ $365 bn · gold price $3,442/oz (2025 average); $4,452/oz (12 months to Sep 2026) · top producer China ≈ 380 t · top consumer China ≈ 792 t · recycling ≈ 27% of supply (WGC) · reserve life ≈ 20 years · central-bank buying 848 t (2025).
Figure 1. Gold at a glance, 2025
Notes to Figure 1
- Mine production — Table 9 · Filed | USGS MCS 2026 | World Mine Production and Reserves | “World total (rounded)” 3,300 t, 2025e | p.91. The World Gold Council’s estimate, which adds artisanal output, is 3,823 t (Table 12).
- Market size — Derived | 3,300 t × 32,150.7 troy oz per tonne × $3,442/oz (Table 3, 2025 average) = $365 bn; the value of one year’s mine output, not of any company.
- Price, 12 months to Sep 2026 — Market | World Bank | “Gold”, monthly | Oct 2025–Sep 2026; the mean of the twelve monthly averages, computed here. For reference only: the 2025 calendar-year average was $3,442/oz (Table 3).
- Top producer — Table 9 · Filed | USGS MCS 2026 | World Mine Production and Reserves | “China” 380 t, 2025e | p.91.
- Top consumer — Table 7 · Derived | 360.1 t jewellery + 431.7 t bars and coins = 792 t, 2025, from WGC GDT FY2025 Tables 2 and 3, row “China, P.R.: Mainland”; India is second at 710.9 t.
- Supply from recycling — Table 12 · Derived | “Recycled gold” 1,403.8 t ÷ total supply 5,172.7 t, 2025 (WGC, data to 30 June 2026).
- Reserve life — Table 11 · Derived | USGS world reserves 66,000 t ÷ 2025e mine production 3,300 t.
- Central-bank buying — Table 8 · Market | WGC supply & demand | “Central banks” 848.2 t, 2025; first printed at 863 t in January 2026.
Figure data: Tables 3, 7, 8, 9, 11 and 12 — USGS Mineral Commodity Summaries 2026: Gold (February 2026; page numbers are those of the full volume), the World Gold Council’s gold supply and demand statistics (data to 30 June 2026) and the World Bank Pink Sheet, monthly; each tile’s citation is in the notes above.
How to read this guide: new to gold? Read straight through from Section 1. Here for the investment case? Jump to Section 4 — how to own it, how to screen the miners, and which regimes favour the metal. Every technical term is defined in the Terms list at the end.
1. Gold & the market basics
1.1 What gold is — physical basics & quality
Gold (chemical symbol Au) is a dense, soft, yellow metal prized for a combination no other element offers: it is chemically inert (it does not corrode, tarnish or react with oxygen), it is scarce but divisible, and it is easy to work. Those properties are why societies converged on gold as money and as a store of value long before modern finance. Today gold earns its value from four overlapping demand pools: monetary reserves (central banks), jewellery (adornment that doubles as savings, especially in Asia), investment (bars, coins and exchange-traded funds) and a small, high-specification industrial slice (electronics, dentistry).
A few terms define gold’s quality and form, each used throughout this report:
- Purity / fineness — how much of a piece is pure gold, expressed as karats (24 karat = pure gold, 18 karat = 75%) or as fineness in parts per thousand (995 = 99.5% pure, the minimum for a London “Good Delivery” bar). Gold is weighed in the troy ounce (oz), heavier than the everyday avoirdupois ounce; the unit convention is set out in Section 1.2.
- Doré — the impure gold-silver alloy bar a mine pours on site, shipped to a refinery to be upgraded to bullion.
- Ore grade — grams of gold per tonne of rock (g/t). Grade, not the gold price alone, decides whether a deposit is economic: open pits mine far lower grades than underground operations because moving rock in bulk is cheaper.
- Primary vs. by-product — gold mined for its own sake versus gold recovered alongside copper or other metals (Section 2.5).
The value chain — from ground to vault. Gold moves along one clean path: exploration → mining → processing → smelting (doré) → refining (bullion) → fabrication (bars, coins, jewellery, components) → the end buyer (investors, central banks, consumers, industry). Alongside that physical chain runs a deep financial market — the London over-the-counter (OTC) market and COMEX futures — where far more gold changes hands on paper than moves physically.
Figure 2. The gold value chain, ground to vault
Chart source: this guide — the seven stages described in Section 1.1, grouped into four steps (smelting sits with processing at the mine; the end buyers follow fabrication); the refined-bar standard is the LBMA Good Delivery List Rules. A conceptual diagram with no values.
From rock to doré. Between “mining” and the first bar, ore runs through a fixed sequence of steps whose names recur in every technical report. It is blasted and hauled from an open pit or an underground stope, then put through comminution: a crusher breaks it to gravel and a mill — typically a SAG (semi-autogenous) mill followed by a ball mill — grinds it to powder, because gold locked in rock must be physically liberated before it can be extracted. Coarse free gold is caught in gravity circuits; sulphide ores may be upgraded by flotation; and the gold itself is dissolved by cyanide leaching, usually run as carbon-in-leach / carbon-in-pulp (CIL/CIP), in which activated carbon adsorbs the dissolved gold from the slurry. Low-grade ore is instead stacked on a lined pad and heap-leached — cheaper, slower and with lower recovery — while refractory ore, whose gold is locked in sulphides, must first be roasted or pressure-oxidised. The gold is then stripped from the carbon and plated out by electrowinning.
Smelting vs. refining — two steps, often confused. Smelting happens at the mine: the gold-rich product is melted with fluxes that carry rock and base metals off as slag, and the molten metal is poured into doré. It concentrates and casts the metal but does not purify it to a tradable grade. Refining happens afterwards at a dedicated refinery: the Miller process bubbles chlorine through molten metal to strip out base metals and silver, and the Wohlwill process electro-refines it to “four-nines” (.9999) gold. The split is commercial as well as technical: doré cannot be delivered into the London bullion market — only a refiner on the LBMA’s Good Delivery List can make a bar that can — which makes refining a concentrated, gatekept link in the chain (Section 2.6).
1.2 Units & measurement conventions
This report uses the precious-metals convention throughout. Gold is weighed in troy ounces (oz) and metric tonnes (t); one tonne is 32,150.7 troy ounces. Company and mine output is in Moz (million troy ounces) and koz (thousand troy ounces); national and global flows are in tonnes, as the U.S. Geological Survey (USGS) and the World Gold Council (WGC) report them. Prices are US dollars per troy ounce unless stated, and grade is grams per tonne (g/t).
Table 1. Gold units and conversions
| Unit | Meaning | Typical use in gold | Conversion | Source |
|---|---|---|---|---|
| troy ounce (oz) | Precious-metal weight | Every price; coin sizes | 31.103 g | Filed | USGS MCS 2026 | Appendix A | “1 troy ounce (tr oz) = 1.09714 avoirdupois ounces, or 31.103 grams” | p.1 |
| tonne (t) | Metric tonne of contained gold | National and global flows | 32,150.7 oz | Filed | USGS MCS 2026 | Gold, footnote 1 | “One metric ton (1,000 kilograms) = 32,150.7 troy ounces” | p.90 |
| Moz | Million troy ounces | Company output and reserves | 31.1 t | Derived | 1,000,000 oz ÷ 32,150.7 oz per tonne |
| koz | Thousand troy ounces | Single-mine output | 0.0311 t | Derived | 1,000 oz ÷ 32,150.7 oz per tonne |
| g/t | Grams of gold per tonne of ore | Ore grade | 0.0322 oz per tonne | Derived | 1 g ÷ 31.103 g per oz |
| Good Delivery bar | The London wholesale bar | Vaulting and settlement | 350–430 fine oz; at least 995 fine | Filed | LBMA GD Rules 2026 | §2.1.7 | “Minimum gold content: 350 fine troy ounces” | p.12 |
Source: USGS Mineral Commodity Summaries 2026, Appendix A and the Gold chapter, and the LBMA Good Delivery List Rules, each row as cited. The Moz, koz and g/t conversions are computed here from the USGS factors and rounded.
Two counts of mine output. The USGS estimates 2025 world mine production at 3,300 t, country by country; the WGC, using Metals Focus data that add estimates of artisanal and small-scale mining, puts it at 3,823 t. This guide uses the USGS for country rankings, reserves and the long-run series (Tables 9–11), and the WGC for the supply–demand balance (Tables 8 and 12), and never mixes the two in one table without a note saying which is which.
Gold figures split into flow and stock, and confusing them is the most common error in gold analysis. Flows are quantities per year: mine production, recycling, demand. The stock is the cumulative quantity that exists at a point in time — all the gold ever mined and still held, the above-ground stock (Section 2.4). Because gold is not consumed, that stock keeps growing and dwarfs the annual flow.
Numbers intuition: the largest single mines produce 500–1,000 koz a year, and the Nevada Gold Mines complex about 2.6 Moz (Table 5); a Good Delivery bar holds 350–430 oz; and a year’s mine output — 3,300 t, or about 106 Moz — was worth about $365 bn at the 2025 average price: large for a metal, small next to the 222,600 t already above ground.
1.3 Pricing & benchmarks
There is no single “gold price” but a tight web of linked benchmarks. The global reference is the LBMA Gold Price, set in two electronic auctions each London business day and administered by ICE Benchmark Administration (IBA); bars are valued and most over-the-counter contracts settle against it. COMEX gold futures (CME Group) are the most liquid exchange contract, where price discovery and speculative positioning concentrate. The Shanghai Gold Exchange (SGE) prices physical gold inside China, and its premium or discount to London is the clearest signal of Chinese demand.
Because gold is fungible and arbitraged across the globe, regional differences are small premiums and discounts — reflecting shipping, financing and local demand — rather than the large quality differentials of oil. In normal markets gold trades in contango (futures above spot), the gap reflecting the cost of carry: storage plus the interest forgone on the cash.
Table 2. Key gold benchmarks
| Benchmark | What it prices | Where and how | Role | Source |
|---|---|---|---|---|
| LBMA Gold Price | Spot gold held in London | Two daily auctions, 10:30 and 15:00 London, in US dollars; run by IBA | The global reference — bars valued and contracts settled against it | Filed | IBA methodology | ¶4 | “The auctions are run at 10:30 a.m. and 3 p.m. London time for gold” | p.1 |
| COMEX gold futures (GC) | 100 troy oz of gold, at least 995 fine | CME Group, New York | Price discovery, hedging, positioning; small premium to spot in contango | Filed | CME Rulebook ch.113 | Rule 113101 | “one hundred (100) troy ounces of gold” |
| SGE Au99.99 | Physical gold in China | Shanghai; 10 g trading lot, delivered as 1 kg bars of at least 99.99% | China’s demand signal; premium or discount to London | Filed | SGE Au99.99 contract details | “Trading Unit: 10 g/lot”; “Deliverable Bullions: 1kg gold ingot” |
| Loco London | Good Delivery bars in London vaults | OTC market; 350–430 oz bars, at least 995 fine | The deepest physical pool — the physical leg of the LBMA price | Filed | LBMA GD Rules 2026 | §2.1.7 | “Maximum gold content: 430 fine troy ounces” | p.12 |
Source: the contract documents named in each row — IBA, LBMA Gold Price and LBMA Silver Price Methodology; CME Group, COMEX Rulebook chapter 113, Gold Futures; the Shanghai Gold Exchange’s Au99.99 contract details; and the LBMA Good Delivery List Rules. The Role column is this guide’s summary of how each benchmark is used, not a field of the documents.
Because a single day’s price goes stale, the meaningful description of the price regime is a set of averages. Gold averaged $3,442/oz in 2025, $2,275 over 2021–2025 and $1,831 over 2016–2025. The annual series below shows the long cycle the rest of this report returns to: two decades of stagnation ending at about $279 in 2000, a bull market to a then-record $1,670 average in 2012, a bear to $1,161 in 2015, and a second bull from 2019 that accelerated in 2024–2025. In inflation-adjusted terms 2024 was the first year to average above 1980 ($2,376 in 2025 dollars), and 2025 cleared it by 45%.
Table 3. Average annual gold price, 2000–2025
| Year | 2000 | 2001 | 2002 | 2003 | 2004 | 2005 | 2006 | 2007 | 2008 | 2009 | 2010 | 2011 | 2012 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Nominal | $279/oz | $271/oz | $310/oz | $364/oz | $409/oz | $445/oz | $604/oz | $697/oz | $872/oz | $973/oz | $1,225/oz | $1,569/oz | $1,670/oz |
| In 2025 dollars | $522/oz | $493/oz | $555/oz | $637/oz | $697/oz | $734/oz | $965/oz | $1,082/oz | $1,304/oz | $1,460/oz | $1,809/oz | $2,246/oz | $2,342/oz |
| Year | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Nominal | $1,411/oz | $1,266/oz | $1,161/oz | $1,249/oz | $1,258/oz | $1,269/oz | $1,392/oz | $1,770/oz | $1,800/oz | $1,801/oz | $1,943/oz | $2,388/oz | $3,442/oz |
| In 2025 dollars | $1,950/oz | $1,722/oz | $1,577/oz | $1,676/oz | $1,652/oz | $1,627/oz | $1,753/oz | $2,202/oz | $2,139/oz | $1,982/oz | $2,053/oz | $2,451/oz | $3,442/oz |
Source: World Bank, Commodity Price Data (Pink Sheet), annual, sheet “Annual Prices (Nominal)”, column “Gold” ($/troy oz), 2000–2025 — calendar-year averages of the London afternoon (PM) fixing, and from June 2025 the World Bank’s spot average of daily rates, as its Description sheet states; the 2025 value differs from the LBMA PM average of $3,431.5 that the WGC prints by $10. The real row is computed here: nominal × (US CPI-U 2025 ÷ CPI-U of the year), from FRED series CPIAUCSL (annual mean of monthly values; 2025 uses 11 months, October 2025 being unpublished). The multi-year averages in the text are computed here from the nominal row; 1980 is $608 nominal in the same file.
Figure 3. Gold’s long cycle, 2000–2025: nominal and inflation-adjusted average price
Figure data: Table 3. The real line deflates each year by US CPI-U to 2025 dollars, so the two lines meet in 2025.
Current situation (2026, unresolved — for reference only): gold has averaged $4,552/oz from January to September 2026, with a record monthly average of $5,020 in February and $4,319 in September; the trailing 12-month average to September is $4,452 (World Bank monthly series, Figure 1, note 3). These are averages of a year still in progress and anchor none of the analysis below.
2. Supply, demand & the market balance
2.1 Where gold is mined — deposits & geology
Gold is found on every continent, which is why no single country dominates supply the way Saudi Arabia or Russia dominate oil. Production concentrates where the right geology meets an investable jurisdiction, and each deposit type brings its own grade, mining method and processing route — and so its own cost (Table 4). The highest-volume complexes are sediment-hosted Carlin-type deposits in Nevada, giant open pits in Central Asia and Russia, and porphyry copper-gold systems where gold is a by-product. Artisanal and small-scale mining (ASGM) — individuals and small groups working with simple tools, often outside formal regulation — supplies up to a fifth of the world’s gold (Section 4.5).
Table 4. Gold deposit types and where they are mined
| Deposit type | What it is | Main districts | Example | Why it matters to the miner |
|---|---|---|---|---|
| Carlin-type (sediment-hosted) | Fine-grained gold disseminated in sedimentary rock | Nevada, United States | Nevada Gold Mines | Much of the ore is refractory and must be roasted or autoclaved |
| Orogenic / greenstone lode | Gold in quartz veins along ancient fault zones | Abitibi (Ontario–Québec), Western Australia, West Africa | Detour Lake, Canadian Malartic, Ahafo | The classic gold-only mine, open pit or underground |
| Porphyry copper-gold | Large low-grade copper deposits carrying gold | Indonesia, Australia, the Andes | Grasberg | Gold is a by-product; output follows the copper mine plan |
| Epithermal | Shallow veins and disseminations in volcanic belts | The Pacific rim, Latin America | Lihir | Can be high grade; some ores refractory |
| Paleoplacer | Ancient river gravels turned to rock | Witwatersrand, South Africa | South Deep | Deep, labour-intensive and high-cost |
| Artisanal (ASGM) | Small-scale, often informal mining of many deposit types | Africa, Latin America, Asia | — | Up to 20% of world supply; mercury use (Section 4.5) |
Source: this guide’s geological summary; the mine descriptions in Table 5 follow the operators’ filings cited there (e.g. Newmont’s 2025 Form 10-K describes Carlin and Turquoise Ridge as sediment-hosted disseminated deposits and Cadia as an alkalic porphyry gold-copper deposit), and the ASGM share is from the U.S. EPA (“producing up to 20% of the world’s gold annually”). No agency publishes world output by deposit type, so the table carries no output column (Section 7.1).
At the asset level a handful of giant operations anchor supply. The two largest are Nevada Gold Mines, Barrick and Newmont’s joint venture, which produced about 2.6 Moz (80 t) in 2025 — half of all US output — and the Muruntau cluster in Uzbekistan, which its state owner describes as the largest gold deposit in the world. Grasberg shows how by-product supply behaves: its gold output halved in 2025 after a mud rush in September shut much of the underground mine — a cut driven by what happened to a copper mine, not by the gold price.
Table 5. Largest gold mines and complexes, 2025
| Mine / complex | Country | Operator / owners | Output, 2025 · 100% basis | Type and note | Source |
|---|---|---|---|---|---|
| Nevada Gold Mines | United States | Barrick 61.5% (operator), Newmont 38.5% | ~2,590 koz | Carlin-type; Carlin, Cortez, Turquoise Ridge | Derived | ABX MD&A FY2025 | Operating highlights | “Nevada Gold Mines (61.5%)” 1,591 | p.12 + NEM 10-K FY2025 | Operating Statistics | “NGM” 999 | p.63 |
| Muruntau cluster | Uzbekistan | NMMC (state-owned) | ~1,640 koz 1 | Giant open pit | Derived | NMMC presentation | Production FY2025 | “Muruntau Cluster” 52% × 3.15 Moz | p.6 |
| Grasberg district | Indonesia | Freeport-McMoRan 48.76% (operator); MIND ID group 51.24% | 937 koz 2 | Porphyry copper-gold; by-product | Filed | FCX 10-K FY2025 | Consolidated mining data | “Indonesia (48.76%)” 937 | p.34 |
| Olimpiada | Russia | Polyus | 927 koz | Open pit | Filed | Polyus FY2025 MD&A | Consolidated operating results | “Olimpiada” 926.5 | p.4 |
| Detour Lake | Canada | Agnico Eagle 100% | 693 koz | Orogenic; open pit | Filed | AEM MD&A FY2025 | Detour Lake | “Gold production (ounces)” 692,675 | p.21 |
| Kibali | DR Congo | Barrick 45% (operator), AngloGold Ashanti 45%, SOKIMO 10% | ~673 koz | Open pit and underground | Derived | AU 20-F FY2025 | Africa region | “Kibali (Attr. 45%)” 303 ÷ 45% | p.83 |
| Ahafo South | Ghana | Newmont 100% | 664 koz | Orogenic; open pit and underground | Filed | NEM 10-K FY2025 | Operating Statistics | “Ahafo South” 664 | p.63 |
| Canadian Malartic | Canada | Agnico Eagle 100% | 643 koz | Orogenic; open pit and underground | Filed | AEM MD&A FY2025 | Canadian Malartic | “Gold production (ounces)” 642,612 | p.19 |
| Pueblo Viejo | Dominican Republic | Barrick 60% (operator), Newmont 40% | ~632 koz | Open pit | Derived | ABX AIF FY2025 | Pueblo Viejo | “produced 379,014 ounces of gold in 2025 (Barrick’s 60% share)” | p.77 + NEM 10-K FY2025 | “Pueblo Viejo (40%)” 253 | p.7 |
| Paracatu | Brazil | Kinross 100% | 601 koz 3 | Open pit, low grade | Filed | K MD&A FY2025 | Paracatu | “Produced” 601,318 | p.16 |
| Lihir | Papua New Guinea | Newmont 100% | 585 koz | Epithermal; open pit, pressure oxidation | Filed | NEM 10-K FY2025 | Operating Statistics | “Lihir” 585 | p.63 |
| Natalka | Russia | Polyus | 567 koz | Open pit | Filed | Polyus FY2025 MD&A | Consolidated operating results | “Natalka” 567.4 | p.4 |
Notes to Table 5
- NMMC publishes no mine-level output; its April 2026 presentation shows the Muruntau cluster as 52% of 2025 production in a chart, so the figure is a rounded estimate (52% × 3.15 Moz).
- Grasberg’s 2024 output was 1,861 koz. On 8 September 2025 a mud rush — about 800,000 tonnes of wet material — entered the Grasberg Block Cave and killed seven people; the mine restarted in stages from late October (FCX 10-K FY2025, pp.23 and 58).
- Gold-equivalent ounces, Kinross’s reporting basis; Paracatu’s silver content is minor.
Source: the operators’ annual filings for the year to 31 December 2025, each row as cited in its Source column — Barrick 2025 MD&A and AIF (Form 40-F), Newmont 2025 Form 10-K, Freeport-McMoRan 2025 Form 10-K, Polyus FY2025 operating and financial results, Agnico Eagle 2025 MD&A, AngloGold Ashanti 2025 Form 20-F, Kinross 2025 MD&A and the NMMC corporate presentation; printed page numbers. Output on a 100% basis; where no filing prints the 100% figure it is derived as marked — the partners’ attributable figures summed, or one partner’s figure divided by its share. Loulo-Gounkoto in Mali (578 koz attributable to Barrick’s 80% in 2024, about 720 koz at 100%) falls out of the table: it was suspended from January 2025 in a dispute with the state and handed back to Barrick in December (Barrick AIF FY2025, p.22).
2.2 Demand & consumption
Gold demand is unusually broad, which makes it resilient: when one pillar weakens, another often takes over. In 2025 total demand reached a record 5,173 t in the WGC’s latest data (first printed at 5,002 t in January 2026, and valued then at a record $555 bn). Of the 5,028 t that can be attributed to a use (Table 6), investment — bars, coins and ETFs — rose 83% to 2,209 t, overtaking jewellery for the first time since 2020: ETFs took in 803 t and bars and coins reached 1,406 t. Jewellery fell 19% to 1,648 t as record prices cut the volume people could afford. Central banks bought 848 t (Section 2.7), and technology — bonding wire, connectors and a little dentistry — held at 323 t.
Table 6. Gold demand by category, 2025
| Category | Demand | Share | Change vs. 2024 |
|---|---|---|---|
| Jewellery fabrication | 1,648 t | 33% | −19% |
| Bars & coins | 1,406 t | 28% | +16% |
| Central banks | 848 t | 17% | −22% |
| ETFs | 803 t | 16% | from −3 t |
| Technology | 323 t | 6% | −1% |
| Gold demand (excl. OTC) | 5,028 t | 100% | +8% |
Source: World Gold Council, gold supply and demand statistics (Goldhub, data to 30 June 2026), annual series “Jewellery fabrication”, “Total bc”, “Central banks”, “Etfs” and “Technology”, 2024 and 2025 — Metals Focus data compiled by the WGC. Shares and changes are computed here; gold demand excluding OTC is the sum of the five categories. OTC and other flows (145 t in 2025) are a residual and appear in Table 12.
Figure 4. Gold demand by category, 2025
Figure data: Table 6.
Consumer demand — jewellery plus bars and coins — is overwhelmingly an Asian and Middle Eastern story. China and India together account for just over half of it, and both bought less in 2025 as prices rose. Behind them the United States, Turkey, Iran, Russia and Saudi Arabia are the next-largest markets; Western demand leans to bars, coins and ETFs rather than jewellery, and in Germany bar and coin buying more than doubled. The ranking covers consumer demand only — central-bank buying and ETF flows are not tied to a country’s consumers.
Table 7. Largest gold-consuming countries, 2025
| Rank | Country | Jewellery | Bars & coins | Consumer demand | Share of world | Change vs. 2024 |
|---|---|---|---|---|---|---|
| 1 | China (mainland) | 360 t | 432 t | 792 t | 27% | −3% |
| 2 | India | 431 t | 280 t | 711 t | 24% | −11% |
| 3 | United States | 117 t | 58 t | 175 t | 6% | −16% |
| 4 | Turkey | 33 t | 71 t | 104 t | 4% | −32% |
| 5 | Iran | 27 t | 46 t | 73 t | 2% | +5% |
| 6 | Russia | 37 t | 36 t | 72 t | 2% | −2% |
| 7 | Saudi Arabia | 44 t | 18 t | 62 t | 2% | −5% |
| 8 | Thailand | 8 t | 51 t | 59 t | 2% | +21% |
| — | World | 1,542 t | 1,374 t | 2,916 t | 100% | −5% |
Source: World Gold Council, Gold Demand Trends: Q4 and Full Year 2025, “Table 2: Jewellery demand in selected countries” (jewellery consumption) and “Table 3: Total bar and coin demand in selected countries”, 2024 and 2025 columns; consumer demand, shares and changes are computed here from the tonnes as printed (one decimal) and then rounded, so rows may not add exactly. This table uses the January 2026 edition, the only one with a country split; its world jewellery figure (1,542 t) is consumption, slightly below the fabrication figure in Table 6.
Over time the composition of demand has shifted more than its total. Jewellery, the largest use for most of the 2010s, has drifted down as prices rose; investment surges in crises (2020) and in strong bull markets (2025) and collapses when sentiment turns (2013, when ETFs shed 929 t); and since 2022 central banks have taken roughly a fifth of identifiable demand, up from about an eighth in 2011–2019.
Table 8. Gold demand by category, 2010–2025
| Category | 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | 2016 | 2017 |
|---|---|---|---|---|---|---|---|---|
| Jewellery fabrication | 2,089 t | 2,133 t | 2,162 t | 2,769 t | 2,544 t | 2,479 t | 2,023 t | 2,267 t |
| Technology | 467 t | 438 t | 390 t | 363 t | 355 t | 338 t | 329 t | 339 t |
| Investment | 1,653 t | 1,780 t | 1,636 t | 804 t | 899 t | 964 t | 1,622 t | 1,322 t |
| of which bars & coins | 1,246 t | 1,538 t | 1,343 t | 1,733 t | 1,061 t | 1,087 t | 1,079 t | 1,051 t |
| of which ETFs | 407 t | 242 t | 293 t | −929 t | −162 t | −124 t | 543 t | 271 t |
| Central banks | 79 t | 481 t | 569 t | 629 t | 601 t | 580 t | 395 t | 379 t |
| Gold demand (excl. OTC) | 4,288 t | 4,832 t | 4,758 t | 4,564 t | 4,399 t | 4,360 t | 4,369 t | 4,307 t |
| Category | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|---|---|---|---|
| Jewellery fabrication | 2,298 t | 2,162 t | 1,332 t | 2,252 t | 2,208 t | 2,208 t | 2,025 t | 1,648 t |
| Technology | 342 t | 333 t | 309 t | 337 t | 315 t | 305 t | 326 t | 323 t |
| Investment | 1,167 t | 1,282 t | 1,808 t | 1,012 t | 1,136 t | 962 t | 1,205 t | 2,209 t |
| of which bars & coins | 1,097 t | 878 t | 915 t | 1,200 t | 1,244 t | 1,204 t | 1,209 t | 1,406 t |
| of which ETFs | 70 t | 404 t | 893 t | −189 t | −108 t | −242 t | −3 t | 803 t |
| Central banks | 656 t | 605 t | 255 t | 450 t | 1,080 t | 1,051 t | 1,086 t | 848 t |
| Gold demand (excl. OTC) | 4,463 t | 4,382 t | 3,703 t | 4,052 t | 4,739 t | 4,526 t | 4,643 t | 5,028 t |
Source: World Gold Council, gold supply and demand statistics (Goldhub, data to 30 June 2026), annual series — the latest vintage, which revises earlier editions; Metals Focus data. Investment is bars and coins plus ETFs; gold demand excluding OTC is the sum of jewellery fabrication, technology, investment and central banks, computed here from the unrounded series; every cell is rounded once to whole tonnes, so rows may not add exactly. Negative ETF values are net outflows.
Figure 5. Gold demand by category, 2010–2025 (tonnes)
Figure data: Table 8 — investment (bars, coins and ETFs together, which never turns negative in a year), jewellery fabrication, central banks and technology, stacked; OTC flows are left out.
2.3 Supply: producing countries
Annual supply has two sources — newly mined gold and recycled scrap (Section 2.5). On the USGS count, China, Russia, Australia, Canada and the United States led 2025 mine output and together supplied about 41% of it — meaningful concentration, but far from a cartel. A state-owned company mines most of Uzbekistan’s gold (NMMC, Table 21) and state-linked groups are prominent in China, while most Australian, Canadian, US and African output comes from listed companies.
Table 9. Leading gold-mining countries, 2024–2025
| Rank | Country | 2024 | 2025e | Share of world, 2025 | Change |
|---|---|---|---|---|---|
| 1 | China | 377 t | 380 t | 11.5% | +1% |
| 2 | Russia | 310 t 1 | 310 t | 9.4% | 0% |
| 3 | Australia | 284 t | 280 t | 8.5% | −1% |
| 4 | Canada | 200 t 1 | 200 t | 6.1% | 0% |
| 5 | United States | 163 t | 160 t | 4.8% | −2% |
| 6 | Ghana | 149 t | 150 t | 4.5% | +1% |
| 7 | Mexico | 140 t | 140 t | 4.2% | 0% |
| 8 | Kazakhstan | 130 t 1 | 130 t | 3.9% | 0% |
| 9 | Uzbekistan | 129 t | 130 t | 3.9% | +1% |
| 10 | Peru | 108 t | 110 t | 3.3% | +2% |
| — | Rest of world | 1,290 t | 1,310 t | 39.7% | +2% |
| — | World total (rounded) | 3,280 t | 3,300 t | 100% | +1% |
Notes to Table 9
- A USGS estimate for 2024. Every 2025 value is an estimate (“2025e”), made before the year’s data were complete.
Source: USGS, Mineral Commodity Summaries 2026: Gold, “World Mine Production and Reserves”, columns “Mine production 2024” and “2025e”, rows 1–10 and “World total (rounded)", p.91 of the full volume (the chapter link opens pp.90–91). Rest of world = world total less the ten rows shown; shares (country ÷ 3,300) and changes are computed here. The USGS’s own text gives the top-five share as 41% (p.90), from unrounded data; the rounded rows give 40.3%.
Figure 6. Leading gold-mining countries, 2025 (tonnes)
Figure data: Table 9, 2025e column.
Over the long run, world mine output has crept up from about 2,590 t in 2000 to 3,300 t in 2025 — about 1% a year — with a trough of 2,300 t in 2008, a record 3,310 t in 2018 and a pandemic dip in 2020. Growth has flattened as large discoveries have thinned and grades fallen: the “peak gold” debate (Section 5).
Table 10. World gold mine production, 2000–2025
| Year | 2000 | 2001 | 2002 | 2003 | 2004 | 2005 | 2006 | 2007 | 2008 | 2009 | 2010 | 2011 | 2012 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| World | 2,590 t | 2,600 t | 2,550 t | 2,540 t | 2,420 t | 2,470 t | 2,370 t | 2,350 t | 2,300 t | 2,490 t | 2,590 t | 2,680 t | 2,740 t |
| Year | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025e |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| World | 2,960 t | 3,040 t | 3,100 t | 3,180 t | 3,260 t | 3,310 t | 3,260 t | 3,050 t | 3,120 t | 3,230 t | 3,250 t | 3,280 t | 3,300 t |
Source: three USGS publications, each the latest to print the year — 2000–2018: USGS Data Series 140, Gold statistics, sheet “Gold”, column “World production”; 2019–2023: USGS Minerals Yearbook 2023, Gold, advance tables, Table 8, row “Total” (kilograms ÷ 1,000); 2024–2025: USGS MCS 2026 Gold, “World total (rounded)", p.91. The first two share one family (Data Series 140 is compiled from the Minerals Yearbook tables), and the join with MCS 2026 is clean: both print 3,250 t for 2023. 2022 is the Minerals Yearbook 2023 revision (3,230 t, up from 3,160 t). Rounded to three significant figures by the USGS.
Figure 7. World gold mine production, 2000–2025 (tonnes)
Figure data: Table 10.
Reserves — the economically mineable part of a measured or indicated resource, an estimate rather than a measured quantity — tell the longevity story, and here the ranking differs from current output. Australia and Russia hold the largest reserves, followed by South Africa, Indonesia, China and Canada. China mines the most gold but has one of the shortest reserve lives in the table, about eight years at 2025 rates, while South Africa’s deep Witwatersrand ore gives it more than 50. World reserves of about 66,000 t equal roughly 20 years of mine supply at the 2025 rate.
Table 11. Gold reserves by country, 2025
| Country | Reserves | Reserve life | Country | Reserves | Reserve life |
|---|---|---|---|---|---|
| Australia | 13,000 t 1 | 46 years | United States | 3,000 t | 19 years |
| Russia | 12,000 t | 39 years | Brazil | 2,500 t | 31 years |
| South Africa | 5,000 t | 56 years | Kazakhstan | 2,300 t | 18 years |
| Indonesia | 3,600 t | 40 years | Peru | 2,200 t | 20 years |
| China | 3,200 t | 8 years | Uzbekistan | 2,200 t | 17 years |
| Canada | 3,200 t | 16 years | World total (rounded) | 66,000 t | 20 years |
Notes to Table 11
- On the JORC-compliant basis Australia’s reserves are 4,500 t, a reserve life of 16 years; the 13,000 t is the USGS’s broader estimate.
Source: USGS, Mineral Commodity Summaries 2026: Gold, “World Mine Production and Reserves”, column “Reserves” and footnote 11, p.91 of the full volume. Reserve life = reserves ÷ the same page’s 2025e mine production for each country (South Africa about 90 t, Indonesia 90 t and Brazil 80 t are on that page but outside Table 9’s top ten), computed here and rounded to whole years. The USGS revised Australia, Brazil, China, Peru and other countries in this edition.
2.4 The supply–demand balance
Gold’s balance works differently from a consumed commodity’s. Because almost nothing is destroyed, there is no meaningful annual “deficit”: central-bank buying, ETF flows and recycling move metal between holders rather than using it up, and the market clears through price — when buyers want more gold than mines and scrap provide, the price rises until some holders sell. The WGC’s accounts make the point: supply and demand are equal by construction, and the difference between supply and the demand that can be attributed to a use is an OTC and other residual — unobserved investment, inventory changes and statistical error.
Table 12. Gold supply and how the market cleared, 2010–2025
| Category | 2010 | 2011 | 2012 | 2013 | 2014 | 2015 | 2016 | 2017 |
|---|---|---|---|---|---|---|---|---|
| Mine production | 2,754 t | 2,877 t | 2,957 t | 3,178 t | 3,284 t | 3,368 t | 3,564 t | 3,653 t |
| Net producer hedging | −109 t | 23 t | −45 t | −28 t | 105 t | 13 t | 38 t | −26 t |
| Recycled gold | 1,671 t | 1,626 t | 1,637 t | 1,195 t | 1,130 t | 1,067 t | 1,232 t | 1,112 t |
| Total supply | 4,317 t | 4,526 t | 4,549 t | 4,346 t | 4,519 t | 4,448 t | 4,834 t | 4,740 t |
| Gold demand (excl. OTC) | 4,288 t | 4,832 t | 4,758 t | 4,564 t | 4,399 t | 4,360 t | 4,369 t | 4,307 t |
| OTC & other (residual) | 29 t | −306 t | −209 t | −219 t | 120 t | 87 t | 465 t | 433 t |
| Category | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 |
|---|---|---|---|---|---|---|---|---|
| Mine production | 3,726 t | 3,666 t | 3,502 t | 3,622 t | 3,710 t | 3,708 t | 3,751 t | 3,823 t |
| Net producer hedging | −12 t | 6 t | −37 t | 12 t | −9 t | 67 t | −56 t | −54 t |
| Recycled gold | 1,132 t | 1,276 t | 1,293 t | 1,136 t | 1,136 t | 1,234 t | 1,365 t | 1,404 t |
| Total supply | 4,847 t | 4,948 t | 4,759 t | 4,770 t | 4,837 t | 5,008 t | 5,060 t | 5,173 t |
| Gold demand (excl. OTC) | 4,463 t | 4,382 t | 3,703 t | 4,052 t | 4,739 t | 4,526 t | 4,643 t | 5,028 t |
| OTC & other (residual) | 384 t | 566 t | 1,055 t | 718 t | 98 t | 482 t | 417 t | 145 t |
Source: World Gold Council, gold supply and demand statistics (Goldhub, data to 30 June 2026), annual series “Mine production”, “Net producer hedging” and “Recycled gold”; total supply and the OTC residual are computed here with the WGC’s identities (total supply = mine production + net hedging + recycling = total demand; OTC = total supply − gold demand excluding OTC, from Table 8), computed from the unrounded series; every cell is rounded once to whole tonnes, so rows may not add exactly. Metals Focus’s mine production includes artisanal estimates and was revised up in 2026 — the January 2026 edition printed 3,672 t for 2025 — so it runs above the USGS series in Table 10.
At the country level the useful split is between net suppliers — Australia, Russia, Canada and the other large miners, whose own consumers buy only a fraction of what they mine — and net buyers, led by India, which mines almost no gold, and China, which is the largest producer and by far the largest consumer. Trade then fills the gap (Section 2.6).
Table 13. Mine production against consumer demand, 2025
| Country | Mine production, 2025e | Consumer demand | Production less demand |
|---|---|---|---|
| Australia | 280 t | 22 t | +258 t |
| Russia | 310 t | 72 t | +238 t |
| Canada | 200 t | 23 t | +177 t |
| Mexico | 140 t | 14 t | +126 t |
| Brazil | 80 t | 17 t | +63 t |
| Indonesia | 90 t | 48 t | +42 t |
| United States | 160 t | 175 t | −15 t |
| China | 380 t | 792 t | −412 t |
| India | n/d 1 | 711 t | ~−711 t |
Notes to Table 13
- The USGS does not list India, whose small mine output sits in “Other countries”; its position is shown as consumer demand alone.
Source: mine production from USGS MCS 2026: Gold, “2025e” column, each country’s row on p.91 of the full volume; consumer demand (jewellery plus bars and coins) from the WGC’s Gold Demand Trends FY2025, each country’s row in Tables 2 and 3. Tables 7 and 9 show only the largest countries, so Australia, Canada, Mexico, Brazil and Indonesia are read from those source rows directly. Production less demand is computed here. It ignores ETF, central-bank and technology demand and the gold that recycling returns, so it shows where newly mined metal must travel, not each country’s full trade balance (Tables 16–17).
Figure 8. Mine production less consumer demand, 2025 (tonnes)
Figure data: Table 13.
Stocks, not shortages. The right way to read gold’s balance is through the stock. About 222,600 t of gold existed above ground at mid-2026 — 58 years of 2025 mine output on the WGC’s count, a stock-to-flow ratio no other commodity approaches. That is why no plausible change in mining can flood or starve the market, and why the price is set by the willingness of existing holders to buy or sell. The visible, mobile part of the stock is what moves the price in the short run: London’s vaults alone hold nearly two years of total annual demand (Table 14).
Table 14. Gold stocks and the years of 2025 mine output they represent
| Stock | Quantity | As of | Years of output | Source |
|---|---|---|---|---|
| Above-ground stock, total | 222,600 t | Jun 2026 | 58 | Filed | WGC above-ground stock | “Total above-ground stock (end-Q2 2026)” 222,600 tonnes |
| Jewellery | ~99,700 t | Jun 2026 | 26 | Filed | WGC above-ground stock | “Jewellery” ~99,700t, 45% |
| Bars and coins | ~47,800 t | Jun 2026 | 13 | Filed | WGC above-ground stock | “Bars and coins” ~47,800t, 21% |
| Central banks | ~39,000 t | Jun 2026 | 10 | Filed | WGC above-ground stock | “Central banks” ~39,000t, 18% |
| Gold ETFs | ~4,000 t | Jun 2026 | 1.0 | Filed | WGC above-ground stock | “Gold ETFs” ~4,000t, 2% |
| London vaults | 9,632 t | Aug 2026 | 2.5 | Filed | LBMA London vault data | “the amount of gold held in London vaults was” 9,632 tonnes |
| Global gold ETF holdings | 4,029 t | Dec 2025 | 1.1 | Market | WGC gold ETF holdings | monthly, regional tonnes summed | Dec 2025 |
| Mine reserves (below ground) | 66,000 t | 2025 | 20 1 | Filed | USGS MCS 2026 | “World total (rounded)” reserves | p.91 |
Notes to Table 14
- Years of USGS mine output (3,300 t); the other rows use the WGC’s 3,823 t, the flow that matches the WGC’s stock estimate. The WGC itself puts reserves at about 61,000 t.
Source: WGC, Above-ground stock (data as of end-June 2026, rounded to the nearest 100 t), LBMA, London Vault Data (end-August 2026 data; end-December 2025: 9,106 t, from the monthly workbook), WGC, Gold ETFs holdings and flows (data to 25 September 2026; the four regions’ end-December 2025 tonnes summed here) and USGS MCS 2026: Gold (p.91 of the full volume) — each row as cited. Years of output = tonnes ÷ 2025 mine production, computed here. London’s vault total includes Bank of England holdings and overlaps the ETF and central-bank rows; the rows are not additive beyond the four holder categories.
Figure 9. Who holds the gold above ground, end-June 2026
Figure data: WGC Above-ground stock, the six holder categories as printed (jewellery, bars and coins, central banks, other, OTC, gold ETFs), shares of 222,600 t; Table 14 lists four of them — jewellery, bars and coins, central banks and gold ETFs.
2.5 Supply structure: mine, recycling & by-product
Gold supply is overwhelmingly mined, with a large, price-sensitive recycling layer no consumed commodity has. In 2025 mines supplied 73% of the 5,173 t total (net of producers’ hedging) and recycled jewellery and electronics 27%. Recycling is the elastic part — but less elastic than the price suggests: with gold up 44% in 2025, recycling rose only 3% to 1,404 t, below the 1,671 t of 2010, when it made up 39% of supply. Holders who expect higher prices keep their gold; households sell in distress or when prices spike suddenly.
Table 15. Gold supply mix, 2025
| Supply source | Volume | Share of supply | Price elasticity |
|---|---|---|---|
| Mine production | 3,823 t | 74% | Low — years from discovery to output |
| Net producer hedging | −54 t | −1% | — |
| Recycled gold | 1,404 t | 27% | High, but damped when prices are expected to rise |
| Total supply | 5,173 t | 100% | — |
Source: World Gold Council, gold supply and demand statistics (Goldhub, data to 30 June 2026), 2025 values of the series in Table 12; shares computed here. The Price elasticity column is this guide’s judgement, not a WGC field.
Figure 10. Gold supply mix, 2025
Figure data: Table 15 — mine supply shown net of producer hedging (3,769 t).
By-product gold is the part of mine supply that does not answer to the gold price at all. No agency publishes a world share. In the United States about 7% of domestic gold comes from base-metal ores, chiefly copper (USGS MCS 2026, p.90), and the largest by-product sources are the porphyry copper-gold mines: Grasberg alone produced 937 koz in 2025 (Table 5), and Newmont’s Cadia — an alkalic porphyry gold-copper deposit — 385 koz. When those mines cut output, as Grasberg did after its 2025 mud rush, gold supply falls whatever the gold price. (By-product economics matter most for copper and silver, where they shape the cost curve.)
2.6 Trade flows: the London–Zurich–Asia axis
Gold’s trade map looks nothing like oil’s. Worth roughly $110 m a tonne at 2025 prices, gold travels in aircraft holds, not tankers, so there are no maritime choke points. The vulnerability is concentration in a few mid-chain hubs. Switzerland’s refiners process about a third of the world’s gold, and five of the 67 refiners on the LBMA’s Good Delivery List for gold are Swiss; London’s vaults hold the metal that backs the OTC market; and Hong Kong and Dubai funnel bullion to China, India and the Gulf. In value terms Switzerland was both the largest exporter and the largest importer in 2025.
Table 16. Largest gold exporters, 2025
| Rank | Country | Exports | Exports | Net trade |
|---|---|---|---|---|
| 1 | Switzerland | $195.8 bn | 1,815 t 1 | −$15.3 bn |
| 2 | United States | $80.6 bn | 777 t 1 | +$50.4 bn 2 |
| 3 | Hong Kong | $80.4 bn | 739 t 1 | −$21.4 bn |
| 4 | United Kingdom | $79.8 bn | 812 t | −$75.1 bn |
| 5 | Australia | $38.4 bn | n/d | +$25.1 bn |
| 6 | Canada | $38.1 bn | 349 t | +$21.8 bn |
| 7 | China | $25.3 bn | n/d | −$76.9 bn |
| 8 | Japan | $25.3 bn | n/d | +$24.1 bn |
Source: UN Comtrade, HS 7108, annual 2025 — each reporter’s exports to the world (flowCode=X); net trade uses the imports in Table 17. Notes and method are under Table 17.
Table 17. Largest gold importers, 2025
| Rank | Country | Imports | Imports | Net trade |
|---|---|---|---|---|
| 1 | Switzerland | $211.1 bn | n/d | −$15.3 bn |
| 2 | United Kingdom | $154.9 bn | 1,376 t | −$75.1 bn |
| 3 | China | $102.2 bn | n/d | −$76.9 bn |
| 4 | Hong Kong | $101.8 bn | 942 t | −$21.4 bn |
| 5 | India | $58.9 bn | 640 t 1 | −$58.0 bn |
| 6 | United States | $30.2 bn | 292 t 1 | +$50.4 bn 2 |
| 7 | Turkey | $23.1 bn | 224 t 3 | −$20.1 bn |
| 8 | Thailand | $20.9 bn | n/d | −$7.9 bn |
Notes to Tables 16 and 17
- Weight estimated by UN Comtrade where the reporter gave none; n/d = no weight reported or estimated.
- US-reported imports miss most of 2025’s inflow: partners reported exports to the United States of 1,046 t ($101.3 bn), against 433 t in 2024, much of it re-exported later in the year — so the US surplus overstates its position.
- Turkey filed no net weight; 224 t is its reported quantity (kilograms ÷ 1,000).
Source: UN Comtrade, HS heading 7108 (gold, unwrought, semi-manufactured or powder, non-monetary), annual 2025, each reporter’s trade with the world — exports (flowCode=X, FOB) and imports (flowCode=M, CIF), public API. Values are primaryValue ÷ 10⁹; tonnes are net weight in kilograms ÷ 1,000. Net trade = exports − imports, computed here. The United Arab Emirates, a major hub, has filed no data for 2024 or 2025 and does not appear; its partner-reported (mirror) flows are in Table 18. China reports values but no weights.
Table 18. Gold’s trading and refining hubs
| Hub | Role | Measure | Value | Source |
|---|---|---|---|---|
| London | OTC trading, clearing and vaulting | Gold in London vaults, end-Aug 2026 | 9,632 t | Filed | LBMA London vault data | “the amount of gold held in London vaults was” 9,632 tonnes |
| Switzerland | Refining | Share of the world’s gold refined | about one-third | Filed | ASMP | The Swiss precious metals industry | “1/3 of the world’s gold is refined in Switzerland” |
| Switzerland | Refining | Gold refiners on the LBMA Good Delivery List | 5 of 67 | Filed | LBMA Good Delivery List – Gold | “67 Gold refiners”; five with Swiss addresses |
| Hong Kong | Gateway to mainland China | Exports to mainland China, 2025 | 425 t ($46.3 bn) | Market | UN Comtrade | HS 7108 | Hong Kong exports by partner | 2025 |
| UAE (Dubai) | Gulf and African entrepôt | Partners’ imports from the UAE, 2025 | 1,043 t ($115.4 bn) | Market | UN Comtrade | HS 7108 | partner-reported imports from the UAE | 2025 |
| New York (COMEX) | Futures delivery | Partners’ exports to the US, 2025 vs. 2024 | 1,046 t vs. 433 t | Market | UN Comtrade | HS 7108 | partner-reported exports to the US | 2024–2025 |
Source: LBMA, London Vault Data (end-August 2026); the Swiss Precious Metals Association (ASMP), The Swiss precious metals industry; the LBMA Good Delivery List – Gold; and UN Comtrade HS 7108 by partner (mirror data where the hub does not report). The ASMP’s one-third is an industry figure; the Swiss Federal Council put Swiss refineries at about 40% of world capacity in 2018, citing the WGC.
The directional flow is consistent. Doré and mined bars leave Africa, Latin America, Australia and Canada for refiners in Switzerland, London’s market and the Gulf; refined kilobars then move east to China, India, Turkey and Thailand, through Hong Kong and Dubai. In 2025 the flow also ran through New York: metal shipped into the United States (Table 18) went back out to Switzerland and London later in the year — a round trip within one year that shows how quickly the hubs can rebalance.
Figure 11. Where gold travels: producers, hubs and consumers, 2025 ($ bn)
- Australia38
- Canada38
- Ghana20
- Switzerland196
- UAE115
- Hong Kong80
- United Kingdom80
- China102
- India59
- Turkey23
- Thailand21
Source: UN Comtrade, HS 7108, 2025 — producers’ and hubs’ exports and consumers’ imports to the world, $ bn (Tables 16–17; Ghana, outside those tables, exported $20.2 bn); the UAE’s value is partners’ imports from it, since it does not report. Links mark the main flows between these nodes, each at least $4 bn in the by-partner data (for example, Swiss imports from Ghana of $4.6 bn and Swiss exports to India of $13.4 bn on the highlighted route); they show topology, not volume.
2.7 Market organisations & the official sector
Gold has no OPEC. Production is spread across dozens of countries and hundreds of companies, and the vast above-ground stock means a coordinated mine cut could be offset by holders selling. Two very different kinds of body shape the market instead.
First, the market organisers: the LBMA (London Bullion Market Association) sets the Good Delivery standards, accredits refiners and oversees the OTC market; ICE Benchmark Administration runs the LBMA Gold Price auction, with fifteen direct participants; CME Group runs COMEX futures; and the Shanghai Gold Exchange runs China’s physical market. They set the rules of trading, not the price level.
Second — and far more important for the price in the 2020s — the official sector. Central banks and governments hold about 39,000 t, some 18% of all the gold above ground. Their holdings have moved through two eras. Through the 2000s the official sector was a net seller: European central banks sold under the Central Bank Gold Agreement of 1999, which capped sales at about 400 t a year, and the reported holdings of the 93 countries that report every year fell from 27,864 t in 2000 to a trough of 25,020 t in 2007. The trend then reversed. Led by emerging-market central banks diversifying away from the dollar, the official sector became a sustained net buyer, and those 93 countries held 31,435 t by 2025. The United States, Germany, Italy, France, China and Russia hold the largest reserves.
Table 19. Largest official gold holders, 31 March 2026
| Rank | Country | Gold reserves | Share of total reserves |
|---|---|---|---|
| 1 | United States | 8,133 t | 83% |
| 2 | Germany | 3,350 t | 83% |
| 3 | Italy | 2,452 t | 80% |
| 4 | France | 2,437 t | 81% |
| 5 | China | 2,313 t | 9% |
| 6 | Russia | 2,305 t | n/d |
| 7 | Switzerland | 1,040 t | 14% |
| 8 | India | 881 t | 18% |
| 9 | Japan | 846 t | 9% |
| 10 | Netherlands | 612 t | 73% |
| 11 | Poland | 582 t | 30% |
| 12 | Turkey | 535 t | 62% |
Source: World Gold Council, Gold reserves by country (Central Banks Dashboard, IMF International Financial Statistics, September 2026 edition), quarterly table at 31 March 2026, columns “Gold Reserves Tonnes” and “Holdings %" — the latest date with every holder reported. Share of total reserves is gold’s share of each country’s official reserve assets at market value; Russia’s is not yet reported (n/d). Countries only; the IMF (about 2,800 t), the ECB and the BIS also hold gold.
Figure 12. Official gold holdings, 2000–2025 (tonnes)
Source: World Gold Council, Gold reserves by country (IMF IFS data), end-year gold reserves in tonnes, summed here for the 93 countries that report a holding in every year from 2000 to 2025 — a constant sample, so a country starting or stopping reporting does not move the line. It excludes the IMF, the ECB, the BIS and countries with gaps (Uzbekistan before 2013, Venezuela after 2017), and so sits below the WGC’s ≈39,000 t official total.
Central-bank buying has become the market’s swing demand: it ran between 255 t and 656 t a year from 2011 to 2021, then above 1,000 t in each of 2022, 2023 and 2024, and 848 t in 2025 (Table 8). The buyers are mostly emerging-market central banks; more than half of 2025’s buying was unreported to the IMF and is estimated by the WGC and Metals Focus.
Table 20. Largest central-bank buyers and sellers, 2025
| Central bank | Net purchases | Central bank | Net purchases |
|---|---|---|---|
| Poland | 102.0 t | Turkey | 26.7 t |
| Kazakhstan | 57.0 t | Czech Republic | 20.4 t |
| Brazil | 42.8 t | Iraq | 12.0 t |
| Azerbaijan (SOFAZ) | 38.2 t | Singapore | −15.1 t |
| China | 26.7 t | Russia | −6.2 t |
Source: World Gold Council, Gold Demand Trends: Q4 and Full Year 2025, Central banks section, chart “Reported central bank net purchases and sales”, series “Net purchases” and “Net sales”, data to 31 December 2025 where available. Reported changes only; the WGC says unreported buying made up 57% of the year’s total.
Figure 13. Central-bank net gold purchases, 2011–2025 (tonnes)
Figure data: Table 8, Central banks row, 2011–2025; the chart starts in 2011, the first year of sustained buying (2010: 79 t).
Current situation (2026, unresolved — for reference only): central-bank buying slowed to 57 t in the first quarter of 2026 — after the WGC moved 187 t of first-print demand into OTC — and recovered to 289 t in the second, as Turkey (83 t) and Russia (44 t) sold and Poland, Uzbekistan and China bought (WGC, Gold Demand Trends Q2 2026).
3. The companies & the value chain
3.1 The largest gold companies
Gold mining is fragmented: the largest producer mines about 5% of world output. The way to size producers in an evergreen guide is by durable fundamentals — production, reserves, reserve life and cost — not market capitalisation or share price, which move daily and date a report instantly. Newmont, enlarged by its 2023 purchase of Newcrest, is the clear leader; Agnico Eagle, Barrick, Uzbekistan’s state-owned NMMC and AngloGold Ashanti follow at about 3 Moz each; China’s Zijin Mining — also a top-tier copper miner — Russia’s Polyus, Gold Fields and Kinross complete the top nine. Their filings count ounces differently — attributable, payable, consolidated or gold-equivalent — so small gaps between them mean little; the notes say which is which. To screen the whole universe on production, reserves and cost, use the Metal Pilot gold screener, or the best gold stocks for five listed producers ranked on 2025 output.
Table 21. Largest gold producers by output and reserves, 2025
| Company | Country | Type | Listing | Production, 2025 | Gold reserves | Reserve life | AISC, 2025 | Source |
|---|---|---|---|---|---|---|---|---|
| Newmont | United States | Major (Au, Cu) | Public (NYSE:NEM) | 5.89 Moz 1 | 118.2 Moz | 20 yrs | $1,358/oz 2 | Filed | NEM 10-K FY2025 | Item 1 | “Attributable gold ounces produced (thousands)” 5,889 | p.7 Filed | NEM 10-K FY2025 | Item 2 | “attributable proven and probable gold reserves of” 118.2 Moz | p.71 Filed | NEM Q4 2025 release | Summary of Results | “Gold By-Product AISC ($/oz)” 1,358 | p.3 |
| Agnico Eagle | Canada | Major (Au) | Public (TSX, NYSE: AEM) | 3.45 Moz 3 | 55.4 Moz | 16 yrs | $1,339/oz | Filed | AEM MD&A FY2025 | Revenues | “Payable production: Gold (ounces)” 3,447,367 | p.9 Filed | AEM YE2025 reserves release | “proven and probable mineral reserve estimate totalled” 55.4 Moz | p.3 Filed | AEM MD&A FY2025 | “All-in sustaining costs per ounce on a by-product basis for the full year 2025 were” $1,339 | p.29 |
| Barrick | Canada | Major (Au, Cu) | Public (TSX: ABX; NYSE: B) | 3.26 Moz 4 | 85 Moz | 26 yrs | $1,637/oz | Filed | ABX MD&A FY2025 | Highlights | “Gold production (thousands of ounces)” 3,255 | p.6 Filed | ABX AIF FY2025 | Gold Reserves | “Barrick’s total proven and probable gold reserves were” 85 Moz | p.32 Filed | ABX MD&A FY2025 | Highlights | “Gold AISC” 1,637 | p.6 |
| NMMC | Uzbekistan | State producer (Au) | State-owned | 3.15 Moz | 64 Moz 5 | 20 yrs | $1,358/oz | Filed | NMMC presentation | Company overview | “Gold produced FY25” 3.15 Moz | p.5 Filed | NMMC presentation | Reserves and Resources | “MINERAL RESERVES” 64 Moz | p.6 Filed | NMMC presentation | “AISC FY25” 1,358 $/oz | p.5 Filed | NMMC affiliated persons | “Owns 100% of the Company’s shares” | p.1 |
| AngloGold Ashanti | United Kingdom | Major (Au) | Public (NYSE: AU; JSE, A2X, GSE) | 3.09 Moz 6 | 36.5 Moz | 12 yrs | $1,709/oz | Filed | AU 20-F FY2025 | Item 5 | "…to 3.09 million ounces in 2025” | p.148 Filed | AU 20-F FY2025 | Mineral Reserve | "…to 36.5Moz at 31 December 2025” | p.131 Filed | AU 20-F FY2025 | “Group AISC of $1,709/oz” | p.200 |
| Zijin Mining | China | Diversified (Cu, Au) | Public (SSE: 601899; HKEX: 2899) 7 | 2.88 Moz 7 | 64.2 Moz | 22 yrs | n/d | Filed | Zijin FY2025 results | Gold | “produced 89,544kg (2,878,919 ounces) of mine-produced gold” | p.49 Filed | Zijin FY2025 results | Resources and reserves | “Gold/t (metal)” 1,996.25 t | p.35 |
| Polyus | Russia | Major (Au) | Public (MOEX: PLZL) | 2.53 Moz | 105.2 Moz 8 | 42 yrs | n/d 8 | Filed | Polyus FY2025 MD&A | “Total gold output, koz” 2,528.8 | p.4 Filed | Polyus reserves page | “Polyus Ore Reserve Estimate as at 31 December 2024” 105.2 Moz Filed | Polyus FS 2025 | Note 1 | "’level one’ listed on the Moscow Exchange” | p.13 |
| Gold Fields | South Africa | Major (Au) | Public (JSE, NYSE: GFI) | 2.44 Moz 9 | 48.3 Moz | 20 yrs | $1,645/oz | Filed | GFI FY2025 results | “Group attributable gold-equivalent production increased 18% to” 2,438koz | p.1 Filed | GFI FY2025 results | “Group gold Reserves increased by 4.0Moz or 9% to” 48.3 Moz | p.4 Filed | GFI FY2025 results | “All-in sustaining costs” $1,645/oz | p.1 |
| Kinross | Canada | Major (Au) | Public (TSX: K; NYSE: KGC) | 2.01 Moz 9 | 20.9 Moz | 10 yrs | $1,539/oz | Filed | K MD&A FY2025 | Highlights | “Attributable gold equivalent ounces … Produced” 2,012,106 | p.3 Filed | K AIF FY2025 | “proven and probable mineral reserves as at December 31, 2025, was” 20.9 Moz | p.8 Filed | K MD&A FY2025 | “Attributable all-in sustaining cost per ounce sold on a by-product basis” $1,539 | p.3 |
Notes to Table 21
- Attributable, including Newmont’s 38.5% of Nevada Gold Mines and its equity-accounted shares of Pueblo Viejo (40%) and Fruta del Norte; Newmont’s reserves include 50% of three undeveloped copper-gold projects (about 21 Moz).
- By-product basis, from Newmont’s results release; the 10-K prints co-product AISC of $1,609/oz.
- Payable production.
- Attributable production, including Barrick’s 61.5% of Nevada Gold Mines and 60% of Pueblo Viejo.
- As at 1 January 2025, from a company presentation that is not an audited filing; NMMC publishes no annual report in English.
- Consolidated: Sukari (50% owned) and Siguiri (85%) are counted in full; reserves are attributable.
- Subsidiaries at 100%; on an attributable basis 75,144 kg (2.42 Moz) and reserves of 1,637.91 t (52.7 Moz), under China’s resource standard (GB/T 17766-2020). Zijin’s largest shareholder, Minxi Xinghang, is state-owned. Zijin reports a unit cost of sales, not AISC.
- JORC reserves at 31 December 2024 on a 100% basis, the latest Polyus has published; Polyus reports total cash costs ($739/oz in 2025), not AISC.
- Gold-equivalent ounces, attributable.
Source: company filings for the year to 31 December 2025, each row as cited — Newmont 2025 Form 10-K and Q4 2025 results, Agnico Eagle 2025 MD&A and year-end reserves release, Barrick 2025 MD&A and AIF, the NMMC corporate presentation and list of affiliated persons, AngloGold Ashanti 2025 Form 20-F, Zijin Mining 2025 annual results (HKEX, 22 March 2026), Polyus FY2025 results, financial statements and reserves page, Gold Fields 2025 results (Form 6-K, 20 February 2026) and Kinross 2025 MD&A and AIF; printed page numbers where the document is paginated. Reserve life = reserves ÷ production, computed here. AISC is each company’s own non-GAAP measure, by-product basis where both are printed. Listing gives the exchanges and tickers; Polyus discloses neither its controlling shareholder nor its ownership of each mine, under a Russian government decree, and its auditor qualified its 2025 opinion on that point (Polyus FS 2025, p.4). No market-capitalisation figures are shown by design.
Figure 14. Largest gold producers, 2025 (Moz)
Figure data: Table 21, production column.
3.2 Company archetypes along the value chain
Gold exposure is not one trade: the business models sit at very different points on the risk/return spectrum, and an investor should match the archetype to the goal. Explorers look for deposits — pure option value, no cash flow, binary outcomes. Developers permit and build mines — high capital risk, with value unlocked at first pour. Producers mine and sell gold — direct exposure to the price, geared by operating leverage (largely fixed costs against a moving gold price), with margins set by their position on the cost curve. Royalty and streaming companies finance miners in exchange for a share of future output or revenue — gold-price upside with capped cost exposure and diversification across many mines; the whole peer group is screened on the Metal Pilot royalties & streaming dataset, the five largest are compared side by side in Precious Metals Royalty Companies Compared, and they are ranked on attributable output in Best Gold Royalty Stocks by GEOs. Physical bullion and ETFs track the metal itself, with no operating risk. Among the companies analysed on this blog are the explorer GoldMining, the developers NexGold Mining and Seabridge Gold, the producers Aris Mining, Artemis Gold, B2Gold, Endeavour Mining, Equinox Gold, Evolution Mining and Northern Star, and the royalty companies Elemental Royalty, Gold Royalty and Vox Royalty.
Table 22. Gold company archetypes
| Archetype | What they do | Revenue model | Price sensitivity |
|---|---|---|---|
| Explorer | Search for deposits | None (raise and spend) | Very high (sentiment) |
| Developer | Permit and build mines | None until production | High |
| Producer | Mine and sell gold | Gold sales less costs | High (operating leverage) |
| Royalty / streaming | Finance miners for a share of output | Royalty and stream income | Medium (costs capped) |
| Bullion / ETF | Hold physical gold | None (price only) | One-for-one with gold |
Source: this guide — a business-model taxonomy; the price-sensitivity column is a qualitative judgement, not a measured figure.
3.3 Infrastructure & balance-sheet assets
What a gold company owns — and how it is measured — determines what its filings are telling you. A producer’s balance sheet rests on its mineral reserves and resources (in Moz, valued through the net present value of the mine plan at a discount rate), its mines and processing plants (open pits, underground mines, mills and leach pads, measured by capacity and throughput) and its land (exploration upside). The number that most often misleads is gross versus net: production and reserves are often quoted on a 100% basis when the company owns only part of a joint venture, so the attributable figure is the one that flows to shareholders — Nevada Gold Mines’ 2.6 Moz is 1.6 Moz to Barrick and 1.0 Moz to Newmont (Table 5). Reserves also depend on the gold price assumed: the majors in Table 21 used between $1,500 and $2,000/oz at end-2025, far below the 2025 average.
Table 23. Gold-company asset types and metrics
| Asset type | What it does | Key metric | Unit |
|---|---|---|---|
| Reserves and resources | The in-ground ounce base | Proven and probable reserves; grade | Moz; g/t |
| Mines (open pit, underground) | Extract ore | Throughput; mine life | t/yr; years |
| Processing plant / mill | Liberate gold from ore | Capacity; recovery | t/yr; % |
| Royalty and stream interests | A claim on others’ output | Attributable ounces | koz/yr |
| Tenements / exploration land | Future discovery | Area; drill results | km²; g/t |
Source: this guide — a summary of what gold-company filings report; reserve definitions follow SEC Regulation S-K subpart 1300 as restated in Newmont’s 2025 Form 10-K, Item 2 (“the economically mineable part of a measured or indicated mineral resource”, p.70); reserve price assumptions from the filings cited in Table 21 (Newmont $2,000, Barrick $1,500, Agnico Eagle mostly $1,600, Kinross $2,000).
4. Investing in gold
4.1 Ways to own it — exposure vehicles
Before any screening framework comes the practical question: what do you actually buy? Gold offers more routes than any other commodity, and they differ more in structure than in exposure. Bars and coins are the direct route: you pay a dealer premium over spot (highest on small coins) and carry storage and insurance yourself. At larger size, vaulted metal splits into allocated holdings — specific numbered bars held in your name, yours even if the custodian fails — and unallocated holdings, a claim on a bank’s general gold pool: cheaper, but you are an unsecured creditor. Bullion-backed ETFs hold allocated bars against shares and track spot closely for an annual fee — the default vehicle for most investors since SPDR Gold Shares listed in 2004. Futures offer direct, leveraged exposure with margin calls and roll mechanics. Miners and royalty and streaming companies are not gold but businesses geared to it (Sections 3.2 and 4.2).
One structural caveat applies across commodities. A fund that holds futures rather than metal must roll its contracts as they expire, and in contango — gold’s normal state — each roll sells the cheaper expiring contract and buys a dearer later one, a recurring cost that compounds into a lasting gap below spot. Gold is the fortunate exception among major commodities: its main ETFs hold physical bullion, so the drag does not apply — but check the structure before assuming it, because a futures-based gold fund carries it. (The oil guide shows what the drag costs in practice.)
Table 24. Ways to own gold
| Vehicle | What you own | Tracks spot? | Key structural cost or risk | Best suited for |
|---|---|---|---|---|
| Bars and coins | Physical metal in hand | Yes, less the premium | Dealer premium, storage, insurance | Long-term store of value |
| Allocated vault account | Specific numbered bars | Yes | Storage fee | Larger long-term holdings |
| Unallocated account | A bank’s promise to deliver | Yes | Counterparty risk (unsecured creditor) | Trading convenience |
| Bullion-backed ETF | Shares in a trust holding bars | Yes, closely | Annual fee; no possession | Most investors’ default |
| Futures | An exchange contract (100 oz) | Yes, plus basis and roll | Leverage, margin calls, roll mechanics | Professionals, hedgers |
| Miner equities | A business geared to gold | No — amplifies | Operating and jurisdiction risk | Leverage to the price |
| Royalty / streaming | Financing claims on many mines | No — smoother | Deal risk, diversified | Lower-risk equity exposure |
Source: contract size from CME Group, COMEX Rulebook chapter 113 (Rule 113101); the ETF listing date from the SPDR Gold Trust 2025 Form 10-K (“initial public offering on November 18, 2004”); the other columns are this guide’s summary of how each vehicle works — structural attributes only, no fees or prices, which change.
4.2 How to value & screen gold miners
The facts above turn into a repeatable checklist. For a producer, the metrics that matter are the reserve and resource base (how many ounces, at what grade), the all-in sustaining cost (AISC) per ounce — the industry’s headline cost measure, covering operating costs, sustaining capital and overheads — reserve life, and the all-in margin (realised price less AISC). For a developer, it is the project’s net present value and internal rate of return at a conservative gold price, the capital cost and the path to permits. For an explorer, it is grade, drill results and management.
The most useful single tool is the cost curve: rank production from cheapest to dearest and draw the long-run average price across it, and you can see who earns a margin and who survives a downturn. Figure 15 does this for the seven of the nine largest producers that publish AISC: in 2025 their costs ran from $1,339 to $1,709/oz, all below the 2016–2025 average price of $1,831 and less than half the 2025 average of $3,442 — margins that would shrink fast if the price returned to its ten-year average. The same screen — reserves, AISC, reserve life — runs across every listed producer in the gold screener linked in Section 3.1.
Table 25. Gold-miner screening metrics
| Metric | What it tells you | Healthy vs. concerning | Where to find it |
|---|---|---|---|
| Reserves and resources (Moz) | Scale and longevity | Larger and higher-grade is better | Reserve statement |
| AISC ($/oz) | Cost competitiveness | Well below the long-run average price (Table 3) vs. near or above it | Annual report or MD&A |
| Reserve life (years) | Runway before replacement | Over 10 years comfortable; under 7 a worry | Reserves ÷ production |
| All-in margin ($/oz) | Profit per ounce | Wider is better | Realised price − AISC |
| Grade (g/t) | Ore quality | Higher means lower cost per ounce | Reserve statement |
| Reserve price assumption ($/oz) | How conservative the reserves are | Well below the current price | Reserve statement |
Source: this guide — the screening framework; the healthy and concerning ranges are judgements, not industry standards. The metrics are defined as the producers report them (e.g. AISC as a non-GAAP measure in Newmont’s 2025 Form 10-K, Results and Highlights, p.4).
Figure 15. Gold cost curve of seven major producers, 2025 (AISC, $/oz)
Figure data: Table 21 — from left to right Agnico Eagle, Newmont, NMMC, Kinross, Barrick, Gold Fields and AngloGold Ashanti, each bar’s width its share of the seven’s 23.3 Moz of 2025 production and its height its 2025 AISC (by-product basis where printed). The dashed line is the 2016–2025 average gold price, $1,831/oz (Table 3); the 2025 average, $3,442, sits far above the chart. Zijin and Polyus publish no AISC and are left out. AISC definitions differ by company, so the curve ranks broad cost positions, not exact ones.
4.3 Macro regimes, rates & correlations
Gold’s behaviour across the economic cycle is the heart of the investment case, and it runs through one channel: gold pays no yield, so its appeal rises and falls with the opportunity cost of holding it — chiefly the real (inflation-adjusted) interest rate — and with the US dollar it is priced in. (For how the major commodities compare across the same regimes, see the macro regime guide.)
Each rating in Table 26 rests on a concluded episode. Falling real rates: the US 10-year real yield fell from 0.40% in 2019 to −0.60% in 2020 and gold’s average price rose 27%. Rising real rates: from −0.48% in 2012 to 0.45% in 2015 the yield rose, the dollar index rose 18%, and gold fell 30%. Inflation alone is not enough: in 2022 US consumer prices rose 8% but the real yield jumped from −0.91% to 0.43%, and gold’s average price was flat. Crises: from the August 2008 average to the March 2009 average gold rose 10% while the S&P 500 fell 41%. Stagflation — high inflation with weak growth — is the regime it has loved most: from 1971 to 1980 its annual average rose from $41 to $608. The great exception is 2022–2025, when the real yield climbed to about 2% and gold nearly doubled anyway, because central-bank buying and demand for assets outside the dollar system overrode the rate signal — a reminder that these are tendencies, not laws.
Table 26. Gold across economic regimes
| Regime | Typical gold performance | Why | Concluded example | Source |
|---|---|---|---|---|
| High inflation with negative real rates | Strong | Store of value, cash losing value | 1971–1980: $41 to $608/oz (annual averages) | Market | World Bank | “Gold”, annual | 1971, 1980 |
| Falling real rates, rate cuts | Strong | Lower opportunity cost | 2019–2020: real yield 0.40% to −0.60%; gold +27% | Market | FRED | “DFII10”, annual means | 2019–2020; gold from Table 3 |
| Recession / risk-off | Strong | Safe-haven demand | Aug 2008 – Mar 2009 (monthly averages): gold +10%, S&P 500 −41% | Market | World Bank | “Gold”, monthly; Shiller | S&P composite, monthly | Aug 2008, Mar 2009 |
| Inflation with rising real rates | Flat to weak | The rate effect dominates | 2022: CPI +8%, real yield −0.91% to 0.43%; gold +0.1% | Market | FRED | “CPIAUCSL”, “DFII10”, annual means | 2021–2022; gold from Table 3 |
| Strong growth, rising real rates, strong dollar | Weak | Higher opportunity cost | 2012–2015: real yield −0.48% to 0.45%, dollar +18%; gold −30% | Market | FRED | “DFII10”, “DTWEXBGS”, annual means | 2012–2015; gold from Table 3 |
| Disinflation, confidence restored | Weak for years | Real yields attractive | 1980–1999: $608 to $279/oz | Market | World Bank | “Gold”, annual | 1980, 1999 |
Source: gold from the World Bank Pink Sheet, annual and monthly averages (Table 3); the US 10-year real yield (FRED DFII10), the nominal broad dollar index (FRED DTWEXBGS) and CPI-U (FRED CPIAUCSL) as annual means of their daily or monthly values; the S&P 500 from Robert Shiller’s monthly data (monthly averages). Changes are computed here; the performance and why columns are this guide’s reading of the examples. Historical, not predictive.
On past performance, gold’s monthly average rose 582% from its April 2001 low to its September 2011 peak, fell 39% into December 2015, then more than tripled by 2025. Over the full 2000–2025 window it comfortably outpaced inflation — its 2025 average was 6.6 times its 2000 average in 2025 dollars (Table 3) — and its drawdowns have rarely coincided with equities’, the source of its diversification value. Past performance is not indicative of future results.
The equity amplifier. Miners are how many readers take the exposure, and operating leverage — largely fixed costs against a moving gold price — makes them a geared version of the metal in both directions. Over that bull market the NYSE Arca Gold BUGS index (HUI) of unhedged miners rose about 1,430%, and in the bear it fell 81% — each measured from the index’s own turning points (its November 2000 low, September 2011 peak and September 2015 low), so the windows differ from gold’s by a few months. Whatever gold does, the miners have tended to do two to two-and-a-half times as much — leverage to be sought or avoided deliberately. (Data: monthly averages of the World Bank gold series and of the HUI’s daily closes, 2000–2016 — concluded episodes; the index is compiled by NYSE Arca, now part of ICE, and its history is taken from the daily bars behind CNBC’s .HUI quote page (Chart tab), since the compiler publishes no free history.)
On correlations (monthly changes, January 2006–December 2025), gold moves against US real yields (about −0.48) and the dollar (about −0.40), has a weak link to inflation (+0.11, comparing year-on-year rates — the hedge works over years), almost none with the S&P 500 (−0.02, the diversification property) and a strong one with silver (+0.77). Traders follow the pair through the gold/silver ratio — ounces of silver per ounce of gold — which averaged about 69 from 2000 to 2025 (monthly averages), from 35 at silver’s April 2011 peak to 111 in the April 2020 panic. The real-rate link held in monthly changes after 2022 (−0.47) but broke in levels: through 2006–2021 gold’s price and the real yield moved almost perfectly opposite (−0.94), while in 2022–2025 they rose together (+0.43). Correlations are sample-dependent and break down in crises, when most assets briefly move together.
Table 27. Gold correlations, monthly, 2006–2025
| Asset | Correlation with gold | Measure | Note | Source |
|---|---|---|---|---|
| Silver | +0.77 | Monthly % change | Moves together; silver more volatile | Market | World Bank | “Silver”, monthly | 2006–2025 |
| US inflation (CPI-U) | +0.11 | Year-on-year rates | A hedge over years, not months | Market | FRED | “CPIAUCSL”, monthly | 2005–2025 |
| S&P 500 | −0.02 | Monthly % change | The diversification property | Market | Shiller | ie_data, sheet “Data”, column B (“P”, S&P Comp. price) | 2006–2025 |
| US dollar (broad index) | −0.40 | Monthly % change | Gold is priced in dollars | Market | FRED | “DTWEXBGS”, daily → monthly mean | 2006–2025 |
| US 10-year real yield | −0.48 | Monthly change in the yield | −0.53 in 2006–21; −0.47 in 2022–25 | Market | FRED | “DFII10”, daily → monthly mean | 2006–2025 |
Source: computed here from monthly data, January 2006–December 2025 (240 months; 239 for the dollar and CPI) — gold and silver from the World Bank Pink Sheet, monthly, “Gold” and “Silver”; FRED DFII10 (monthly mean of daily yields) and DTWEXBGS; FRED CPIAUCSL (October 2025 unpublished); the S&P 500 monthly average from Shiller’s data. Pearson correlations of gold’s monthly percentage change with each series’ change as shown; for CPI, gold’s year-on-year change against year-on-year inflation. Using month-end rather than average yields weakens the real-yield figure to −0.28.
Figure 16. Gold’s correlations with major assets, monthly, 2006–2025
Figure data: Table 27, ordered from most positive to most negative; shading deepens with the size of the correlation and saturates at |0.75|.
4.4 Price drivers & cycles
Stripped of noise, the gold price answers to a short list of forces, and the clearest evidence comes from concluded episodes. On the demand side: real interest rates (the dominant macro lever), the dollar, investment and ETF flows, central-bank buying and crisis hedging. On the supply side: the slow, capital-intensive response of mines — years from discovery to first pour — and the faster but damped response of recycling. The recurring pattern is that gold leads at monetary turning points and lags real economic activity.
The settled cases: in 1933 the United States suspended the gold standard; from 1944 the Bretton Woods system fixed the dollar at $35 an ounce; when the US ended the dollar’s convertibility into gold in August 1971, gold floated and the 1970s bull began, peaking at a monthly average of $675 in January 1980 (World Bank monthly series) amid stagflation — followed by a two-decade bear as high real rates restored confidence in the dollar. The 2001–2011 bull (the dot-com bust, the 2008 financial crisis, quantitative easing and negative real rates) and the 2013–2015 bear that followed the “taper tantrum” — the 2013 bond sell-off when the Federal Reserve signalled it would slow its bond buying — are the episodes measured in Section 4.3, as is the 2020 pandemic rally, when real yields turned negative. Each is resolved history, and the lesson is durable: gold rallies when confidence in real yields or in the financial system erodes, and falls when both are restored.
Figure 17. Six dates that made the modern gold market
- 1933US suspends gold standardA proclamation of 20 April formally suspends the gold standard in the United States.
- 1944Bretton WoodsThe postwar system makes the dollar convertible to gold at a fixed $35 an ounce.
- 1971Convertibility endsThe US ends the dollar's convertibility into gold in August, and the price floats.
- 1999Sales cappedEuropean central banks agree that annual gold sales will not exceed about 400 t.
- 2004SPDR Gold Shares listA bullion-backed ETF lists on the NYSE on 18 November; ETFs become most investors' default.
- 2022Central banks buy 1,080 tOfficial buying passes 1,000 t for the first time in the series (from 2010) and stays above it through 2024.
Source: Federal Reserve History — Roosevelt’s Gold Program (“On April 20, President Roosevelt issued a proclamation that formally suspended the gold standard”), Creation of the Bretton Woods System and Nixon Ends Convertibility of US Dollars to Gold (“convertible to gold at a fixed exchange rate of $35 an ounce”); the ECB’s Joint statement on gold (26 September 1999: “Annual sales will not exceed approximately 400 tons”); the SPDR Gold Trust 2025 Form 10-K (NYSE listing, 18 November 2004); and the WGC’s central-bank series (Table 8, 1,080 t in 2022). Concluded events only; the events are evenly spaced, so the dates carry the time.
Table 28. Gold price drivers
| Driver | Direction of effect | Why | What to watch |
|---|---|---|---|
| Real interest rates | Lower real rates, higher gold | Cuts the cost of holding a non-yielding asset | US 10-year TIPS yield (FRED DFII10) |
| US dollar | Weaker dollar, higher gold | Gold is priced in dollars | Broad dollar index (FRED DTWEXBGS) |
| Central-bank buying | More buying, higher gold | Structural, price-insensitive demand | WGC quarterly demand; IMF reserve data |
| Investment and ETF flows | Inflows, higher gold | The marginal buyer at turning points | WGC monthly ETF holdings |
| Crisis and geopolitics | Risk-off, higher gold | Safe-haven demand | Equity drawdowns, conflict |
| Mine and scrap supply | More supply, mild drag | Slow and inelastic | USGS and WGC supply data |
Source: this guide’s synthesis of the episodes in Sections 4.3–4.4 and Tables 8, 12 and 26–27; the series named in the last column are the FRED and WGC data used there.
4.5 Risks, controversies & ESG
The bull case has real counterweights. The main financial risk is opportunity cost: gold pays nothing, so a sustained rise in real yields with a strong economy is its natural enemy — and can mean years of dead money, as 2013–2019 showed, when the annual average never regained its 2012 level. A strong dollar is a parallel headwind. A specific 2020s risk is central-bank reversal: official buying has become a major prop, and a slowdown or reversal — like the 2026 sales by Turkey and Russia in Section 2.7’s aside — would remove a pillar of demand. Regulatory and fiscal risk sits with producing states: Mali’s dispute with Barrick kept Loulo-Gounkoto shut for most of 2025, until it was handed back to Barrick in December (Table 5), and royalties, export rules and mining codes can change faster than a mine plan.
Gold mining also carries serious ESG exposure. Artisanal and small-scale gold mining accounts for 38% of all man-made mercury emissions to the atmosphere, according to the U.S. EPA, and the industry at large faces scrutiny over cyanide, tailings-dam safety, water use, deforestation and community relations — and, as Grasberg showed in 2025, worker safety underground. On the other side, gold’s defenders note that recycling supplies about a quarter of the market with a fraction of mining’s footprint, and that a tonne of gold disturbs far less ground per dollar of value than bulk commodities. There is little substitution risk — gold’s monetary role is unique — though thrifting — designing components to use less gold — reduces the gold content of electronics over time. These are contested questions that reasonable analysts weigh differently.
Figure 18. Gold risk map — likelihood vs. impact
Chart source: this guide’s qualitative assessment of the risks in Section 4.5 — positions are judgements, not measurements, and the shaded quadrant marks risks that are both likely and high-impact; the mercury share is from the U.S. EPA (“ASGM accounts for more than a third (38%) of all anthropogenic mercury emissions to the atmosphere”).
5. Future outlook & forecasts
Gold is unusual: it is accumulated, not consumed, so no agency publishes the multi-decade demand forecast that exists for oil or the industrial metals. Its forward view is a set of conditions and directions — official-sector buying, real interest rates, investment appetite — rather than a 2050 tonnage.
5.1 Demand
The clearest forward signal is official-sector demand. In the WGC’s 2026 survey of central banks, 89% expect global central-bank gold reserves to rise over the next 12 months, a record 45% expect their own reserves to rise, and 84% expect gold to be a larger share of total reserves in five years — while 74% expect a lower dollar share. The buying behind those answers is set out in Section 2.7. Investment demand is counter-cyclical — it rises as real rates fall and uncertainty climbs — while jewellery is price-sensitive and softens at record prices, as in 2025.
5.2 Supply: the “peak gold” question
Supply is famously inelastic. Mine output sits near a plateau — the USGS puts 2025 output at 3,300 t, still just below the 3,310 t of 2018, and the WGC warns that revisions could undo its own 2025 record — because grades are falling, large discoveries are thinning, and new mines take a decade or more to permit and build. The flexible part is recycling (Section 2.5). Because the above-ground stock dwarfs annual flows, gold never physically “runs short”: the market clears through price, which is why gold, unlike a consumed commodity, has no supply-versus-demand deficit to chart.
Table 29. Gold’s forward indicators (drivers, not a tonnage forecast)
| Driver | Recent level | Forward signal | Source |
|---|---|---|---|
| Central-bank buying | 848 t in 2025 (Section 2.7) | 89% of reserve managers expect global holdings to rise; 45% expect their own to rise | Market | WGC S&D | “Central banks” (Table 8); Filed | WGC CB survey 2026 | Perspectives on gold reserves | Charts 3–4 |
| Investment (bars, coins, ETFs) | 2,209 t in 2025, the highest in Table 8 | Counter-cyclical; rises as real rates fall and uncertainty climbs | Market | WGC S&D | “Investment” (Table 8) |
| Jewellery (India, China) | 1,648 t in 2025, −19% | Softens at record prices | Market | WGC S&D | “Jewellery fabrication” (Table 8) |
| Mine supply | ~3,300 t (USGS), near a plateau | Flat to slightly higher — the “peak gold” thesis | Filed | USGS MCS 2026 | “World total (rounded)" | p.91 (Tables 9–10) |
| Recycling | 1,404 t in 2025 | Price-elastic, damped when prices are expected to rise | Market | WGC S&D | “Recycled gold” (Table 12) |
Source: World Gold Council, Central Bank Gold Reserves Survey 2026 (16 June 2026; 76 responses, fieldwork 5 February–19 May 2026), “Perspectives on gold reserves” (Charts 1–4; the dollar-share answer is Chart 2); recent levels from Tables 8–10 and 12. Survey answers are expectations, not measured data; the forward-signal column for the last four rows is this guide’s reading.
5.3 Catalysts to watch
In the near term, the watch-list is the path of Federal Reserve policy and real yields, the pace of central-bank buying (especially the emerging-market buyers and any further sellers), ETF flows — which turned strongly positive in 2025, lifting holdings to 4,029 t by year-end (Table 14) — and geopolitical and de-dollarisation developments. Over 3–10 years, the structural themes are continued reserve diversification, the “peak gold” question as grades fall, and the demand pull from a growing Asian middle class. What would confirm the bull thesis: persistent official-sector buying and falling real rates. What would break it: a durable return to high real rates with a strong dollar, renewed confidence in financial assets and central banks turning sellers.
Table 30. Gold catalyst calendar
| Catalyst | Timing | Why it matters | Watch | Source |
|---|---|---|---|---|
| Federal Reserve rate decisions | Eight scheduled meetings a year | Sets the real-rate path | FOMC statements; DFII10 | Filed | Fed FOMC calendars | “eight regularly scheduled meetings during the year” |
| WGC Gold Demand Trends | Quarterly | Official-sector and demand data | gold.org | Filed | WGC GDT | Q4/FY2025 (29 Jan 2026) and Q2 2026 (30 Jul 2026) editions |
| Central-bank reserve data | Monthly (IMF IFS) | Structural demand | WGC reserves dashboard | Filed | WGC Gold reserves by country | IMF IFS, September 2026 edition |
| Gold ETF holdings | Monthly | The marginal investor | WGC ETF data | Market | WGC gold ETF holdings | monthly series | to 25 Sep 2026 |
| LBMA London vault data | Monthly, fifth business day | Physical liquidity in London | lbma.org.uk | Filed | LBMA London vault data | published on the fifth business day of each month |
| USGS Mineral Commodity Summaries | Annual (January–February) | Mine supply and reserves | usgs.gov | Filed | USGS MCS 2025 (January 2025) and 2026 (February 2026) editions |
Source: the Federal Reserve’s FOMC calendar (“The FOMC holds eight regularly scheduled meetings during the year”), the WGC’s Gold Demand Trends, reserves and ETF pages, the LBMA’s London Vault Data page and the USGS MCS — release schedules as stated on each.
6. Summary
Gold is the rare commodity that is also money. Physically it is a near-indestructible precious metal whose value rests on scarcity, durability and long monetary convention; nearly all of it still exists, some 222,600 t above ground. Its price is set not by producers but in London and on COMEX, and is best quoted as an average — $3,442/oz in 2025, a record in real terms, and $4,452 over the 12 months to September 2026. It is mined everywhere — China, Russia and Australia lead, but no country tops about 12% — so there is no cartel, and supply is anchored by a stock-to-flow ratio near 58 on the WGC’s count and a mine plateau around 3,300 t on the USGS’s. Demand rests on jewellery in India and China, investment, technology and, decisively in the 2020s, central banks, which bought more than 1,000 t a year in 2022–2024 and 848 t in 2025, when investment became the largest category. The market clears through price and the OTC residual rather than through shortage, and its vulnerabilities are the refining and vaulting hubs of Switzerland and London, not a shipping lane. The companies that mine it — Newmont, Agnico Eagle, Barrick, NMMC and peers — are best compared on production, reserves and AISC, never on a fast-moving market cap, and range from binary explorers to royalty firms; their shares have amplified gold’s moves two- to two-and-a-half-fold. For most investors the simplest exposure is a bullion-backed ETF or allocated metal, with miners and royalty companies as the geared and smoothed alternatives; the risks to weigh are opportunity cost, a reversal in official buying, and mining’s ESG record, from mercury in artisanal mining to tailings and underground safety. Gold’s regime is clear: it rewards falling real rates, a weak dollar, stagflation and crises, and struggles when real yields and the dollar both rise — with central-bank demand the force that broke that rule in 2022–2025. The variables to watch are the real interest rate and official-sector buying.
To go from this big-picture view to the actual companies — screening every gold producer by reserves, AISC and reserve life — explore Metal Pilot’s gold mining stock screener.
Terms in this post (61)
- Above-ground stock
- All gold ever mined and still held. Gold’s defining feature.
- AgEq, GEO (equivalent ounces)
- Silver-equivalent or gold-equivalent ounces: a mine’s other metals converted into ounces of its main metal at a ratio of prices, so a mixed output sums to one number.
- AISC (all-in sustaining cost)
- The cost to produce an ounce or pound and keep the mine running: operating costs, royalties, sustaining capital and corporate overhead, net of by-product credits, on the World Gold Council’s non-GAAP definition. Lower is better. In depth: Gold market guide — How to value & screen gold miners
- All-in margin
- The realised gold price less AISC, per ounce. The cash a producer keeps per ounce sold.
- Allocated / unallocated
- Allocated metal is specific bars held in the owner’s name; unallocated metal is a claim on a bank’s pool. Unallocated metal carries the bank’s credit risk.
- ASGM
- Artisanal and small-scale gold mining — often informal, low-tech mining by individuals and small groups, a meaningful share of mined gold outside company reporting.
- Attributable / payable / gold-equivalent
- A company’s share of output / what it is paid for / other metals converted to gold ounces. Why company totals are not strictly comparable.
- Block cave
- An underground mining method that undercuts an ore body and lets it collapse under its own weight. It mines very large, low-grade deposits at a low unit cost after a long and costly build.
- By-product credit
- Revenue from secondary metals netted against the cost of the primary metal. Large credits lower a mine’s reported unit cost but tie it to other metals’ prices.
- Carlin-type deposit
- Fine-grained gold disseminated in sedimentary rock, as in Nevada. Hosts the world’s largest gold complex.
- Central Bank Gold Agreement
- The 1999 pact capping European central banks’ gold sales. Shaped official selling in the 2000s.
- CIL / CIP
- Carbon-in-leach / carbon-in-pulp: cyanide leaching with activated carbon to capture gold. The standard way gold is extracted.
- COMEX
- The metals exchange of CME Group in New York; its gold, silver and copper futures are the US benchmarks (copper in cents a pound, converted here to dollars a tonne). In depth: Copper market guide — Pricing & benchmarks
- COMEX gold futures
- CME Group’s 100 oz gold futures contract in New York. Where price discovery and positioning concentrate.
- Comminution
- Crushing and grinding rock to fine particles to free the valuable minerals — usually the largest energy cost of a mill.
- Contango / backwardation
- The slope of a futures curve: contango when later-dated contracts cost more than nearer ones, backwardation when they cost less. Backwardation usually signals a tight market.
- Cost curve
- Producers ranked from the lowest to the highest unit cost against their cumulative output. The cost of the marginal producer tends to set the floor for prices over a cycle.
- Doré
- An impure gold-silver bar poured at a mine and sent to a refinery for purification.
- Electrowinning
- Plating dissolved gold onto cathodes with an electric current. The last step before doré is poured.
- Epithermal deposit
- Shallow gold veins and disseminations formed by hot fluids in volcanic belts. Can be high grade; some ores refractory.
- ETF (bullion-backed)
- A fund holding allocated bars against its shares. Bullion exposure without storage.
- Fineness / karat
- Purity in parts per thousand (995, .9999) or in 24ths (24 karat = pure). Distinguishes investment bars from jewellery.
- Flotation
- A froth process that separates sulphide minerals from crushed ore into a concentrate, using air bubbles and reagents.
- Gold/silver ratio
- The number of ounces of silver one ounce of gold buys. A high ratio means silver is cheap against gold. In depth: Silver market guide — Macro regimes, rates & correlations
- Good Delivery bar
- The standard bar of the London market, made by an LBMA-accredited refiner to a set weight and fineness.
- Grade
- The concentration of metal in the ore — grams a tonne for gold and silver, per cent for base metals and uranium. A higher grade means more metal per tonne mined and usually a lower unit cost. In depth: Gold market guide — What gold is — physical basics & quality
- Heap leach
- Stacking crushed low-grade ore on a lined pad and trickling a solution through it to dissolve the metal. It is cheap but recovers less than a mill.
- Hedge
- Contracts — swaps, forwards, options — that fix the price of future sales or set a floor under it, trading some upside for certainty.
- Kilobar
- A 1 kg bar, the standard unit of Asian demand. What refiners ship east.
- LBMA Gold Price
- The benchmark price of gold in London, set twice each business day, at 10:30 and 15:00, in an electronic auction run by ICE Benchmark Administration; in dollars a troy ounce. In depth: Gold market guide — Pricing & benchmarks
- Loco London
- Gold held and settled in London vaults. The deepest physical liquidity pool.
- Miller / Wohlwill processes
- Chlorine refining to about 99.5%; electrolytic refining to 99.99%. How doré becomes bullion.
- Mirror data
- Trade figures for a country built from its partners’ reports of trade with it, used where its own data are missing or unreliable.
- NPV / IRR
- Net present value is a project’s future cash flows discounted to today; the internal rate of return is the discount rate at which that value is zero — the two headline returns of a project study.
- Official sector
- Central banks, governments and public institutions such as the IMF, as holders, buyers, sellers or lenders.
- Operating leverage
- Profits that move more than revenue because costs are largely fixed, which makes a producer’s earnings swing harder than the commodity price.
- Ore grade
- The metal content of mined rock — a percentage for base metals, grams per tonne for precious metals. A falling grade raises the unit cost.
- Orogenic deposit
- Gold in quartz veins along ancient fault zones. The classic gold-only mine.
- OTC and other
- The WGC’s residual between supply and identifiable demand. Unobserved investment and stock changes.
- Paleoplacer
- An ancient river gravel turned to rock, as on South Africa’s Witwatersrand. Deep, labour-intensive, high-cost mining.
- Porphyry copper-gold
- A large low-grade copper deposit carrying gold. The main source of by-product gold.
- Primary vs. by-product
- Gold mined for itself vs. recovered while mining another metal. By-product supply ignores the gold price.
- Production
- The metal, oil or gas the company produced in the period at its own share of each mine or field — ounces, pounds or tonnes, or barrels of oil equivalent a day — before or after royalties as it reports; for a royalty or stream holder, its share of the operators’ output. In depth: The Oil & Gas Value Chain: Upstream — Units & reading a production number
- Proven and probable reserves
- The two confidence categories of mineral reserves; together they are the reserve base a mine plan is built on.
- Quantitative easing
- A central bank creating money to buy government and other bonds, to push down long-term interest rates.
- Real interest rate
- The nominal interest rate less inflation, actual or expected. A low or negative real rate lowers the cost of holding gold and other assets that pay no interest.
- Real yield
- A Treasury yield after inflation: the yield on Treasury Inflation-Protected Securities (TIPS), whose principal rises with the consumer price index. In depth: Commodities Across the Cycle — The dollar and real rates: the cross-asset hinge
- Recycling / scrap
- Gold recovered from old jewellery and electronics. The elastic part of supply.
- Refractory ore
- Ore whose gold is locked in sulphide minerals and needs roasting or oxidation before it can be leached, which raises its processing cost.
- Reserve life
- Years the reserves last at the current or guided production rate: reserves divided by a year’s output, on one reserve category. In depth: The Oil & Gas Value Chain: Upstream — How to read & screen a producer
- Reserves
- What a company can show it will produce at a profit from known deposits or fields, at stated prices and costs, estimated under a disclosure code — NI 43-101, JORC or S-K 1300 for mines; SEC rules or NI 51-101 for oil and gas. In depth: The Oil & Gas Value Chain: Upstream — Reserves vs resources
- Reserves vs. resources
- Reserves are the economically mineable part of an estimate; resources are the broader geological estimate of what is in the ground.
- Royalty
- A share of a mine’s or well’s revenue or profit paid to a government or a royalty holder, ahead of the operator’s own margin. In depth: The Oil & Gas Value Chain: Upstream — Fiscal regimes
- Royalty / stream
- A financing claim, bought upfront, on a share of a mine’s revenue (a royalty) or of its metal at a fixed price (a stream).
- Stagflation
- High inflation combined with weak growth.
- Stock-to-flow
- Above-ground stock ÷ annual mine production. Why gold supply is so price-inelastic.
- Stope
- An underground opening from which ore is mined. Underground mining method drives cost.
- Stream
- An upfront payment for the right to buy a share of a mine’s future metal at a fixed, low price: the operator gets capital now, the stream holder metal later. In depth: Silver market guide — Company archetypes along the value chain
- Taper tantrum
- The 2013 bond sell-off when the Fed signalled it would slow its QE purchases. Helped drive the 2013–2015 gold bear.
- Thrifting
- Redesigning a product to use less of a costly material per unit.
- Troy ounce
- The unit of weight for precious metals, 31.103 g — heavier than the ordinary ounce.
7. Sources, methodology & disclaimer
7.1 Sources, methodology & data vintage
Methodology. Data as of 7 October 2026; the guide is refreshed annually, after the USGS Mineral Commodity Summaries (January–February) and the WGC’s full-year Gold Demand Trends. Market structure is for full-year 2025. Prices are calendar-year, multi-year and trailing-12-month averages of the World Bank’s monthly series, never spot quotes; real prices deflate by US CPI-U to 2025 dollars. Two mine-production counts are kept apart: the USGS (3,300 t in 2025) for countries, reserves and the long-run series, and the WGC/Metals Focus (3,823 t, including artisanal estimates) for the supply–demand accounts and the stock-to-flow ratio. WGC supply and demand use the latest vintage (data to 30 June 2026) except the country split (Table 7), which exists only in the January 2026 edition. Correlations use monthly changes (CPI: year-on-year rates), January 2006–December 2025, and are computed in a script kept with this guide’s evidence; regime examples use annual means. Shares, changes, reserve lives, derived 100% mine figures and the official-holdings sample are computed here from the unrounded source figures; judgement columns are labelled. Company figures are each filer’s own basis (attributable, payable, consolidated or gold-equivalent), noted per row. Reserves, resources and survey expectations are estimates, not measured facts.
Data gaps & substitutions.
- WGC data files. The WGC’s workbooks and PDFs require a Goldhub login, so the annual supply and demand series, central-bank holdings and ETF holdings are taken from the data behind the public Goldhub pages, and the January 2026 report from its web edition; WGC figures therefore carry no page numbers.
- End-2025 above-ground stock and the world official total. Not reachable without the gated files; Table 14 uses the end-June 2026 stock, and Figure 12 a constant 93-country sample instead of the IMF-inclusive total.
- COMEX warehouse stocks and US trade. CME blocks scripted downloads of its stock reports, so COMEX inventories are not shown, and the rulebook’s contract terms are quoted as published but could not be re-checked by script. US-reported gold imports understate 2025’s inflow, so the surge is measured with partners’ reports (Tables 16–18).
- UAE and China trade. The UAE has filed no Comtrade data for 2024–2025, and China reports values without weights; mirror data stand in where noted.
- Deposit-type output. No agency publishes world mine output by deposit type, so Table 4 has no output column.
- By-product share. No agency publishes a world share of by-product gold; Section 2.5 uses the US share and mine examples.
- NMMC and Polyus. NMMC’s figures come from an April 2026 presentation, not audited accounts, and its Muruntau output is read from a chart; Polyus’s latest reserves are at end-2024 and it publishes no AISC.
- HUI history. The index compiler (NYSE Arca, part of ICE) publishes no free history; the HUI series is CNBC’s chart data (daily closes), averaged by month here.
- October 2025 US CPI. Not published; the 2025 CPI average and the CPI correlation skip that month.
Refresh log.
- 7 Oct 2026 — full refresh. Moved every series to the newest editions: USGS MCS 2026 (2025 estimates), the WGC’s supply and demand data to June 2026, the 2026 central-bank survey, UN Comtrade 2025 and the miners’ 2025 filings; prices to September 2026 from the World Bank. Rebuilt the price series on one downloadable source with a real-price row, the production series annually from 2000, the demand and supply tables annually from 2010; added the deposit-type, mine, consumer, balance, stocks, exporter, importer and hub tables, the cost curve from company AISC, and recomputed correlations; pinned every figure to its document with Source columns, numbered notes and this register; rebuilt the charts as live, theme-aware figures.
- 21 Jun 2026 — first published; revised 18 Jul 2026 (data as of 25 June 2026).
The register below is the guide’s reference list: one band per document, in order of first use, each linked to the edition cited.
Table 31. Source register — the document behind every table and figure
| Table / figure | What it takes | Locator | Period |
|---|---|---|---|
| USGS, Mineral Commodity Summaries 2026: Gold — page numbers are those of the full volume | |||
| Figure 1; Tables 9, 13; Figures 6, 8 | Mine production by country and world | World Mine Production and Reserves, “Mine production 2024” and “2025e”, p.91 | 2024, 2025e |
| Table 1 | Tonne-to-ounce factor | Footnote 1, “One metric ton (1,000 kilograms) = 32,150.7 troy ounces”, p.90 | — |
| Table 10; Figure 7; Table 29 | World mine production | “World total (rounded)”, p.91 | 2024–2025 |
| Table 30 | Release timing of the annual edition | Footer, “February 2026” | — |
| Tables 11, 14; Figure 1 | Reserves by country and world; Australia’s JORC basis | Column “Reserves” and footnote 11, p.91 | 2025 |
| Sections 2.3, 2.5 text | Top-five share; Nevada share; US by-product share | Domestic Production and Use; Events, Trends, and Issues, p.90 | 2025 |
| World Gold Council, gold supply and demand statistics (Goldhub) — data to 30 June 2026 (Metals Focus) | |||
| Figure 1; Tables 6, 8, 12, 15; Figures 4, 5, 10, 13 | Annual supply and demand by category | Annual series: “Mine production”, “Net producer hedging”, “Recycled gold”, “Jewellery fabrication”, “Technology”, “Investment”, “Total bc”, “Etfs”, “Central banks”, “LBMA” | 2010–2025 |
| Tables 14, 29 | 2025 mine output, demand and recycling used in the cover and outlook tables | Same series | 2025 |
| World Bank, Commodity Price Data (Pink Sheet), annual | |||
| Table 3; Figure 3; Figure 1 (market size) | Annual average gold price | Sheet “Annual Prices (Nominal)”, column “Gold” | 1971–2025 |
| Tables 25–26; Figure 15 | Multi-year averages and regime examples | Same column (via Table 3) | 1971–2025 |
| World Bank, Commodity Price Data (Pink Sheet), monthly | |||
| Figure 1; Section 1.3 aside | Trailing 12-month and 2026 averages; monthly record | Sheet “Monthly Prices”, “Gold” | Jan 2025–Sep 2026 |
| Tables 26–27; Figure 16; Sections 4.3–4.4 text | Gold and silver monthly prices; episodes; gold/silver ratio | Sheet “Monthly Prices”, “Gold”, “Silver” | 1980–2025 |
| World Gold Council, Gold Demand Trends: Q4 and Full Year 2025 — web edition | |||
| Figure 1; Tables 7, 13; Figure 8 | Jewellery and bar-and-coin demand by country | “Table 2: Jewellery demand in selected countries”; “Table 3: Total bar and coin demand in selected countries” | 2024–2025 |
| Section 2.2 text; Figure 1, note 8; Table 12, its note | First-print totals: demand, value, central banks, mine output | “Table 1: Gold supply and demand by sector”; Highlights | 2025 |
| Table 20; Section 2.7 text | Reported central-bank buyers and sellers; unreported share | Central banks section, chart “Reported central bank net purchases and sales” | 2025 |
| Table 30 | Publication date of the full-year edition | Page date, 29 January 2026 | — |
| LBMA, Good Delivery List Rules | |||
| Figure 2; Tables 1, 2, 31 | Good Delivery bar weight and fineness | §2.1.7 Specifications for a Good Delivery Bar, p.12 | — |
| USGS, Mineral Commodity Summaries 2026, Appendixes | |||
| Table 1 | Troy ounce in grams | Appendix A, Abbreviations and Units of Measure, p.1 | — |
| ICE Benchmark Administration, LBMA Gold Price and LBMA Silver Price Methodology | |||
| Table 2; Section 1.3 text | Auction times and currency | ¶4, p.1 | — |
| CME Group, COMEX Rulebook chapter 113 — Gold Futures | |||
| Tables 2, 24 | Contract unit and fineness | Rule 113101 | — |
| Shanghai Gold Exchange, Au99.99 contract details | |||
| Table 2 | Trading unit and deliverable bar | “Trading Unit: 10 g/lot”; “Deliverable Bullions: 1kg gold ingot” | — |
| FRED, Consumer Price Index for All Urban Consumers (CPIAUCSL) — U.S. BLS via the St. Louis Fed | |||
| Table 3; Figure 3; Tables 26–27 | CPI-U to deflate prices; inflation rates | CPIAUCSL, monthly, annual means | 1971–2025 |
| Newmont, 2025 Form 10-K and Q4 and full-year 2025 results release | |||
| Table 4 | Deposit descriptions (Carlin, Cadia) | Item 2, Properties | — |
| Table 5; Section 2.5 text | NGM, Ahafo South, Lihir and Cadia output; Pueblo Viejo share | Operating Statistics, p.63; Item 1, p.7 | 2025 |
| Table 21; Figure 14 | Attributable production; reserves; co-product AISC (Table 21, note 2) | 10-K Item 1, p.7; Item 2, p.71; Results and Highlights, p.4 | 2025, end-2025 |
| Table 21; Figure 15 | By-product AISC | Results release, Summary of Results, “Gold By-Product AISC ($/oz)”, p.3 | 2025 |
| Tables 23, 25, 31 | Reserve definition; AISC as a non-GAAP measure; reserve price | 10-K Item 2, p.70; Results and Highlights, p.4 | — |
| U.S. EPA, Reducing Mercury Pollution from Artisanal and Small-Scale Gold Mining | |||
| Tables 4, 31; Section 4.5; Figure 18 | ASGM share of gold supply and of mercury emissions | “producing up to 20% of the world’s gold annually”; “more than a third (38%)” | — |
| Barrick Mining, 2025 MD&A and Annual Information Form — Form 40-F | |||
| Table 5 | NGM attributable output; Pueblo Viejo; Loulo-Gounkoto suspension | MD&A p.12; AIF pp.22, 77 | 2025 |
| Table 21; Figures 14, 15; Section 3.3 | Production, AISC, reserves, listing, reserve price | MD&A Highlights, p.6; AIF Gold Reserves, p.32; Market for Securities, p.137 | 2025, end-2025 |
| Navoi Mining & Metallurgical Company (NMMC), corporate presentation; with list of affiliated persons | |||
| Tables 5, 21; Figures 14, 15 | Production, AISC, reserves, Muruntau share; state ownership | Presentation pp.5–6; affiliated persons p.1 | 2025; reserves at 1 Jan 2025 |
| Freeport-McMoRan, 2025 Form 10-K | |||
| Table 5, note 2; Sections 2.1, 2.5 | Grasberg gold output; ownership; mud rush | Consolidated mining data, p.34; pp.22–23, 58 | 2024–2025 |
| Polyus, FY2025 operating and financial results; with IFRS financial statements 2025 and reserves page | |||
| Tables 5, 21; Figure 14 | Output by mine and total; listing; reserves; cash costs | MD&A p.4 and p.2; FS Note 1, p.13, auditor’s report p.4; reserves page | 2025; reserves end-2024 |
| Agnico Eagle, 2025 MD&A (Form 40-F, 19 Mar 2026) and year-end 2025 reserves release | |||
| Table 5 | Detour Lake and Canadian Malartic output | MD&A pp.19, 21 | 2025 |
| Table 21; Figures 14, 15; Section 3.3 | Payable production, AISC, reserves, reserve price | MD&A pp.9, 29; release p.3 | 2025, end-2025 |
| AngloGold Ashanti, 2025 Form 20-F | |||
| Table 5 | Kibali attributable output | Africa region, p.83 | 2025 |
| Table 21; Figures 14, 15 | Production, reserves, AISC, listing | pp.148, 131, 200; cover | 2025, end-2025 |
| Kinross Gold, 2025 MD&A and Annual Information Form — Form 40-F | |||
| Table 5 | Paracatu output | MD&A p.16 | 2025 |
| Table 21; Figures 14, 15; Section 3.3 | Gold-equivalent production, AISC, reserves, reserve price | MD&A p.3; AIF p.8 | 2025, end-2025 |
| USGS, Data Series 140 — Gold statistics | |||
| Table 10; Figure 7 | World mine production | Sheet “Gold”, column “World production” | 2000–2018 |
| USGS, Minerals Yearbook 2023 — Gold, advance tables | |||
| Table 10; Figure 7 | World mine production | Table 8, row “Total” (kilograms) | 2019–2023 |
| World Gold Council, Above-ground stock — data to end-June 2026 | |||
| TL;DR; Table 14; Figure 9; Section 2.7 | Above-ground stock and its holders | “Total above-ground stock (end-Q2 2026)” and holder categories | end-June 2026 |
| LBMA, London Vault Data — page and monthly workbook, data to end-Aug 2026 | |||
| Tables 14, 18, 30 | Gold held in London vaults | Page text; workbook row “2025-12”, column “Gold” (thousand troy ounces) | Dec 2025, Aug 2026 |
| World Gold Council, Gold ETFs holdings and flows — data to 25 Sep 2026 | |||
| Table 14; Section 5.3; Table 30 | Global gold ETF holdings | Monthly holdings by region (tonnes); year-to-date total | Dec 2025, Sep 2026 |
| UN Comtrade, HS 7108 (gold, non-monetary) — public API, annual data (imports query) | |||
| Tables 16, 17 | Exports and imports by reporter, value and net weight | Reporter × world, flowCode=X and M | 2025 |
| Table 18; Figure 11; Section 2.6 | Partner flows of Switzerland, the UK, Hong Kong, India; mirror flows to the US, China and the UAE; Ghana’s exports to the world (Figure 11) | Reporter × partner; partner-reported (mirror) queries | 2024–2025 |
| Swiss Precious Metals Association (ASMP), The Swiss precious metals industry | |||
| Table 18; Section 2.6 | Switzerland’s share of gold refined | “1/3 of the world’s gold is refined in Switzerland” | — |
| LBMA, Good Delivery List — Gold | |||
| Table 18; Section 2.6 | Number of accredited gold refiners; Swiss refiners | “67 Gold refiners”; refiner addresses | Oct 2026 |
| Swiss Federal Council, Goldhandel und Verletzung der Menschenrechte — report on postulate 15.3877 | |||
| Table 18 (source line) | Swiss share of world refining capacity | “etwa 40 Prozent der weltweiten Kapazitäten”, p.5 | 2018 |
| LBMA, LBMA Gold Price | |||
| Section 2.7 text | Number of direct auction participants | “There are fifteen direct participants who have been on-boarded to contribute to the LBMA Gold Price” | Oct 2026 |
| World Gold Council, Gold reserves by country (Central Banks Dashboard) — IMF IFS, September 2026 edition | |||
| Tables 19, 30 | Official gold holdings and share of reserves; monthly IMF data | Quarterly table, “Gold Reserves Tonnes”, “Holdings %” | 31 Mar 2026 |
| Figure 12; Section 2.7 text | End-year holdings, 93-country constant sample | Annual table, “Gold Reserves Tonnes” | 2000–2025 |
| European Central Bank, Joint statement on gold | |||
| Section 2.7; Figure 17 | The Central Bank Gold Agreement sales cap | “Annual sales will not exceed approximately 400 tons” | 1999–2004 |
| World Gold Council, Gold Demand Trends: Q2 2026 | |||
| Section 2.7 aside; Table 30 | 2026 central-bank buying and sellers; Q1 revision | Central banks section; “Table 1: Quarterly gold supply and demand by sector” | Q1–Q2 2026 |
| Zijin Mining, annual results announcement for 2025 — HKEX | |||
| Table 21; Figure 14 | Mine-produced gold; gold reserves; substantial shareholder | pp.49, 35, 109 | 2025, end-2025 |
| Gold Fields, Reviewed Results 2025 — Form 6-K | |||
| Table 21; Figures 14, 15 | Gold-equivalent production, AISC, reserves | Salient features, p.1; Mineral Resources and Mineral Reserves, p.4 | 2025, end-2025 |
| SPDR Gold Trust, 2025 Form 10-K | |||
| Table 24; Section 4.1; Figure 17 | NYSE listing date of SPDR Gold Shares | Item 5, “initial public offering on November 18, 2004”, p.28 | 2004 |
| FRED, Market Yield on U.S. Treasury Securities at 10-Year Constant Maturity, Inflation-Indexed (DFII10) | |||
| Tables 26–28; Section 4.3; hook | US 10-year real yield | DFII10, daily; monthly and annual means | 2006–2025 |
| FRED, Nominal Broad U.S. Dollar Index (DTWEXBGS) | |||
| Tables 26–28 | The dollar | DTWEXBGS, daily; monthly and annual means | 2006–2025 |
| Robert Shiller, U.S. stock market data (ie_data.xls) | |||
| Tables 26–27 | S&P 500 monthly average | Sheet “Data”, column B (“P”, S&P Comp. price) | 2006–2025 |
| CNBC, NYSE Arca Gold BUGS Index (.HUI) chart data — daily closes | |||
| Section 4.3 (equity amplifier); Summary | HUI lows and highs (index compiled by NYSE Arca, part of ICE) | Chart tab, daily bars, averaged by month | 2000–2016 |
| Federal Reserve History, essays on Roosevelt’s gold program, Bretton Woods and the end of convertibility | |||
| Section 4.4; Figure 17 | 1933, 1944 and 1971 events | The three essays as linked in Figure 17’s source line | 1933–1971 |
| World Gold Council, Central Bank Gold Reserves Survey 2026 | |||
| Section 5.1; Table 29 | Central banks’ expectations for gold reserves | “Perspectives on gold reserves”, Charts 1–4; methodology | Fieldwork Feb–May 2026 |
| Federal Reserve, Meeting calendars (FOMC) | |||
| Table 30 | Number of scheduled meetings | “eight regularly scheduled meetings during the year” | 2026 |
| USGS, Mineral Commodity Summaries 2025: Gold | |||
| Table 30 | Release timing of the annual edition | Footer, “January 2025” | — |
| This guide — derived, estimated or illustrative | |||
| Figure 1 (notes 2, 5, 6, 7) | Market size, top-consumer total, recycling share, reserve life | Computed from Tables 3, 7, 9, 11, 12 | 2025 |
| Figure 2; Tables 4, 22–25, 28; Figure 18 | Value chain, deposit types, archetypes, asset and vehicle summaries, screening ranges, drivers, risk map | This guide’s summaries and judgements | — |
| Tables 2, 15, 26, 29 | The role, elasticity, performance and forward-signal columns | Judgement columns, labelled in each source line | — |
| Tables 3, 5–7, 9, 11–14, 16–17, 21; Figures 3, 12, 15 | Real prices, derived 100% mine output, shares, changes, residuals, net positions, reserve lives, cover, constant sample, cost-curve widths | Operations stated in each source line | 2000–2025 |
| Table 27; Figure 16; Section 4.3 | Correlations, gold/silver ratio, amplifier returns | Computed from the monthly series above | 2000–2025 |
Source: compiled from the source lines and Source columns under each table and figure above; each band links the edition cited.
7.2 Disclaimer & disclosure
This report is for informational purposes only and is not investment advice, a recommendation, or an offer to buy or sell any security or commodity. Gold prices are volatile, and the figures here are estimates as of the stated date that will change; reserves, resources, correlations and regime averages are estimates and historical observations that may not persist. Do your own research and consult a licensed financial adviser before acting. This report was prepared with the assistance of AI; its figures were sourced from the documents in the register above and reviewed, but readers should verify any number before relying on it. The author holds no position disclosed as a conflict in respect of the companies named.