Canton was engineered for a specific class of problem: enabling multiple regulated parties to settle complex transactions atomically while preserving privacy and retaining control over their own data and validation. That design has delivered. Institutions including DTCC, Goldman Sachs, BNY, Broadridge, JPMorgan and Visa now operate on the network. Monthly volumes in the trillions of dollars of repo and collateral activity are no longer projections.
The same architecture, however, has produced a durable tension that remains unresolved.
The Structural Gap
Because Canton prioritises privacy by default, party-based coordination and the absence of a global mempool, many actions that feel trivial on other networks become non-trivial here.
A significant portion of existing wallets can reliably execute plain fund transfers. They struggle with the richer multiparty patterns that native Canton applications require. The network excels at institutional settlement, but remains awkward for the repeated, low-stakes, permissionless activity that retail users and smaller builders treat as baseline.
This is not simply a temporary interface problem. It is a direct consequence of the properties that make Canton valuable to institutions in the first place.
The Clearest Response So Far
OneSwap has been the most deliberate response to that gap.
Its core design treats the limitations of retail wallets as a first-class constraint rather than an inconvenience. Users express intent, send tokens to a deposit party, and receive the output back through the same party. The application absorbs much of the complexity that most wallets cannot yet sign directly.
The recent sequence of upgrades — single atomic settlement, a built-in platform wallet, prepaid gas, an approximately 80% reduction in network fees, and the announcement of the first Pump.fun-style launchpad — can be understood as successive attempts to shrink the distance between Canton’s institutional settlement quality and ordinary user expectations.
The real product is therefore less the AMM itself than the abstraction layer around it: a surface that allows both end-users and other applications to ignore as much of the underlying complexity as possible.
What Still Holds It Back
The limitations remain material.
Liquidity is still early. LP ownership continues to be tracked primarily off-ledger. The launchpad has been announced but is not yet live. Even after the fee reduction, network costs remain higher than on pure L1 environments optimised for high-frequency retail use.
More importantly, it is still unclear whether this abstraction can sustain meaningful volume and native asset depth at scale. That is an empirical question, and the market has not answered it yet.
The Deeper Question
The deeper issue is structural.
Canton has demonstrated that it can host serious institutional infrastructure. It has not yet demonstrated that it can host a low-friction consumer surface without eventually altering the character of the network itself.
If the abstraction succeeds at scale, the surface may begin to reshape what the network is used for — and by whom.
Can a system optimised for controlled multiparty privacy and institutional-grade atomic settlement develop a retail layer that feels natural without diluting the very properties that attracted institutions in the first place? Or will the interface, over time, start to redefine the network?
That question is no longer abstract.
The next phase of surface-layer development will begin to answer it.
