Ask a room of gold owners where their metal came from and most will name a dealer, a mint, or a grandparent’s safe deposit box. Ask where it came from before that and the answers thin out quickly. It is a fair blind spot. The gold market talks about interest rates, the dollar, and central bank buying almost every day, and it talks about supply almost never.
That silence is informative on its own. Supply gets ignored largely because supply barely moves. Gold traded around $4,300 an ounce in late September 2026, and through all the years it took to get there, the volume of new metal coming out of the ground shifted by a couple of percent annually. Anyone holding coins or bars owns a claim on a stock that grows slowly no matter what the price does.
How much gold is there, actually?
The World Gold Council puts the total ever mined at about 222,600 tonnes as of mid-2026. Gold does not rust, burn, or get used up the way a barrel of oil does, so nearly all of it still exists in some form. Melted into one block, the Council reckons the entire human inheritance of gold would form a cube roughly 22.6 metres on a side. For a metal that has underwritten money for several thousand years, that is a startlingly modest pile.
Two-thirds of it has been mined since 1950, which says less about gold than about industrial-era mining. Where it sits now breaks down roughly like this:
- Jewellery: about 99,700 tonnes, or 45 percent
- Bars and coins: about 47,800 tonnes, or 21 percent
- Central bank reserves: about 39,000 tonnes, or 18 percent
- Other uses, including electronics and dentistry: about 22,100 tonnes, or 10 percent
- Exchange-traded funds: about 4,000 tonnes, or 2 percent
The bars-and-coins line deserves a second look. Private holders of physical metal collectively own more gold than every central bank on earth, an unusual feature of this market and one reason retail buying gets watched as closely as it does.
Where new gold comes from each year
Annual supply arrives through two channels. Mine production is the larger one, typically about 75 percent of the total, with recycled metal supplying most of the rest. In the second quarter of 2026, according to the Council’s demand trends data, miners delivered 965.6 tonnes and recycling returned 326.1 tonnes, for total supply just under 1,270 tonnes. Full-year mine production hit 3,672 tonnes in 2025, a record, though a record reached by a margin of about one percent over the prior year.
Mining is also remarkably spread out. Operations run on every continent except Antarctica, and China, the largest producing country, accounts for only around a tenth of global output. There is no OPEC for gold, no small group able to open or close the taps together. Political trouble in any one jurisdiction dents the total without breaking it.
Are we running out of gold?
This is the question the numbers get recruited for most often, usually in service of an argument that prices must therefore rise. The honest answer is less dramatic. Metals Focus estimated proven reserves, meaning gold that is economic to extract with current technology at current prices, at 54,770 tonnes at the end of 2025. The United States Geological Survey puts the figure nearer 64,000 tonnes. Broader resources, the category that includes deposits not yet proven economic, come to roughly 132,110 tonnes.
Set 54,770 tonnes against annual production near 3,650 tonnes and you get roughly fifteen years of reserves, a figure that sounds alarming until you notice the pattern behind it. Reserve estimates have held remarkably stable for decades even as enormous quantities were dug up and sold. Reserves do not measure how much gold exists. They measure how much gold companies have bothered to prove up, and companies prove up about as much as they need. Higher prices and better technology keep pulling marginal deposits across the line.
So the case for running out of gold does not really hold. The Council’s own reading is that global mine production will gradually plateau rather than peak and fall away. A plateau matters, but it is not scarcity arriving on a deadline.
Why a higher price does not summon more metal
Here is the part that matters most for anyone thinking about gold in a portfolio. In most commodity markets, a price spike calls forth extra supply, which in time caps the spike. Gold answers that signal so slowly that the feedback loop nearly stops working.
S&P Global Market Intelligence, studying 127 mines, found the average journey from discovery to first production ran 15.7 years. Exploration, drilling, feasibility studies, permitting, financing, and construction each take their turn. The World Gold Council makes a similar point from the other direction, estimating that mined production lags the gold price by at least six years. A company that decided in 2026 that today’s prices justified a new mine would be pouring its first bar somewhere in the late 2030s, by which point the price that prompted the decision is ancient history.
Ore grades add to the drag. The rich, shallow deposits were found first, and what remains tends to be deeper, more remote, or leaner. Major discoveries have been declining for years. None of that halts production, but each additional tonne costs more effort than the last.
The practical consequence: when gold moves, supply is almost never the reason. Price is set by what buyers are willing to pay for a stock that was already sitting there. If you want to understand a move, the more productive place to look is real interest rates, the dollar, and the expected path of Fed policy, whether that path points toward tightening or toward the balance sheet expansions covered in our look at quantitative easing.
Recycling is the part that flexes
If mine supply is a glacier, recycled gold is weather. At least 90 percent of it comes from old jewellery, and it responds to price almost immediately, because the decision to sell a chain sitting in a drawer takes an afternoon rather than fifteen years.
The second quarter of 2026 showed the mechanism plainly. Recycling fell 13 percent from the previous quarter, to 326.1 tonnes, as softer prices during the period discouraged people from parting with old pieces. When prices run, scrap flows back in; when they ease, it dries up. Recycling functions as a shock absorber, damping moves at the margin without ever being large enough to set the price.
What the supply picture means for holders
Run the arithmetic and one number falls out that is worth remembering. Annual mine production of roughly 3,650 tonnes against an above-ground stock of about 222,600 tonnes means the world’s gold supply grows a little over 1.5 percent a year. It has grown at approximately that pace for a long time, and the structural obstacles described above make a sudden acceleration implausible.
That steadiness is arguably the whole proposition. We are not running out of gold in any sense that should drive a decision. The binding constraint sits elsewhere: nobody, including any government, can resolve to make meaningfully more of it next year. The durability that draws people to the metal rests on that constraint rather than on any forecast about depletion.
It also sets expectations sensibly. Because supply is close to fixed, gold’s price becomes a referendum on demand, which is why the metal can fall for long stretches while the mining story never changes. Supply gives you a floor of sorts, since the metal content of a bullion coin cannot be printed away, but it offers nothing in the way of a timeline. Readers wanting the current state of play can watch live metal prices, and anyone newer to physical ownership may find our overview of how buying works a useful next stop.
None of this is investment advice, and the data cuts both ways: the same inelasticity that prevents a flood of new gold also prevents supply from rescuing the price when demand goes quiet. What the numbers offer is a sense of proportion. The world’s gold is a 22.6 metre cube growing by a sliver each year, and it has been doing that patiently for a very long time.
