For most of the post-war era, the question of where a central bank kept its gold was one of the most boring in finance. The answer, overwhelmingly, was somebody else’s vault — the Bank of England, the New York Fed, a handful of trusted hubs. Custody was a settled convenience. The 2026 evidence says it is settled no longer.
The World Gold Council’s annual Central Bank Gold Reserves Survey, published in June, drew a record 76 respondents and was conducted between 5 February and 19 May 2026 (World Gold Council). Two numbers get the headlines: 89% of reserve managers expect global central bank gold holdings to rise over the next 12 months, and a record 45% expect their own institution’s reserves to increase. The buying story is real and well told. The custody story underneath it is not.
Gold is coming home
Ask reserve managers where their gold actually sits, and the traditional hierarchy is still standing — but cracking. The Bank of England remains the single most popular location, named by 57% of respondents. Domestic storage is now a close second at 49%. The Bank for International Settlements sits at 16%. The Swiss National Bank has halved, from 12% in the 2025 survey to 6% in 2026.
The static snapshot understates the motion. Over the past 12 months, 9% of respondents increased the share of gold they hold domestically — up from 5% a year earlier — and 10% diversified into additional overseas locations, up from just 2%. Looking forward, 7% plan to increase domestic storage and 9% plan to spread across more jurisdictions. These are minority moves. But they are minority moves that roughly doubled in a single year, in an activity that used to change on a scale of decades.
Why custody became a risk
The trigger is not subtle. In February 2022, a G7-led coalition froze roughly US$300 billion of the Central Bank of Russia’s reserves — about half of its total holdings — most of it immobilised inside European securities depositories (Brookings). That single act rewrote the risk model for every reserve manager watching. It demonstrated that foreign-held reserves are not unconditionally the owner’s property; they are a claim that a host jurisdiction can suspend.
Gold is the asset least exposed to that risk — but only if you can touch it. Bullion sitting in an allocated account at the Bank of England is safer than a bond in a sanctionable depository, yet it is still gold you do not physically control. Bring it home and it becomes something a foreign government cannot freeze, seize, or condition. The survey’s storage numbers are the logical tail of the survey’s demand numbers: if you are buying gold specifically as crisis insurance, insurance you cannot access in a crisis is a strange thing to pay for.
The same survey shows the wider retreat from counterparty exposure. Over a five-year horizon, 74% of respondents expect a moderately or significantly lower share of reserves held in US dollars. Gold is the mirror image of that trade — a reserve asset with no issuer, no coupon, and no jurisdiction. Holding it at home closes the loop.
The buyers behind the numbers
The flows corroborate the intentions. Poland’s central bank has been the most aggressive accumulator, adding 45 tonnes in the year to May 2026 to reach 595 tonnes — now roughly 30% of its total reserves (World Gold Council). The People’s Bank of China bought 8 tonnes in April, its largest monthly purchase since December 2024, extending an unbroken run of 18 months and lifting official holdings to 2,322 tonnes. The Czech National Bank added 3 tonnes in April, its 38th consecutive monthly purchase. Uzbekistan, whose gold is now 88% of its reserves, remains a heavyweight even after trimming a tonne.
Not everyone is buying: Russia — the country whose frozen reserves started this — has been a net seller, offloading 22 tonnes in the year to May, drawing on the metal precisely because its other reserves are inaccessible. That is the whole thesis in one line item. Gold held at home is the reserve you can still spend when the rest is locked.
Implications
The comfortable read is that this is a slow-motion, low-stakes reshuffle. Minority percentages, a few tonnes moved between vaults. That underrates two second-order effects.
First, the plumbing. London’s role as the center of the gold market rests on the assumption that central-bank metal stays put and can be lent, leased, and settled against. A structural drift toward domestic vaulting thins that pool. It does not have to be large to widen lease rates and settlement frictions in the London bullion market; marginal supply sets the price of liquidity.
Second, the signal. Reserve custody is a statement of trust in the host jurisdiction. When 10% of reserve managers spend a year diversifying away from single-location foreign storage, they are pricing the probability that the rules of custody change under them. That is not a vote against London or Basel specifically. It is a vote for optionality — and optionality, once reserve managers start paying for it, tends to compound.
Watch the 2027 survey for whether the “increase domestic storage” line keeps climbing off this year’s 9%. If it does, the story stops being about how much gold central banks own and becomes about who they are willing to trust to hold it. The buying was always the easy part. Where the metal sleeps is the harder question, and for the first time in a generation, a growing minority of central banks are answering it themselves.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.