Custody is a Decision About Everything That Comes After It

One of the biggest mistakes banks tend to make when entering digital assets is believing they are deciding on coins, clients and use cases when, in reality, they are making decisions about operating models. Discussions around board tables tend to centre on familiar questions: which assets should we support (Bitcoin? Stablecoins? Tokenised real-world assets?); which clients should we target; is there sufficient demand; and how quickly can we get a custody offering to market? While these are all perfectly reasonable commercial questions, they are not the ones that will determine whether a digital asset strategy succeeds over the next five or ten years. Instead, the decision that matters most is usually treated as an implementation detail, quietly made somewhere between procurement, architecture and governance, and by the time anyone realises its significance, it has already become embedded across the organisation.

To some extent, the financial services industry has fallen into the trap of viewing digital asset capabilities as a sequence of discrete products. First comes custody, then perhaps staking. Then tokenised collateral, stablecoin payments, off-exchange settlement or some future capability that has yet to emerge. Every stage is treated as a fresh business case with its own budget, governance process and implementation project. That mindset is understandable because it mirrors how financial institutions have traditionally introduced new products. The problem is that digital assets do not behave like traditional products. They share the same infrastructure, the same governance requirements and, increasingly, the same operational workflows. They are not isolated capabilities that happen to sit alongside one another; they are different expressions of a common operating model.

The irony is that custody, widely regarded as the safest and simplest place to begin, is often the point at which institutions unknowingly narrow almost every option they will have in the future. You cannot launch a custody service without deciding who is authorised to move assets, how approval workflows operate, where policies are going to be enforced, how risks are managed, what operational controls sit around transactions and which system becomes the source of truth. None of these decisions feel particularly strategic when the objective is simply to hold assets securely on behalf of clients. They are viewed as sensible implementation choices made in the context of a single product. But operating models have a habit of outliving the products they were originally designed to support.

This is no doubt why the banking industry has been asking the right questions but in the wrong order. Instead of starting out by asking what assets to support, banks should be asking which type of operating model will allow them to support every capability they are likely to offer over the next decade. That is a much more difficult conversation because it requires organisations to think long-term and strategically and to look well beyond today’s commercial priorities to imagine how digital assets may evolve. Yet that is precisely the exercise they need to undertake if they are serious about building institutional businesses rather than launching isolated products.

The reason this matters is that custody is not the destination. It is simply the first step. Once custody is live, clients inevitably begin asking for more. They want to stake idle assets. They want tokenised collateral to move seamlessly between counterparties. They want stablecoins to support treasury operations and real-time settlement. They want programmable payments and automated workflows. None of these developments should come as a surprise because the ability to move, programme and mobilise assets is exactly what makes digital assets valuable in the first place.

Unfortunately, this is often where the cracks begin to appear. Operating models that were designed around static asset safekeeping suddenly find themselves supporting assets that need to move continuously between wallets, venues, counterparties and protocols. Governance processes that once felt appropriately robust become operational bottlenecks. Approval models designed for infrequent transfers struggle when transactions become continuous and automated. Controls that worked perfectly well for custody alone become increasingly awkward as institutions layer staking, settlement and collateral management on top. What looked like incremental product expansion gradually becomes a series of expensive redesign projects, not because the technology has failed but because the original operating model was never designed to stretch beyond the first use case.

Often these problems are attributed to software limitations or vendor capability, and the assumption is that a different custody platform or an additional specialist provider will resolve the next phase of the roadmap. Yet, technology is rarely the primary constraint. The constraint is the architecture of decision-making that was established when the first product went live. Once systems become embedded, they acquire an inertia that is remarkably difficult to overcome. Every subsequent capability has to work around those earlier decisions, even when they were never intended to support anything beyond basic custody.

What makes this particularly frustrating is that the assets themselves are likely the least durable element of the entire strategy. Five years from now, the institutions entering digital assets today will almost certainly support products that barely feature in current board discussions. Stablecoins and tokenised money market funds will mature and evolve. New cryptocurrencies and blockchains will emerge and regulatory frameworks will change. Client demand is likely to shift in ways that are difficult to predict today. The assets will change because markets always change. The operating model, however, will remain. Governance frameworks, approval structures, operational controls and organisational responsibilities tend to endure long after the commercial assumptions that created them have disappeared.

That is why banks should spend considerably less time debating which asset comes first and considerably more time designing an operating model that assumes every capability will eventually arrive. This does not mean building for every conceivable use case on day one or investing in technology that may never be needed. It means recognising that custody is not a standalone product but the first expression of a much broader institutional capability. If the operating model is designed with that destination in mind, every subsequent service becomes an extension of the same framework rather than an exercise in dismantling and rebuilding what already exists.

Digital assets are often described as a technology transformation but that framing may not be helpful. Blockchain technology is advancing rapidly, and specialist providers continue to solve increasingly complex problems. The harder challenge is organisational – resisting the temptation to optimise for today’s product launch at the expense of tomorrow’s flexibility. Banks have spent decades learning that operating models are strategic assets in their own right. Digital assets do not change that lesson; if anything, they make it even more important.

Custody, then, should never be viewed simply as the first product on a digital asset roadmap. It is the moment an institution decides, consciously or otherwise, how every capability that follows will be governed, controlled and operated. Get that sequence right and each new service becomes another configuration on an architecture that was designed to evolve. Get it wrong and every new capability becomes another transformation programme, another governance review and another costly reminder that the most important decision was never just about custody at all.