Connecting financial systems has become an industry in itself. The next phase of infrastructure makes that industry obsolete.
If you look at the last three decades of financial infrastructure spending, a huge proportion has gone into one activity: making different systems talk to each other.
Correspondent banking networks, payment gateways, central counterparties, cross-chain protocols, synchronised settlement platforms. All of them solve the same problem. Two ledgers need to reach a shared outcome, and neither can do it alone.
The solution has always been to insert something in between.
A bridge.
This approach has worked well enough to become invisible. It is simply how the industry operates. But it is now hitting structural limits, limits that faster settlement cycles, richer data standards, and better consensus mechanisms cannot fix on their own.
The Problem Is Architectural
The problem is not the performance of the bridge. It is the architecture of bridging itself.
A bridge, in financial infrastructure terms, is any external component that coordinates a state change across two sovereign transaction systems. It holds liquidity, interprets commitments, manages timeouts, and resolves disputes. It sits outside the finality boundaries of both systems and therefore introduces a separate point of risk.
Even the most modern cross-border projects follow this pattern. A shared platform sits between national ledgers. A common escrow contract locks funds on both sides. A central sequencer orders transactions.
The technology is new. The architecture is not. The bridge remains.
The industry has spent its energy making bridges more reliable. Better legal frameworks. Prefunded collateral. Stronger consensus algorithms. These are genuine improvements, but they do not change the fundamental dynamic.
Each bridge still requires its own governance, its own liquidity pool, its own failure procedures. As the number of distinct transaction systems grows — CBDC platforms, tokenised asset ledgers, programmable bank networks — the number of bridges needed multiplies.
The result is a system where complexity grows faster than capability.
A Different Starting Point
What if the coordination logic did not sit in a bridge at all?
What if it sat in the systems themselves?
This is the shift from exogenous to endogenous interoperability.
Instead of placing a coordinator between systems, each system gains the native ability to participate in cross-system settlement directly. Two transaction engines, each operating under its own governance and legal framework, can lock assets, exchange proofs, and finalise a transfer without an intermediary platform orchestrating the steps.
This is not a technical tweak.
It is a different way of designing transaction infrastructure.
The capacity to settle across boundaries becomes a feature of the system, not a service provided by a third party.
Settlement Grammar
The building block of this approach is what we can call a settlement grammar.
It is not a messaging standard. The industry already has ISO 20022 for that.
A settlement grammar is smaller and more fundamental. It defines a handful of primitive operations:
- Lock an asset under a condition
- Prove that the lock exists
- Release on matching proof
- Revoke on timeout
These are the atomic moves from which any bilateral or multilateral settlement can be constructed.
When two systems implement the same settlement grammar natively, they do not need a bridge to translate between them.
They share the basic vocabulary of finality.
The transaction logic can be different, the assets can be different, the legal underpinnings can be different. But the commitment protocol is shared, and that is enough.
Why This Changes the Shape of the System
Bridges produce a hub-and-spoke topology. Central platforms concentrate settlement activity, manage risk, and charge for the service.
Endogenous interoperability points toward a different structure: a mesh.
Any compliant node can settle directly with any other.
This does not remove the need for central bank money or high-quality settlement assets. Those remain essential. But the settlement asset becomes a parameter of the transaction rather than the venue for the transaction.
A wholesale CBDC can be the asset being locked and released across two private-sector ledgers, without either ledger needing to join the central bank's platform.
The Role of Central Banks
For central banks, this reframes the role.
The question becomes less:
“What platform should we build?”
and more:
“What settlement grammar should we require, and how do we certify that participants implement it correctly?”
The central bank moves from system operator to protocol governor.
It ensures the grammar is safe, the asset is sound, and the implementations meet a minimum bar. It does not need to host every transaction.
The Role of Commercial Banks
For commercial banks and infrastructure operators, the shift is equally significant.
The transaction engine — the software stack that manages balances, executes transfers, and maintains finality — becomes the strategic asset, not the network membership.
A bank that runs a transaction engine with native cross-system settlement capability can connect to any counterparty without:
- Negotiating a new bilateral agreement
- Onboarding to a new platform
- Funding a new bridge
Not a Universal Ledger
This is worth distinguishing from the universal ledger vision that has gained attention in recent years.
A universal ledger solves the bridging problem by removing the need for bridges, but it does so by moving everyone onto the same platform.
That demands:
- A single governance structure
- A shared legal framework
- A common technical stack
The history of financial infrastructure suggests this is extremely difficult across jurisdictions with different legal traditions, monetary policies, and supervisory mandates.
Endogenous interoperability takes a different path.
It accepts that multiple sovereign systems will persist. It does not try to consolidate them.
It gives them a shared grammar so that fragmentation becomes operationally manageable.
This is closer to how the internet works: not by running every computer on the same operating system, but by ensuring they all speak TCP/IP.
What Needs to Happen
1. Programmable Transaction Systems
Core transaction systems are becoming programmable.
Whether through DLT, cloud-native ledger architecture, or institutional smart contract platforms, the ability to embed commitment logic directly into the transaction engine is no longer theoretical.
Banks and central banks are already building systems that can support this.
2. The Limits of Bridges Are Becoming Visible
Cross-border CBDC experiments have shown that even when you build a shared platform, the legal and governance complexity does not disappear.
It simply moves into the design of the platform itself.
Some architects are now asking whether the platform has become the bottleneck.
3. Behavioural Standards Matter More Than Data Standards
The industry conversation is shifting from data standards to behavioural standards.
ISO 20022 tells you what a payment instruction should look like.
It does not tell you how two systems should resolve that payment against each other when something goes wrong.
The next frontier is standardising settlement behaviour — the protocol of finality, not just message formats.
The Practical Path
None of this means bridges disappear overnight.
They are embedded in legal contracts, operational processes, and balance sheets.
But the strategic direction is becoming clear.
The industry will spend less time building and maintaining connectors, and more time ensuring that transaction systems are natively capable of cross-system settlement.
The institutions that understand this shift early will be the ones that shape the settlement grammar and the governance framework around it.
Those that continue to treat interoperability as a connection problem will find themselves maintaining an increasingly expensive and complex web of bridges while others settle directly.
The bridging era served the industry well.
It was the right architecture for a world where transaction systems were closed, proprietary, and incapable of externalised commitment logic.
That world is passing.
Conclusion
The architecture that replaces it is not a better bridge.
It is systems that no longer need one.