For the first time since 1996, the world's central banks hold more gold than US Treasuries. The European Central Bank's June 2026 report put Treasuries at 22% of global reserves, down from 25%, while gold's share climbed past it, as reported by the European Central Bank. The World Gold Council reached the same conclusion in January, valuing foreign central bank gold at roughly $4 trillion against about $3.9 trillion in US government bonds.
Central banks buying gold at this scale is now a four-year pattern rather than a headline. The World Gold Council's own 2026 survey puts the average at about 1,000 tonnes a year over the past four years, double the roughly 500 tonnes a year of the preceding decade.
The most instructive element of that record is the timing. This accumulation continued while gold traded some 27% below its January record, at approximately $4,052 an ounce on 24 July 2026 per Trading Economics. Institutions with the longest investment horizons in finance continued to buy through the decline, and the reasoning behind that translates directly to how an individual investor might construct a position.
This analysis covers the verified data, the reasoning behind it, the balancing evidence, and the practical question of what an individual version of this strategy involves. Every figure is sourced to the World Gold Council, the ECB, the IMF, or OMFIF.
Informational only. This is research, not financial advice. Figures verified as of 26 July 2026.
The numbers require a baseline to be meaningful. The pre-2022 figures establish how substantial an annual rate of 1,000 tonnes actually is.
| Year | Net central bank gold purchases (tonnes) |
|---|---|
| 2014 | 601.2 |
| 2015 | 579.6 |
| 2016 | 394.9 |
| 2017 | 378.6 |
| 2018 | 656.2 |
| 2019 | 605.4 |
| 2020 | 254.9 |
| 2021 | 450.1 |
| 2022 | 1,080.0 |
| 2023 | 1,050.8 |
| 2024 | 1,089.4 |
| 2025 | about 850 |
The 2014 to 2024 figures come from Visual Capitalist's decade chart built on World Gold Council data, which notes that 2022 to 2024 purchases totaled 3,220.2 tonnes against 1,575.7 tonnes across 2014 to 2016. The 2025 figure comes from the ECB report cited above.
Read against that baseline, 2022 represents a structural shift rather than a single-year spike. Three consecutive years above 1,000 tonnes were followed by a moderation to approximately 850 tonnes, a level that remains some 80% above the pre-2022 decade average. The floor has moved.
The 2026 data continues the pattern. Central bank demand reached 244 tonnes in the first quarter, up 17% on the previous quarter and 3% year on year, per the World Gold Council's Q1 2026 Gold Demand Trends. April brought a further 19 tonnes of net buying, a rebound after net sales in March.
Cumulatively, central banks now hold more than 36,000 tonnes, which the ECB notes is approaching levels last seen during the Bretton Woods era. That is roughly a fifth of all the gold ever mined.
This is a two-sided market, and examining both sides establishes the trend's credibility.
The buyers, as of April 2026 (World Gold Council monthly statistics):
Poland added 14 tonnes in April, bringing year-to-date purchases to 45 tonnes and holdings to 595 tonnes, around 30% of its total reserves. Largest single buyer for a second year running.
China added 8 tonnes, its largest monthly purchase since December 2024, extending a run to 18 consecutive months. Official holdings sit at roughly 2,322 tonnes, about 9% of total reserves.
Uzbekistan sold 1 tonne in April but remains a net buyer year to date at 24 tonnes, with holdings of 414 tonnes representing an extraordinary 88% of its total reserves.
Czech Republic bought 3 tonnes, its 38th consecutive monthly purchase, taking holdings to 79 tonnes.
The sellers:
Russia posted net sales for a fourth consecutive month, 6 tonnes in April, 22 tonnes year to date.
Turkey sold or loaned roughly 130 tonnes in early 2026 after accumulating about 220 tonnes since 2022, according to the ECB report.
Azerbaijan's SOFAZ also reduced holdings during the first quarter.
The reasons behind the disposals are informative. Russia's sales track fiscal requirements. Turkey's reflects domestic demand dynamics and reserve management operations involving gold swaps. Both are treasury decisions rather than revisions of view on the metal, which is consistent with the function of a mature reserve asset: it exists to be drawn upon when a sovereign requires liquidity.
The United States remains the largest official holder at roughly 8,133 tonnes, followed by Germany, Italy, and France, with Russia and China next. The World Gold Council maintains the full country-by-country table, updated monthly.
The answer to which country has the most gold reserves has been stable for decades. The more revealing measure is gold as a share of each country's reserves, where the dispersion is considerable:
Uzbekistan: about 88%
United States: about 70%
Poland: about 30%
China: about 9%
Switzerland: about 5%
That dispersion becomes a practical allocation framework later in this analysis, and it is among the most directly applicable insights the central banks buying gold data offers an individual investor.
The headline answer to why central banks are buying gold is diversification. The causal chain beneath that term is more specific, and more instructive.
In February and March 2022, the US and its allies froze roughly $300 billion of Russian central bank reserves following the invasion of Ukraine. It was the first time a G20 central bank's reserves had been immobilized at that scale.
Reserve managers globally drew a common conclusion: a foreign currency reserve constitutes another party's liability, whereas gold is held outright.
Gold is the only major reserve asset that is nobody's liability. Purchases tripled from 450 tonnes in 2021 to 1,080 tonnes in 2022 and have never returned to the old baseline.
The proposition is supported in the academic literature. Federal Reserve research staff examined the question in working paper 1420 on de-dollarization and reserve diversification, and Brookings cites research findings that financial sanctions correlate with higher gold allocations among geopolitically misaligned states. The de-dollarization gold thesis, therefore, rests on published research rather than on market commentary alone.
The ECB is blunt about the driver. "Geopolitical tensions continue to drive strong central bank demand for gold," wrote ECB President Christine Lagarde in the June 2026 report.
The World Gold Council's 2026 Central Bank Gold Reserves Survey, published 16 June 2026, drew 76 responses, the highest participation in the survey's nine-year history. It was fielded between 5 February and 19 May 2026, with most responses arriving after the Middle East conflict began.
The headline findings:
89% expect global central bank gold reserves to rise over the next twelve months
A record 45% expect their own reserves to rise. Just 1% expect a decrease
Crisis performance, portfolio diversification, and inflation hedging rank as the primary reasons for holding gold
Geopolitical risk hedging and reserve diversification policy features are the key reasons for increasing allocations
The Official Monetary and Financial Institutions Forum found the same thing from a different sample. OMFIF's 2026 Global Public Investor report surveyed 74 central banks managing more than $10 trillion, and found 82% now hold physical gold, up from 71% a year earlier, as covered by Kitco.
Andrea Correa, OMFIF's Head of Research, put it plainly to Kitco: "Gold is not moving anywhere. Reserve managers of the central banks are still very bullish on gold. Despite the fact that the gold value itself keeps rising, they are still demanding it."
A secondary shift accompanies the buying. Central banks have moved from owning gold to holding it directly.
Repatriation is accelerating alongside the buying, and the two trends are connected. The Reserve Bank of India moved over 100 tonnes back to domestic vaults. Several European central banks have increased the share stored at home rather than abroad. The WGC survey tracks a measurable rise in central banks planning to increase domestic storage or diversify their overseas vaulting locations.
The reasoning transfers directly to individual investors. Central banks are not only acquiring gold but also reducing counterparty exposure to where the gold is held. Custody forms part of the position rather than an administrative detail, a point developed as rule four below.
The gold vs US treasuries reserve asset crossover is real, and one piece of context makes it far more useful to understand.
The claim. Per the ECB's June 2026 report, US Treasuries fell to 22% of global central bank reserves from 25%, while euro-denominated reserves held steady at 15%, and gold's share rose past Treasuries. By dollar value, the World Gold Council put foreign central bank gold near $4 trillion against roughly $3.9 trillion in US Treasuries, the first crossover since 1996.
The context. Both factors contributed. Gold approximately doubled between the end of 2023 and its 2026 peak, and the ECB notes that appreciation raised gold's share alongside sustained buying from China, Poland, Turkey, and India. The crossover, therefore, reflects an appreciating asset held in increasing size, a stronger combination than either factor in isolation.
Also worth noting: dollar-denominated assets in aggregate still represent the largest share of global reserves at 42%.
The de-dollarization argument is frequently presented without qualification. The following figure adds the necessary precision, and precision is what makes the trend actionable.
The dollar's share of allocated foreign exchange reserves stood at 56.77% in the most recent IMF COFER data, with the euro at 20.25% and the renminbi at 1.95%, per COFER data. Total allocated reserves sit at around $13.1 trillion across 149 reporting central banks.
Two observations hold simultaneously:
The dollar's share has fallen roughly 9.5 percentage points since 2015, which is a real and sustained decline.
The renminbi remains under 2% after fifteen years of internationalization efforts, and no currency is meaningfully replacing the dollar as the world's transaction and invoicing currency.
The reconciliation: gold is displacing Treasuries inside reserve portfolios. It is not displacing the dollar as the global unit of account. Those are different claims, and conflating them is what makes most de-dollarization commentary unreliable. The de-dollarization gold story is real at the portfolio level and considerably weaker at the currency level, and both halves belong in any honest account.
This qualification is included because the more precise version of the analysis is the more useful basis for an allocation decision.
This is the question most readers arrive at, and the answer is among the more useful conclusions in this analysis.
Gold set an intraday record above $5,500 an ounce in January 2026 and traded near $4,052 in late July, a drawdown of roughly 27%. Central banks kept buying throughout. Four reasons that is not a contradiction:
1. Horizon. Reserve managers hold positions for decades, within which a 27% drawdown is a short-term movement. Individual investors typically operate under greater pressure to react to such moves.
2. No mark-to-market pressure. A central bank faces no redemptions, no quarterly performance review, and no investment committee. It can hold an underwater position for years without consequence.
3. Price weakness is the buying condition. The People's Bank of China recorded its largest monthly purchase since December 2024 during the decline, and Poland continued accumulating toward its 700-tonne target. This is counter-cyclical accumulation operating as designed.
4. Their objective is resilience rather than short-term return. They hold an asset with no counterparty that performs when other assets are under pressure. Six months of price action is not the timeframe that the mandate is measured on.
The complete picture sharpens the thesis. OMFIF found that a minority of central banks report higher prices moderating their purchases, and gold ETF inflows were subdued in May 2026. Sovereign demand functions as substantial support rather than a guarantee, so the question of whether you should buy gold rests on the individual investor's horizon rather than on official sector behaviour.
The central banks themselves are bullish. OMFIF's survey found the majority expect gold to trade between $5,000 and $6,000 an ounce within twelve months.
Wall Street is split, and the spread is unusually wide:
| Institution | Year-end 2026 target |
|---|---|
| JPMorgan | $6,000 |
| Goldman Sachs | $4,900 (trimmed by $500 in June) |
| Bank of America | $4,800 |
That $1,200 range comes from investingLive's July 2026 summary, which also flags the most relevant detail for this article: Goldman revised its central bank demand model upward to roughly 60 tonnes a month through 2026, from an earlier 29-tonne pace. Sovereign buying is now the structural anchor of the most-cited bull case on the street.
A $4,800 to $6,000 range indicates that the sell side anticipates upside while disagreeing on its magnitude. Any gold price forecast 2026 presenting a single figure is compressing a genuine range. Forecasts are best treated as context, with attention directed to the central bank data, since that is the variable on which the models depend, and it has remained consistently strong. Our gold price forecast 2026 breaks down every major bank target and the five forces behind them, and how high will gold go in 2026 maps the bull, base, and bear scenarios. For the longer arc, see our gold price prediction to 2030.
Individual investors cannot acquire 100 tonnes. They can, however, adopt the four behaviours that distinguish reserve managers from retail buyers, each of which runs counter to common individual practice.
This section translates the central banks buying gold data into a set of applicable decisions.
The evidence. The Czech National Bank has bought gold in 38 consecutive months. Poland has accumulated through peaks and drawdowns without pausing. Neither is timing the market.
The translation. Fixed-interval purchases, automated where possible. A central bank buying monthly through 2026 will achieve a better average cost than an individual waiting for a low point that is identifiable only in retrospect.
The practical note. Format determines whether this is viable. Purchasing $300 of gold monthly through a dealer incurs fractional-coin premiums of 12% to 18%. Tokenized gold carries no minimum bar size and no per-purchase dealer markup, which makes scheduled accumulation efficient at any amount. Our guide to investing in gold ranks every route against this criterion.
The evidence. Poland has a stated 700 tonne target, and the WGC notes the bank appears focused on reaching it even after Governor Adam Glapiński raised the possibility of selling some holdings. Gold is now about 30% of Polish reserves. Uzbekistan runs 88%. China sits under 10%.
The translation. There is no universally correct allocation, and every serious institution operates to a defined target. Select a percentage of net worth, commonly 5% to 15% in professional portfolio construction, and rebalance to it. An undefined allocation is not a strategy.
The evidence. China's largest monthly addition since December 2024 landed during the 2026 price decline. Central bank buying tripled after the 2022 shock rather than during the 2020 euphoria.
The translation. Where the horizon is measured in decades, price declines represent the accumulation window, and the discipline to continue buying through one distinguishes the institutional approach from the retail one.
The requirement. This approach depends on the ability to hold through a full cycle. Gold has recorded extended flat periods as well as record runs, so it should be funded with capital not required within five years.
The evidence. Repatriation is accelerating across the official sector. India moved 100 tonnes home. Central banks increasingly want allocated metal in named vaults rather than pooled claims held elsewhere.
The translation, set out from weakest to strongest:
Weakest: an unallocated gold account or pooled certificate, under which the holder is an unsecured creditor.
Stronger: a physically-backed ETF, offering institutional custody without delivery and limited to market hours.
Stronger still: allocated, serial-numbered bars in a named vault, which requires substantial capital to arrange properly.
Strongest for most investors: self-custodied tokenized gold with allocated backing and published attestations. The holder retains the keys, the metal is identified, and no storage contract is required.
Central banks acquire gold and hold it for decades. Review what holding it on-chain involves over a comparable horizon.
The ECB's June 2026 report contains a finding that has received limited attention:
Tether, the stablecoin issuer, was the single largest buyer of gold in 2025, acquiring more than 100 tonnes.
That is more than every sovereign central bank bought that year. A crypto company ran the central bank playbook and outbought the central banks.
The implication merits consideration. When the largest issuer in digital assets required a hard reserve asset, it allocated the marginal dollar to metal, and holds that metal behind a tokenized product. This constitutes the clearest third-party validation of the thesis available, and it originates with the European Central Bank.
The structural features reserve managers require are now available at retail scale:
Allocated, serial-numbered bars in LBMA-standard custody, the same arrangement a central bank demands
Published attestations, and increasingly on-chain proof of reserve, exceeding the disclosure most central banks provide
Self-custody without a vault contract, storage costs or a minimum tonnage requirement
Divisibility to a thousandth of an ounce, which makes rules 1 and 2 executable at any position size
Our tokenized gold explainer covers how the backing works, and our gold-backed cryptocurrency guide scores the major tokens against a seven-point safety test.
A central bank's gold is held in a vault generating no income, which is appropriate where the mandate is stability rather than return and the balance sheet is measured in trillions.
An individual investor operates under a different mandate. Staked tokenized gold provides the same exposure with a yield component, sourced from corporate gold-collateral lending rather than from token emissions. The mechanics are set out on the yield strategy page, and the risks, including smart-contract and counterparty exposure, are on the security and custody page.
Two points are relevant here: a stated APR is a target for which the mechanics are published, and the counterparty relationship underlying the lending strategy is named and documented. Both pages set this out in full.
The National Bank of Poland has been the world's largest official buyer for two consecutive years. It added 31 tonnes in the first quarter of 2026, taking holdings to 582 tonnes, then a further 14 tonnes in April to reach 595 tonnes, roughly 30% of total reserves, against a stated 700-tonne target.
What distinguishes this as a case study rather than a statistic is the consistency. Poland bought through the January peak and through the subsequent 27% decline without altering pace, and continued purchasing after Governor Glapiński publicly raised the possibility of sales.
This represents rules 1, 2, and 3 executed simultaneously by one institution over four years. The target determined the behavior rather than the price.
The People's Bank of China has reported net purchases for 18 consecutive months as of April 2026, including its largest single month since December 2024 delivered during the price drawdown.
China has purchased continuously for eighteen months without a reported pause. The more revealing figure is the denominator. Gold represents approximately 9% of China's total reserves against roughly 70% for the United States and 30% for Poland. Whatever target China is working toward, considerable capacity remains, and the buying recorded to date has moved the percentage only marginally.
What it demonstrates: China buying gold is a multi-year structural program rather than a trade. It also explains why monthly headlines about China buying gold move markets less than you would expect. The pace has been steady for a year and a half, and steady is the point.
The ECB identified Tether as the largest single gold buyer in 2025, acquiring over 100 tonnes. Set against the roughly 850 tonnes bought by the entire official sector that year, one private issuer accounted for more than a tenth of global central bank-scale demand.
Metrics to present alongside this: tonnes acquired relative to the top ten sovereign buyers of 2025, and the growth in tokenized gold market capitalization over the same period. Together, they frame central banks' buying gold as one instance of a broader pattern rather than a sovereign-specific phenomenon.
What it demonstrates: the reserve-diversification logic driving central bank gold reserves in 2026 extends well beyond sovereigns. Any institution holding large balances in someone else's liability faces the same problem and reaches for the same solution. Read the central bank gold reserves 2026 data as a signal about balance-sheet design rather than about geopolitics alone.
The evidence for central banks buying gold is consistent and drawn from primary sources: four years averaging approximately 1,000 tonnes annually, gold now exceeding US Treasuries within reserve portfolios, a widening buyer base, and 89% of surveyed central banks anticipating further increases.
The precise position belongs alongside that. The dollar's share of FX reserves remains near 57% and it continues as the world's transactional currency. Higher prices moderate some official buying. Gold has recorded extended flat periods as well as record runs. This is a durable, multi-year reallocation, and describing it accurately is what makes it actionable.
What transfers to the individual investor is the behaviour. Buy on a schedule. Size to a target allocation. Treat declines as opportunity. Control the custody. Those four rules are why central banks buying gold through a 27% decline reflects discipline rather than contrarianism, and they carry no cost to adopt.
One asymmetry favours the individual. Central banks accept zero yield as the cost of holding an asset with no counterparty. An individual investor need not. See the current gold staking rates, or read the full yield mechanics first if you would rather check the plumbing before the price.
