Designing an Operating Model for Digital Assets

Leading banks have already moved past the question of whether to integrate digital assets. JPMorgan put a commercialization mandate on Kinexys. HSBC and a Standard Chartered–led group hold Hong Kong's first stablecoin licenses. Société Générale's FORGE is live in MetaMask. The harder question, and the one facing every other bank, is how that integration actually works inside the rest of the institution.

This is the harder problem, and it's where digital asset programs stall. Alison Kaufman, Circle's head of solutions engineering and financial partnerships, took up the question at our last Current event.

“Five years ago, the questions were more around if and when banks should be engaging with these types of assets,” said Kaufman. “The questions now are very much, ‘so we know we want to be engaging, and in many cases we already are, but how do we actually scale those solutions?’"

Digital assets don't scale at the edges

“When we are discussing funds flows and technical architecture with our financial institution partners, that's when you know things are getting very real,” said Kaufman. “Once we arrive at this stage of discussion, we're moving beyond the conceptual and into how this actually operates inside your systems.”

The complexity comes from the interdependencies. Choosing a blockchain affects custody. Custody affects how assets can be offered to clients. Settlement design affects liquidity management and funding models. Every decision creates second-order effects across the system. Pilots often clear technical feasibility only to stall here.

The operating-model question is also a differentiation question. Banks that resolve the interdependencies early build the architecture for what their institution will offer next: corridor-specific liquidity, programmable custody, real-time treasury reporting. 

The stack is the strategy

Bank digital asset programs treat the technology stack as the strategy: which blockchain, which custodian, which token. Those choices are the visible part of the work. The harder design question sits underneath, in the operating-model layers most banks haven't named.

“Scaling to production is not an easy conversation to have,” said Kaufman. “We see questions all the way from the institutional infrastructure, what blockchains should we be building on, to how we actually interoperate those digital assets, to what those settlement and abstraction layers look like so you're not disrupting the client experience as you implement this new technology.”

The operating model is the layer most programs underestimate. Treasury operations have to move from cut-off-driven daily cycles to continuous coverage; most US bank treasury floors aren't staffed for 24/7 operations. Custody architecture has to support multichain, multi-asset key management at the same risk standards as the existing book, and the existing custody organization wasn't built for it. Internal audit has to opine on smart-contract, oracle, and bridge risk, and most current IA functions cannot. The OCC's spring 2026 stablecoin NPRM made the gap concrete: it requires permitted payment stablecoin issuers to "demonstrate the operational capability to access and monetize reserve assets," a capability most institutions would have to build.

Interoperability is the differentiator

Most banks we work with are experimenting with tokenized deposits. The hard question we always ask: does the asset stay inside the bank's network, or does it interoperate with everything else?

“In some cases, it's about how do we take digital assets we're issuing today in a closed loop within our own network and interoperate with the broader ecosystem,” said Kaufman. “That’s how you take full advantage of what blockchain technology has to offer.”

A tokenized deposit settles transfers inside the bank, which is real value for treasury and client servicing. The strategic capabilities sit one layer out: cross-border exposure denominated in a regulated multi-currency stablecoin, public corridors running 24/7, settlement against a counterparty's tokenized cash equivalent at the moment of execution. Each of those is a client capability, and each requires connectivity to assets and infrastructure the bank does not own.

The market is voting for open architecture. Public stablecoin rails carried an estimated $350 billion in payment volume in 2025. Visa's settlement network now spans nine blockchains, including Circle's Arc. The bank that holds the closed-loop position too long pays the connectivity cost later.

The onchain decision is offchain

The hardest decisions sit inside the bank's own operating model (treasury, custody, AML, audit, core systems) and in the interoperability choices that connect those layers to the broader market. The patterns set inside leading institutions over the next four to six quarters will likely become the patterns the rest of the market builds against.

The practical move is to pull operating-model design and interoperability into a single workstream, with shared owners and shared milestones. The next four quarters will sort the institutions that built the operating model from those that bolted onchain capability onto the old one.

‍Explore more insights from Circle Current