Tokenised Gold: The Collateral Case the FCA Is Testing

Tokenised gold has moved off the conference panel and onto the regulator's desk. On 14 September the FCA opened a Call for Input asking whether tokenising the metal could improve how it is traded, transferred, pledged and held in UK markets. Responses close 23 October.

The same day it published the feedback statement from its May call for input with the Bank of England. There were 123 responses. Most agreed the post-trade side is where the opportunity sits, and specifically that the prize is how collateral moves between parties. The second half of that is the one that pays.

What sits in the vault and how little of it moves

LBMA's London vault data records 9,632 tonnes of gold held in London vaults at the end of August 2026, valued at $1.4 trillion, or roughly 770,571 bars. That is metal under trusted custody, with the Good Delivery framework and a deep network of dealers and clearing members behind it. The FCA calls London the biggest centre for spot gold trading in the world, and the vault numbers back it.

Then there is how little of that metal moves in a day. LBMA clearing data for July 2026 shows a daily average of 15.8 million ounces transferred, worth $64.5 billion, across 6,169 transfers and averaging 2,566 ounces each. That is the net physical settlement layer, running between the four banks that own London Precious Metals Clearing Limited.

Every one of those 6,169 transfers is a reconciliation between two sets of books and a vault, and a place where the metal and the cash can fall out of step.

Upstream of that rail, LBMA Trade Data reports a 12 week moving average weekly gold turnover of $935.08 billion for the period ending 11 September 2026, covering spot, swaps and forwards, options, and lease and deposit flows. That measure is gross rather than net, so it does not compare with the clearing figure on a like for like basis. It still shows how much trade sits on top of the physical rail that has to settle it.

Where the current structure binds

The FCA's own paper is blunt about the mechanics, and the bluntness is the reason to take it seriously. Allocated gold gives the strongest claim, because the custodian records specific bars as belonging to the account holder. It is slower and more costly to mobilise, which makes it awkward as collateral. Unallocated gold is liquid and moves by account entry, but it is a claim on the account provider, so it carries counterparty exposure and is a weaker collateral asset.

Then this line from the paper, which belongs on a whiteboard in every treasury that pledges metal: "the metal and cash legs of an OTC gold transaction settle through separate systems, creating timing, reconciliation and operational risks."

Tokenisation is a credible answer to that specific problem and not much else. Represent the claim, transfer it against payment in a single movement, keep it reconciled to the vault record, and you have removed a reconciliation event from a market that processes thousands of them a day.

The FCA sets out five questions that need answering first. The nature and enforceability of token holder rights against the underlying physical gold. The treatment of gold tokens in insolvency. Whether a secured party can exercise control and realise the asset. How settlement finality is determined when transfers happen on-chain. And how on-chain records reconcile with custody and vault records.

Four of those five are questions for lawyers. The reconciliation one is not, and it decides whether a gold token can be pledged at all.

Classification decides whether the product exists

If you build or distribute a gold product, the perimeter question decides whether the thing is viable. A UCITS scheme cannot be dedicated to gold. A CIS dedicated to gold may be an unregulated CIS, which limits who it can be marketed to. The definitions of CIS and AIF are broad and they overlap, and where a gold token lands turns on its legal and operational structure rather than the pitch deck.

The FCA has put a recognised "eligible gold token" classification on the table as one of five possible responses, with objective conditions covering rights, custody, redemption, governance and technology. It is listed as a Treasury and FCA led option. The fifth option goes wider: a bespoke regime for tokenised gold, or for tokenised commodities more broadly, agreed across all three authorities.

There is competitive pressure underneath this. The FCA notes that CME has introduced 1 ounce gold futures trading 24 hours a day, and that Kalshi has sought approval in the US for perpetual futures tied to gold and other precious metals. Extended hours are already pulling price discovery away from the London session.

The Bank has been laying rail for a while

The May joint statement with the Bank of England is the part people skim, and it should not be. Sixteen firms are working through the Digital Securities Sandbox on the live issuance and settlement of tokenised assets. The Bank has committed to a live synchronisation service targeted for 2028, and to work enabling tokenised equivalents of already eligible assets to be used as collateral at central counterparties and in its own operations. It also supports the Treasury's pilot issuance of a digital gilt instrument.

Sarah Breeden, the Bank's deputy governor for financial stability, described the point the sector has reached as the one where it moves "from pilots to production".

The FCA has said it will consult on rules for safeguarding relevant specified investment cryptoassets in the first half of 2027, and that a joint roadmap with the Bank setting target dates is coming later this year. Whatever comes out of the 23 October deadline lands inside a timetable that is already public.

What to do before 23 October

Write the response if you settle or pledge gold. The questions are specific enough that a short submission from a desk that actually moves metal carries weight. The alternative is a regime designed without your constraints in it.

Assume you will be classified, then price it. If there is any CIS or AIF ambiguity in your structure, work out now what marketing restrictions and additional authorisation would cost, rather than discovering the number at launch.

Fix reconciliation before the token. If the ledger cannot be tied back to the vault record on demand and to an auditor's satisfaction, the collateral case falls over the first time somebody asks.

Build to the 2028 horizon. A synchronisation service, a digital gilt pilot and a sandbox cohort are the rails, and a product designed around a single ledger architecture will struggle to plug into them.

The settlement problem underneath all of this is the same one I have been circling for months. I wrote earlier about why settlement, not issuance, is tokenisation's real bottleneck, and then about how tokenised collateral became the constraint once settlement was addressed. Gold is the cleanest test of both arguments, because the custody is old and the demand for usable collateral is already sitting in the vault.