Written by: Agata Wojtas, Chief Commercial Officer, Digital Colliers
De Nederlandsche Bank just moved €10 billion of gold reserves out of the US and Canada. The stated reason was geopolitical instability. The move took months of planning and involved repatriating physical assets from Federal Reserve Bank vaults in New York and Bank of Canada vaults in Ottawa. For most mid-market banks, the more uncomfortable question is this: if your board asked today where every material asset is held and which jurisdictions touch your custody chain, how long would it take you to produce a definitive answer?
Why Jurisdiction Is a Board Question Now
DORA has been in force across the EU since 17 January 2025. The Digital Operational Resilience Act makes third-party risk and operational continuity a regulatory obligation, not a treasury nicety. Your board is now personally accountable for understanding concentration risk in your service providers, correspondents, and custodians. Custody jurisdiction sits at the centre of that obligation. If a single custodian holds 40% of your reserve assets and that custodian operates under a jurisdiction facing sanctions risk or capital controls, you cannot discover that fact three weeks into a board inquiry. You need the answer in minutes.
The DNB move is the visible tip. Below the surface, every mid-market bank is running the same arithmetic: where do we have counterparty concentration, which jurisdictions create exposure we have not priced, and can we prove we know the answer before the regulator asks.
The Data Problem Nobody Owns
The information exists. It lives in your vendor register, your custody agreements, your correspondent banking contracts, and your reserve asset reporting. The problem is that those four sources do not talk to each other. Treasury owns the custody agreements. Compliance owns the vendor register. Finance owns reserve reporting. Operations owns correspondent banking relationships. When the board asks for a jurisdiction map, you are asking four teams to reconcile data by hand, usually in spreadsheets.
Most mid-market finance teams still run month-end close in spreadsheets, pulling numbers across systems by hand. Typical month-end close runs eight to ten days. A custody jurisdiction map is worse because it crosses more teams and the source data is less structured. Custody agreements are PDFs. Vendor registers are often still in SharePoint. Correspondent bank relationships live in email threads and treasury management systems that do not export clean CSVs. You are looking at two to three weeks of manual work to produce a board-ready report, and it is stale the moment you finish it.
What a Semantic Layer Solves
A semantic layer is middleware that sits between your source systems and your reporting tools. It defines the business logic once: which fields mean the same thing across systems, how to join vendor records to custody records to reserve holdings, which jurisdiction mappings apply. Once the semantic layer is in place, the board question becomes a query. You can produce a jurisdiction map in minutes because the joins and definitions are already encoded.
The technical shape is straightforward. You connect your custody management system, your vendor register, your treasury management system, and your general ledger to the semantic layer. You define the entities: custodian, correspondent bank, reserve asset, jurisdiction. You define the relationships: which custodian holds which assets, which jurisdiction governs which custodian, which correspondent banks intermediate which payment flows. The semantic layer materialises those relationships as a queryable graph. Your board asks where the gold is. You run a query that aggregates reserve holdings by custodian jurisdiction and returns a ranked list in seconds.
The Questions Boards Are Starting to Ask
Once you can produce a jurisdiction map on demand, the follow-on questions get sharper. What percentage of our reserve assets sit in jurisdictions with recent sanctions activity? Which single custodian represents our largest concentration risk? If we lost access to our primary correspondent bank in jurisdiction X, which payment flows would break and how quickly could we reroute them? How many of our top-ten vendors by spend operate under jurisdictions we have flagged as higher risk?
These are not theoretical questions. DNB answered them before moving €10 billion of physical gold. Your board will ask them in 2026, and the answer cannot be we will get back to you in three weeks. The operators who ship jurisdiction mapping as a repeatable capability this year will spend 2026 refining their risk models. The operators who are still reconciling spreadsheets in June will spend 2026 explaining why they are behind.