Most of the World’s Gold Cannot Actually Be Bought or Sold

Roughly $33 trillion worth of gold sits above ground right now. That figure appears in headlines, investment summaries, and macro comparisons so often that most readers accept it as a measure of supply. It is not.

The $33 trillion counts everything: temple offerings in India, industrial traces in circuit boards, your grandmother’s bracelet, bars in central bank vaults that have not moved since the Cold War. It is a measure of accumulation over several thousand years, not a measure of what is available to buy or sell in response to a price signal today.

The gap between “above-ground stock” and “available gold supply” is the single most consequential misread in gold market analysis. Here is the framework, anchored in specific numbers from the CPM Group and the World Gold Council, for understanding what fraction of the world’s gold actually moves markets, and why that fraction is far smaller than the headline stock suggests.

Above-ground gold stock: what the $33 trillion figure actually counts

Start with the full number. Approximately 220,700 tonnes of gold, roughly 7.1 billion ounces, is estimated to exist above ground according to the World Gold Council, Metals Focus, and Refinitiv GFMS. At spot prices near $4,450-$4,600 per ounce in 2026, that aggregate value approaches $33 trillion.

As a macro comparison metric, the number is valid. It tells you how much gold humanity has accumulated. As a supply metric for understanding how gold prices or how investment flows interact with available metal, it is almost entirely misleading.

The reason sits in the composition. That 220,700 tonnes breaks down into four broad categories, each with a very different relationship to the market.

Before you even begin asking how much of that metal is mobile, subtract what is gone for good.

Approximately 700 million ounces of historically mined gold, roughly 10% of the total, is presumed permanently lost to shipwrecks, landfills, wear, and irretrievable industrial use. That leaves an estimated 6.3 billion ounces still in existence.

The CPM Group, led by founder Jeffrey Christian, draws a distinction that shapes the rest of this analysis: the difference between gold that “exists” and gold that “trades.” The $33 trillion tells you the first. Understanding the second requires looking at who holds each category, and why most of them will not sell.

Locked away by design: how gold ownership structure limits price-responsive supply

The categories that look largest on the balance sheet are the ones most firmly locked away. Working through them in order reverses the intuition that more gold means more supply.

Jewellery accounts for 45% of all above-ground gold, approximately 99,700 tonnes, and is heavily concentrated in emerging markets. In countries across South Asia, the Middle East, and parts of Africa, gold jewellery functions as dowry, generational savings, and cultural identity simultaneously. A modest fraction becomes scrap each year, and scrap flows do respond to price. But the overwhelming majority of jewellery stock is not sitting in a drawer waiting for the right bid. It is embedded in social and economic structures that make it almost entirely non-responsive to short-term price signals.

Central bank reserves hold roughly 38,600 tonnes, approximately 1.24 billion ounces. According to CPM Group, central banks collectively hold about 1.2 billion ounces of that total. These are policy assets, monetary backstops held as a strategic store of value. Since the global financial crisis, official institutions have been persistent net buyers rather than sellers. When central banks do sell, the decision is episodic and policy-driven, not price-driven. You cannot model these reserves as supply that continuously refills the market.

Central bank reserve accumulation has accelerated sharply since 2022, reinforcing the argument that official sector gold is becoming even less price-responsive over time as sovereign holders treat their positions as permanent monetary infrastructure rather than liquid inventory.

• Jewellery (~99,700 t): Cultural, dowry, and generational savings asset; almost entirely non-responsive to short-term price signals
• Central banks (~38,600 t): Policy asset and monetary backstop; net buyer since the global financial crisis; sales are episodic and policy-driven
• Long-term private holders (portion of ~51,000 t bars/coins/ETFs): Strategic reserve or family wealth with very long horizons and high sale thresholds; behaves more like quasi-permanent reserve than dealer inventory

Where investor gold sits on the liquidity spectrum

The bars, coins, and ETF category at 23% of the total is where most people assume the tradeable market lives. But this category spans a wide spectrum. At one end: short-term traders, leveraged products, and actively managed ETF holdings that respond to price over days or weeks. At the other: family wealth stored in vaults for decades, institutional allocations with extraordinarily long horizons, and private holdings whose owners have no intention of selling below a price threshold that may never arrive.

Aggregating all of it as “investible” obscures the structural immobility of a large share. According to CPM Group, investors have been net buyers in nearly every year since the mid-1960s, though gross sales occur continuously alongside gross purchases. The fraction that genuinely responds to price over short horizons is the only portion that belongs in the tradeable float calculation.

The spectrum from vaulted bars to ETF shares to futures contracts matters here, because the distinction between physical versus digital gold determines whether a position can contribute to the price-responsive float or simply represents a derivative claim layered on top of existing metal.

The World Gold Council defines an “investible” gold pool of roughly 100,000+ tonnes, worth approximately $15 trillion. That is a useful upper bound by category, but it still overstates functional liquidity because it classifies metal by ownership type rather than by whether the owner has demonstrated any willingness to sell at prevailing prices.

Quantifying the price-responsive pool: how large is the genuine bullion float

So what is left? After stripping away jewellery, central bank reserves, long-term private holdings, and permanently lost metal, the pool of gold that is genuinely available to respond to market signals is far smaller than any category-level aggregate suggests.

CPM Group defines the tradeable float as bullion held in deliverable form by actors whose mandate involves trading, leasing, or collateralising gold, and who have demonstrated willingness to transact at prevailing prices over relatively short horizons.

Of all the gold estimated to still exist, roughly 5-8% constitutes the true price-responsive supply base: several hundred million ounces out of 6.3 billion, according to CPM Group founder Jeffrey Christian.

That estimate is the number that actually matters for price analysis. Three characteristics qualify bullion as part of this float:

1. It is held in deliverable bullion form (bars, allocated metal, ETF backing, vaulted OTC holdings)
2. It is controlled by actors with a trading or leasing mandate, not simply holding it as a policy or legacy asset
3. It has demonstrated historical willingness to enter and exit the market as prices, spreads, and funding conditions change

A concrete example of this float in action: LBMA-reported London bullion inventories climbed back toward the highs recorded in 2020-2021, with a substantial portion of that recovery driven by metal returning from New York once the arbitrage premium that had incentivised outbound shipments faded. The metal that moved between hubs in response to spread and delivery dynamics was institutional bullion held specifically for trading. Central bank reserves did not move. Household jewellery did not move. That is the operational definition of float.

The LBMA structure complicates this picture further, because a large share of what traders count as ‘bullion exposure’ is actually unallocated gold ownership, a claim on a pool rather than a title to specific bars, which means the apparent float is thinner still when you strip out positions that cannot be physically delivered without a conversion process.

For scale, consider the supply side. Record global mine output in 2025 totalled approximately 118 million ounces (roughly 3,672 tonnes), generating revenues of around $544 billion. To put that figure in perspective, US federal net interest payments in the prior year came to approximately $970 billion, meaning the entire planet’s annual gold production at record levels covered only around half of that single budget line. Even at record levels, new supply additions are small relative to both the existing float and the scale of global financial needs.

What a structurally thin float means for gold pricing and investor positioning

Understanding the float’s size is the foundation. The consequences of that size are what change how you interpret gold market data.

When only 5-8% of above-ground gold is truly price-responsive, incremental buying by ETFs, sovereign wealth funds, or central banks reallocating reserves can represent a meaningful share of the entire available pool. A central bank allocation shift that looks modest as a percentage of $33 trillion is potentially very large as a share of several hundred million ounces of genuinely mobile bullion. This structural thinness amplifies the price impact of capital flows relative to what raw tonnage figures would suggest.

The float framework also forces a sharper read on demand data. According to CPM Group, gold fabricators working with casting techniques lose roughly 50% of input material as production scrap, while sputtering target processes generate scrap at around 85%, producing a blended average recovery rate across all fabrication of approximately 67-68%. That recovered material is re-refined and re-enters the market, making fabrication largely a recycling loop rather than a true consumption channel. This dynamic means that headline delivery figures from exchanges such as the Shanghai Gold Exchange substantially overstate end-user consumption, and CPM Group notes that some analysts have made precisely this error when estimating Chinese fabrication demand, treating gross throughput as net offtake rather than accounting for the volume that cycles straight back as refined metal.

• Price amplification: Incremental buying from ETFs, sovereign wealth funds, or central banks can represent a significant share of the float, producing price moves disproportionate to the absolute tonnage involved
• Gross-vs-net demand distinction: Fabrication and exchange delivery data must be read through the scrap recovery lens; gross flows overstate net demand, particularly in Chinese market data
• Producer equity differentiation: Mine supply is one of the few mechanisms that genuinely expands the tradeable float; producer equities carry a structural property that synthetic exposures cannot replicate

Producer equities and the float expansion argument

In periods of monetary or credit stress, the narrow slice of institutionally held, deliverable, actively traded bullion relative to global financial balance sheets can produce sharp and rapid repricings. The relevant constraint is not total gold stock but the small float available to absorb sudden demand.

Mine supply is one of the few ways to genuinely expand that float over time. Producers with growing output are not simply leveraged gold price plays; they are participating in float creation. ETFs and futures redistribute claims on existing metal. They do not add new metal to the price-responsive pool. That structural distinction gives producer equities a leverage dimension that synthetic gold exposures cannot replicate, and it becomes more consequential precisely when the float is under pressure.

Production-stage gold equities carry this float-expansion property at a moment when margins are historically wide, meaning investors gain both a leveraged price claim and exposure to the one mechanism that structurally enlarges the pool of deliverable metal over time.

Applying the float framework: a more accurate lens for gold market analysis

Return to where you started. The $33 trillion figure is a valid macro comparison, a measure of humanity’s accumulated gold across millennia. It is an actively misleading supply metric when used as a denominator for understanding why gold prices move as sharply as they do.

The gold market is a large, highly traded, but structurally constrained arena. Approximately 220,700 tonnes of gold exists above ground. After accounting for permanent losses, roughly 6.3 billion ounces remain. Of that, several hundred million ounces, approximately 5-8%, constitutes the genuinely price-responsive float. The World Gold Council’s $15 trillion “investible” pool is a useful but still overstated upper bound compared with CPM Group’s tighter float definition.

The next time you encounter a headline about central bank purchasing, ETF inflows, or sovereign wealth fund allocation to gold, measure it against the float, not the aggregate stock. A demand development that looks small relative to $33 trillion can be very large relative to several hundred million ounces of genuinely mobile bullion. That recalibration is where the apparent mismatch between “so much gold in the world” and “sharp, sustained price moves” resolves immediately.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions.